Search is not available for this dataset
article
stringlengths 4.36k
149k
| summary
stringlengths 32
3.35k
| section_headings
listlengths 1
91
| keywords
listlengths 0
141
| year
stringclasses 13
values | title
stringlengths 20
281
|
---|---|---|---|---|---|
The nature of the optical cycle of photoactive yellow protein ( PYP ) makes its elucidation challenging for both experiment and theory . The long transition times render conventional simulation methods ineffective , and yet the short signaling-state lifetime makes experimental data difficult to obtain and interpret . Here , through an innovative combination of computational methods , a prediction and analysis of the biological signaling state of PYP is presented . Coarse-grained modeling and locally scaled diffusion map are first used to obtain a rough bird's-eye view of the free energy landscape of photo-activated PYP . Then all-atom reconstruction , followed by an enhanced sampling scheme; diffusion map-directed-molecular dynamics are used to focus in on the signaling-state region of configuration space and obtain an ensemble of signaling state structures . To the best of our knowledge , this is the first time an all-atom reconstruction from a coarse grained model has been performed in a relatively unexplored region of molecular configuration space . We compare our signaling state prediction with previous computational and more recent experimental results , and the comparison is favorable , which validates the method presented . This approach provides additional insight to understand the PYP photo cycle , and can be applied to other systems for which more direct methods are impractical .
The view emerging from previous experiments and calculations is that the pG free energy basin is stabilized by the hydrogen bonding network within the chromophore binding pocket , and disrupting these bonds creates population in an alternative minimum: the signaling state . Within the energy landscape theory perspective [35] , PYP can be seen as a somewhat frustrated energy landscape with two main basins . Alteration of the isomerization state of the chromophore and consequent disruption of the hydrogen bonding network shifts the relative stability of the basins from pG ( for trans pCA ) to pB ( for cis pCA ) . The main coarse features of this landscape should be captured with an appropriate coarse-grained potential , and we use such a potential below to initiate the search for a signaling state ensemble in the pB basin . Coarse-grained modeling has become a popular technique [36] , [37] due to the complexity and long timescales involved in biological systems . Such methods speed computational simulation times by combining several atoms into a single bead . The method we use here was first presented in reference [38] , and is hereafter called the DMC method ( after Das , Matysiak , and Clementi ) [39] , [40] . The DMC model represents each amino acid type as a different ‘color’ bead , centered on the backbone , and with color-specific interactions between beads . The fact that alteration of a single PYP residue produces such a large change in the global free energy landscape , namely the population of an alternative minimum , suggests that non-native interactions are an important feature for this system . Therefore , we expect that structure-based models in which only native contacts are energetically favorable [41] are unable to capture the essential features of this system . Indeed application of such models to PYP did not produce any pB-like minimum in the free energy , only a pG-like and a globally unfolded minimum . Multiple-basin structure-based models have been developed [42]; however these require the structures in each minimum to be known a priori . The only information the DMC model explicitly requires is the pG state configuration ( taken from the PDB ) and the knowledge that the chromophore is exposed to the solvent in the pB signaling-state configuration . The DMC model can be considered a “first-order” correction to structure-based models , taking into account non-native interactions . Indeed , as discussed below , the DMC model produces a free energy landscape with a partially unfolded minimum in between the folded ( pG ) and unfolded states . Of course the DMC model of PYP does not capture the fine details of the PYP free energy landscape . Rather the model can be considered a starting point to select candidate signaling-state configurations for further analysis and exploration . In order to find such candidate structures , the free energy landscape of the DMC PYP system is analyzed with the locally scaled diffusion map ( LSDMap ) [43] . LSDMap is a dimensionality reduction technique that extracts collective variables directly from simulation trajectory data , without the need for other input information such as reaction paths , intuitive coordinates , etc–which is why the technique is particularly useful for this system . The method approximates a numerical solution for the eigenfunctions of the Fokker-Planck operator , and the resulting diffusion coordinates ( DCs ) represent collective motions that correspond to barrier crossing processes in the system . This method was first tested on alanine dipeptide and a DMC model of src-homology 3 domain ( SH3 ) [43] , and has been applied to understand polymer reversal inside a nanopore [44] , the folding pathways of a miniprotein [45] , and the interaction of anthramycin and DNA [46] . From this analysis of the DMC free energy landscape , we ‘zoom in’ on the region likely to contain potential pB-like structures using an all-atom reconstruction technique: the Reconstruction Algorithm for Coarse-Grained Structures ( RACOGS ) [47] . There are many algorithms for reconstruction of protein side chains ( e . g . [48]–[51] ) . The RACOGS method includes a side-chain minimization step that allows the side-chains to move continuously in space , rather than only changing between different rotamers in a library . Since rotamer libraries are typically built from datasets of native or near-native structures , this additional step makes RACOGS less likely to be biased toward native-like side-chain placements . Such a feature is important in reconstructing the non-native pB state of PYP . The all-atom free energy landscape in the pB region of configuration space is expected to be rough [32]–[34] . Therefore to explore the area around the reconstructed structures we use an enhanced sampling algorithm , Diffusion Map-directed-Molecular Dynamics ( DM-d-MD ) [52] . DM-d-MD is an iterative method that uses the ideas of the diffusion map , in particular that the slowest barrier crossing timescale corresponds to the first DC ( DC1 ) , to enhance the sampling by increasing the probability that a system will cross free energy barriers . At each iteration , a short swarm of MD trajectories are run from an initial point , a diffusion map calculation is performed on that swarm , and the configuration with the largest DC1 is selected as the ‘frontier point’ , which is used as the initial point in the next iteration . DM-d-MD has been illustrated in alanine dipeptide and alanine-12 [52] , in which there is a three-orders-of-magnitude speedup of the sampling in comparison to standard MD . We use this novel combination of techniques to obtain a signaling state ensemble of WT-PYP . This method is unique in that , as far as we can tell , this is the first time results of dynamics with a coarse-grained potential are used to reach a non-native basin , and from coarse-grained structures in that basin all-atom reconstructions are used to initiate a more detailed exploration of the new basin . Our results are in agreement with previous calculations and more recent experimental data . We anticipate the overall strategy presented here to be applied to other systems for which conventional techniques are impractical or impossible .
We used the DMC method to construct a coarse-grained potential for the ‘activated’ state of PYP , i . e . the state after photo-absorbtion . The details of the construction of the model are given in the Materials and Methods section . Briefly , the DMC model treats each amino acid as a single bead , with non-bonded interactions between beads dependent on the type of two amino acids . To model the photo-activated state of PYP , the nonbonded interactions between the chromophore residue and all others was set to zero , which roughly models the disruption of the hydrogen bonding network within the chromophore binding pocket . A simulation of was performed using GROMACS [53] , with data collected every 50 ps , and at a temperature sufficient to have many folding/unfolding events . In Figure 1 the free energy is shown in terms of the first and third DCs . The slowest collective motion of this DMC system corresponds to a global unfolding of the protein . Structures with a large positive DC1 correspond to configurations very similar to the pG native state , while configurations with a large negative DC1 are unfolded . The figure shows an additional minimum in the intermediate region of the free energy , and configurations within this minimum are good candidate pB-state configurations . Representative structures from each region are presented in Figures S1 , S2 , and S3 in Text S1 . In addition , histograms of the Cα RMSD to the NMR pG structure are shown in Figure S4 in Text S1 . Approximately 1000 coarse-grained structures were collected from the local minimum in free energy near DC1 = 2 . 5 for further analysis . The free energy is shown in terms of DC1 and DC3 to allow for a clearer view of the intermediate region . The DMC model supplied the ( Cα ) positions for the candidate signaling-state configurations , for which we want to recover the atomic details to more accurately explore the pB region of configuration space . This is accomplished with the Reconstruction Algorithm for Coarse-Grained Structures ( RACOGS ) [47] , which is specially designed to recover all-atom details not only in the native basin , but anywhere in configuration space . One example reconstruction is shown in Figure S5 in Text S1 . The ≈1000 reconstructed all-atom configurations are then solvated and equilibrated using previously established protocol [34] . From these , we used the criterion of lowest protein-only potential energy to select 10 structures for further analysis . To explore the molecular configuration space around these solvated structures , DM-d-MD [52] is initiated from each equilibrated structure . Previous work has suggested two collective variables: the root mean square deviation ( RMSD ) of the α3 helix ( residues 41–53 ) with respect to an ideal helix , and the distance between the GLU46 residue and the pCA chromophore , are good collective variables in which to visualize the system , and that there are a few metastable states in between the pB and pG configurations [34] . These coordinates are used in Figure 2 , along with the underlying black free energy contours obtained from previous parallel tempering calculations ( See Figure 1 from Vreede , et al . [34] ) . In Figure 2 the initial point for the DM-d-MD is shown as a gold circle , the minimum-energy structure from the DM-d-MD as a gold star , and the other DM-d-MD points in light blue . Stand-alone DM-d-MD is an exploratory procedure , and does not yield a Boltzmann distribution of configurations ( however techniques such as umbrella sampling can recover the Boltzmann distribution from a set of DM-d-MD frontier points [52] ) . To recover a local , quasi-equilibrium distribution of the pB signaling state , the lowest energy DM-d-MD frontier point is used to initiate approximately 100 runs of ordinary MD simulations , the results of which are shown in light purple . For the purpose of determining the “lowest-energy” point , the energy was calculated for the protein only , using the Gromos96 43a1 [54] force field . The average length of the runs was 22 ns , and data was collected every 50 ps after the initial 2 ns , yielding a total of 40 , 209 configurations . This is our pB signaling state ensemble . It should be noted that while the MD results overlap with the metastable minima obtained from previous calculations , as shown in the figure , projection onto a two-dimensional coordinate system can be misleading , and we rely on further analysis below to verify our pB-state ensemble . The choice of the “lowest energy” DM-d-MD frontier point is simply a convenient choice for the purposes of this study . We show in the Figure S6 in Text S1 that the next few low-energy DM-d-MD points yield similar overall structure by comparing the secondary structure content of various configurations using the Stride algorithm [55] . As discussed below , the secondary structures are all similar to one another , and–with the notable exception of the first 25 residues–similar to the experimental structure . For comparison , the coordinates of the experimental Δ25 and WT-PYP pB-state structures are shown as dark blue and red * , respectively , in Figure 2 The Δ25 structures are much more scattered than the experimental WT-PYP configurations , due to the larger variation in the Δ25 configurations compared to the WT-PYP . Within the 20 structures in the Δ25 set , the relative Cα RMSD average and standard deviation is 0 . 41+/−0 . 07 nm , while for the 14 WT-PYP structures the average and standard deviation is only 0 . 16+/−0 . 06 nm . This difference is potentially due to the lack of long-range information in the NMR Δ25 experiments [28] . Interestingly the DM-d-MD explores mostly the upper right-hand region of the RMSD α3 – dXE space , and the minimum energy structures are located in the middle of the putative pB region in these coordinates .
Figure 3 displays a configuration in the pG state for reference , a configuration from the experimental WT-PYP pB state , a configuration from the experimental signaling state of Δ25 , and a configuration from our pB ensemble . In the pG configuration , the chromophore is tucked inside the chromophore binding pocket , the α3 helix ( colored in blue ) is well formed , and the binding pocket cap ( residues 98–103 colored in green ) is in place . All three signaling state structures display features known to be associated with the signaling state: the α3 helix is unformed and the chromophore is exposed to the solvent . The pB state is in general less well structured than pG configurations , while some secondary structure elements remain intact . Visually , the secondary structure in the Δ25 configuration looks more well-ordered than the experimental WT-PYP and our pB ensemble . In comparing our result with experimental WT-PYP , the amount of retained secondary structure is similar; however the location of the 25 N-terminal residues differs ( discussed more below ) . To quantify the degree of structural similarity between the pG and pB states , we have computed the relative fluctuations of the ( Cα ) atoms for the various datasets with respect to those of the pG state . Because of the flexible nature of much of the protein in the pB state , this is a better metric than the RMSD between different structures . These fluctuations are computed by first aligning the corresponding Cα's to the last 100 residues of the pG configuration ( model 11 of PDB ID: 3PHY [18] ) , and then calculating the displacement of the Cα's from that pG configuration . This was done for the 20 Δ25 structures , the 14 WT-PYP configurations , and the pB ensemble resulting from our method . Only the last 100 residues are used in the alignment and calculation because 1 ) the first 25 are not present in Δ25 and 2 ) there is a large difference in the location of these residues in previous calculations and experiment ( see below ) . Figure 4 compares the results . For regions of the protein in which the pG secondary structure is preserved in the pB state , for example the α helix formed by residues 79–84 , the Cα displacement is minimal . However in regions of the protein where structure is lost , for example the α3 helix in residues 79–84 , the fluctuations are larger . There is general agreement among all three datasets . The two main conserved regions are the helix in residues 76–85 , and the central β sheet , and can be seen in all three signaling-state structures in Figure 3 . From Figure 4 , near the chromophore region the experimental WT-PYP configurations are more similar to the pG state , the Δ25 configurations fluctuate most , and our configurations are in between . In the experimental WT-PYP configurations , the loop containing the chromophore only moves enough to allow the chromophore to be flipped out of the binding pocket , while in the Δ25 configurations the structure is comparatively more extended , leading to larger deviations from the pG dark state . The most significant difference between the WT-PYP pB state and our ensemble is the location of the first 25 N-terminal residues . The experimental configuration shows an open binding pocket , and the N-terminal residues across the pocket [28] . Our configurations , as well as previous calculations [32] , have the N-terminal residues behind the central β sheet , and away from the binding pocket . This difference is most likely due to the force-field used in simulation . Due to hydrophobicity , it is unlikely that a molecular dynamics simulation will explore open configurations such as that shown in panel b of Figure 3 . Indeed we have solvated and equilibrated one of the configurations from PDB ID 2KX6 ( one of the red *s in Figure 2 ) and performed 20 independent 30-ns simulations . In most of the simulations , the extended N-terminal tail moves toward and into the open chromophore binding pocket . Representative configurations are shown in Figure S8 in Text S1 . Figure S7 in Text S1 shows the results of a Stride [55] secondary structure calculation at various snapshots both during one of these simulations and during a simulation from our signaling ensemble . The location and interaction of the first 25 N-terminal residues with the rest of the protein is still an open question for this system . These interactions are important for understanding WT-PYP , and the different kinetics in the Δ25 system , which has a pB state lifetime roughly 100 times longer than that of WT-PYP [22] . Obviously the difference in Δ25 and WT-PYP kinetics is due to the absence of the first 25 N-terminal residues in Δ25 . Experimental results on Δ25 show that even in the pG state the α3 helix is unstable compared to the pG state in WT-PYP [27] . In addition , Δ25 even exhibits unfolding in biological conditions without any photo activation [22] , which is not observed for WT-PYP . Previous calculations show reformation of the α3 helix is a bottleneck in recovery of the pG state from the pB , and have suggested that the chromophore cannot form the needed contacts in the binding pocket if this helix is not well formed [33] . One possibility is that the N-terminal residues in WT-PYP facilitate the formation of this helix , increasing the recovery rate of pG relative to Δ25 . Our pB ensemble shows interactions between the N-terminal residues and the α3 helix in the form of hydrogen bonds . 70% of the configurations have hydrogen bonds between ASN43 and GLN22 , 59% have hydrogen bonds between ALA44 and ASP24 , and 54% have hydrogen bonds between ALA44 and GLY25 . It is possible that these interaction slow the reformation of the helix , speeding the recovery of pG in WT-PYP . In the pG state , the first 25 residues are separated from the chromophore binding pocket by the central β sheet ( see Figure 3 ) . It is proposed that interactions through this β sheet lead to a stabilization of the chromophore binding pocket . In our simulations ( and in those reported previously [32] ) , the N-terminus in the pB ensemble is in van der Waals contact with the back side of the central β sheet . Such contact may encourage the reformation of the chromophore binding pocket , increasing the pG recovery rate in PYP . There is a possibility of allosteric interactions that drive the transition to and from the signaling state . To begin to investigate this issue , we have analyzed our signaling state ensemble using a generalized correlation [56] , [57] . The results of this analysis are shown in the Figures S9 and S10 in Text S1 , and suggest a potential bridging interaction of the chromophore binding cap with the α3 helix and the chromophore region . This is a direction we plan to pursue in future work . As pointed out above , the location of the N-terminal residues differs between experiment and theory . At this point it is unknown if these differences are due to the force fields used in calculation , differences in the experimental sample preparation details compared to those of the calculation ( the specifics of the PYP photo cycle are known to be environmentally dependent [1] ) , or to something else entirely . More work , both experimental and computational , is required to fully understand these aspects of the WT-PYP and Δ25 photo cycles . We have presented a novel combination of techniques: DMC coarse-grained modeling , LSDMap , RACOGS all-atom reconstruction , and DM-d-MD to obtain an ensemble of signaling state structures for WT-PYP . This amalgam of methods allows for an initial bird's-eye view of the free energy landscape , followed by a “zooming-in” to a region of interest . Such a process is ideal for the PYP system for several reasons . The timescales present are too long for conventional MD , and even advanced sampling methods such as parallel tempering [32] and transition path sampling [34] methods are pushed to their limits , making coarse-grained methods useful . The coarse-grained method we employ includes the effect of non-native interactions , which are thought to be important here , and only requires the structure of the pG dark state as an input . The LSDMap analysis method extracts collective coordinates directly from the results of the coarse-grained simulation with which to analyze the free energy landscape , without the need to rely on calculation [34] or chemical intuition to arrive at collective coordinates in which to analyze the free energy . Once coarse-grained structures are found in a region of interest in the LSDMap free energy , the RACOGS method is used to recover all-atom structural details . This method is designed to work well not only near native-like configurations , but to provide physically realistic structures anywhere in the landscape . Finally DM-d-MD allows for a rapid exploration of the newly located region of configuration space . This procedure should prove useful in other systems with long timescales and unknown free energy minima . Our results compare favorably with previously reported experimental and computational findings , which serves as a validation of our techniques and support for previous results . However uncertainties remain concerning the PYP photo cycle , in particular the role of the first 25 N-terminal residues . There is agreement between the current and previous computational results [32]; however both sets of computational results differ from experiment in the structure of the fist 25 N-terminal residues [9] , [28] . More work is needed to understand this aspect of the PYP photo cycle .
We use a coarse-grained modeling method [38] , termed the DMC method , to model the PYP protein in its “activated” state , i . e . after photon absorption and isomerization of the chromophore . The method itself is generally applicable to any system for which a native structure is available , and accounts for ( in an approximate fashion ) both the geometrical differences between various residue types and minimizes the energetic frustration of the folded structure . The DMC method treats each residue as a single bead , with the potential energy written as ( 1 ) The local interactions comprise bonds , angles , and dihedral terms , ( 2 ) ( 3 ) ( 4 ) The nonbonded terms ( 5 ) are chosen as follows . The distances are determined both by the type of residue-residue interaction , i . e . their color , and their relative distance along the chain . The value of for a particular pair of residues is extracted from a probability distribution . For each possible interaction , three different distance histograms are constructed: one for , one for , and one for , based on the frequency of the Cα - Cα distances between residue pair types in native structures in the PDB . The value of is then taken from the appropriate histogram for the two residues in question . The well depth for each pair of residues , and a factor that determines if the interaction is attractive or purely repulsive , or 1 , is different for each type of interaction , independent of their relative distance along the chain . These values are determined through an iterative procedure using Monte Carlo simulated annealing and perceptron learning [58] to maximize the energy gap between the native structure of the protein and similar globular misfolded states [35] , [59] , [60] . See Das , et al . [38] for further details on the general algorithm for the DMC coarse-grained modeling technique . The procedure outlined above was used here , with the crystal structure of the pG state as the native structure . The result is a DMC model for the dark state . In order to arrive at a model for the photo-activated state , i . e . after photon absorption , the nonbonded interactions between the chromophore residue ( CYS69 ) and all of the other residues were set to zero , i . e . , . Turning off these interactions approximates the known behavior of photo-activated WT-PYP , namely that the hydrogen bonding network around the chromophore is disrupted as a consequence of photo-activation . We use the LSDMap [43] technique to understand the free energy landscape mapped out during the DMC coarse-grained simulation . LSDMap is a kernel-based method of approximating numerical solutions to the Fokker-Planck equation . The method takes as input a set of molecular coordinate trajectories and outputs a set of collective coordinates , termed diffusion coordinates , which are ordered in terms of their relative timescales: the first diffusion coordinate corresponds to the barrier crossing with the longest rate , the second diffusion coordinate is the second longest , etc . The LSDMap code is available via SourceForge [61] . We use the reconstruction algorithm for coarse-grained structures ( RACOGS ) [47] in order to recover the all-atom configuration of candidate signaling structures from the DMC coarse-grained model . This method was developed to provide physically realistic reconstructions in any region of configuration space ( i . e . not only near or within the native-state basin ) . The technique involves several steps: 1 ) backbone reconstruction , 2 ) side chain placement , 3 ) side chain minimization , 4 ) addition of hydrogens , and 5 ) all-atom minimization . This minimization in step 3 detects and performs energy minimization directly on high-energy side chains . This has the effect of reducing any bias that might be present from using a rotamer library taken from native or near-native configurations , therefore making the method more likely to produce realistic structures outside of the native basin . See reference [47] for further details . The vacuum structures resulting from the RACOGS algorithm are solvated with water and Na+ counter ions in GROMACS [53] , using the same force field ( Gromos96 43a1 [54] , Simple Point Charge ( SPC ) water model [62] ) , and general topology as previous work [32] . The system is equilibrated as in the initial preparation of reference [34] . Of the approximately 1000 structures reconstructed , the lowest energy configurations after equilibration were used to initiate DM-d-MD calculations described below . We use the recently proposed diffusion map-directed-molecular dynamics ( DM-d-MD ) [52] procedure to explore the region of configuration space around the reconstructed all-atom configurations . DM-d-MD is an iterative enhanced sampling method in which a swarm of short molecular dynamics simulations are performed at each iteration , a diffusion map calculation is performed on the resulting trajectories , and the furthest point from the swarm is determined from the first diffusion coordinate . This furthest point ( the “frontier” point ) is then used to initialize the swarm for the next iteration . By restarting the next iteration from the frontier point , the technique significantly increases the likelihood that the system will escape from local free energy minima .
|
Many protein systems of biological interest undergo dynamical changes on a time scale too long to be modeled using standard computational methods . One example is photoactive yellow protein ( PYP ) , found in several bacterial species . Blue light , potentially harmful for DNA , triggers several structural changes in PYP , eventually resulting in a conformation that changes the swimming behavior of bacteria . This conformation is difficult to investigate , as it is too short lived . In addition , understanding this “signaling state” is computationally difficult because of the long timescale of the transition . We overcome this by constructing a coarse-grained model to rapidly induce transitions to the signaling state . We then reconstruct and further sample the all-atom configurations from these coarse-grained representations . Our results are consistent with all available experimental and computational evidence .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"physics",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"biophysics",
"biophysical",
"simulations"
] |
2014
|
Multiscale Approach to the Determination of the Photoactive Yellow Protein Signaling State Ensemble
|
Lymphatic filariasis is widely endemic in Myanmar . Despite the establishment of an elimination program in 2000 , knowledge of the remaining burden of disease relies predominantly on programmatic information . To assist the program , we conducted an independent cross-sectional household cluster survey to determine the prevalence of filariasis infection , morbidity and mass-drug administration coverage in four townships of the Mandalay Region: Amarapura , Patheingyi , Tada-U and Wundwin . The survey included 1014 individuals from 430 randomly selected households in 24 villages . Household members one year and older were assessed for antigenaemia using immunochromatographic test cards and if positive , microfilaraemia by night-time thick blood smear . Participants 15 years and older were assessed for filariasis morbidity by ultrasound-assisted clinical examination . The overall prevalence of infection was 2 . 63% by antigenaemia ( 95% confidence interval ( CI ) 1 . 71–4 . 04% ) and 1 . 03% by microfilaraemia ( 95%CI 0 . 59–1 . 47% ) . The prevalence of hydrocoele in adult males was 2 . 78% ( 95%CI 1 . 23–6 . 15% ) and of lymphoedema in both genders was 0% ( 95%CI 0–0 . 45% ) . These results indicate the persistence of filarial infection and transmission despite six rounds of annual mass drug administration and highlight the need for further rounds as well as the implementation of morbidity management programs in the country .
Lymphatic filariasis ( LF ) remains a major cause of permanent disability in tropical and sub-tropical countries[1] . Chronic infection with mosquito-transmitted filarial worms leads to lymphatic dysfunction , resulting in progressive , irreversible swelling of the limbs , breasts or genitals . This disfigurement causes significant pain , disability , social stigma and economic sequelae for sufferers . Worldwide , the South-East Asian Region has the highest burden of LF , accounting for half of the lost disability-adjusted life years globally . [1 , 2] Within the region , Myanmar is one of the worst affected countries . [3 , 4] In response , Myanmar established a National Program to Eliminate LF ( NPELF ) in 2000 as part of the World Health Organization’s ( WHO ) broader Global Program to Eliminate LF . The NPELF drew upon historical reports , national data and a 1997 WHO multi-country prevalence mapping survey to define endemic districts ( the implementation unit ( IU ) used in the country ) . [5 , 6] Forty-five of the 65 districts in Myanmar were classified as LF-endemic , with 47 million people at-risk ( 85 . 5% of the total population ) . [4 , 5 , 7] Baseline , pre– mass drug administration ( MDA ) , surveys indicated high levels of antigenaemia ( 20–30% ) in the central and western dry zones , consistent with that seen in Myanmar migrants living in Thailand . [3 , 5 , 6] Meanwhile , the northern , eastern and southern areas of the country were less endemic or free from filariasis . [5 , 6] In Mandalay Region , the site of this study , the baseline prevalence by microfilaraemia ( mf ) was 5 . 2% ( range: 0 . 2 to 14 . 7% ) and by antigenaemia ranged from 2 to >25% . [5–7] The parasite Wuchereria bancrofti is the sole documented cause of LF in Myanmar , where it is transmitted by the mosquito Culex quinquefasciatus . [3 , 5 , 7 , 8] Under the first component of the Global Program to Eliminate LF , Myanmar began conducting annual , single-dose MDA with albendazole and diethylcarbamazine ( DEC ) in 2001 . [4 , 5] Since then , the coverage has been progressively expanded , reaching all endemic districts by 2015 . [7] With the exception of six districts ( three in Sagaing Region , two in Mandalay Region and one in Magway Region ) that have passed Transmission Assessment Surveys ( TAS ) , all are continuing MDA rounds . [5 , 7] Whilst the WHO recommends annual MDA rounds for at least five consecutive years , several challenges resulted in two to three year intervals between rounds . At the national level , problems in the supply of DEC precluded MDA in 2005 and 2008 , whilst financial and logistical issues later prevented MDA in 2012 . [4 , 5 , 7] Additionally in the Mandalay Region , MDA did not occur in 2006 because of the reported incidence of serious adverse reactions during the preceding rounds , and in 2010 due to further DEC supply issues . [5] As a result , six non-consecutive rounds of MDA had been completed in Mandalay Region over an 11-year period from 2004 to 2014 inclusive . Although MDA has been or is now being conducted in all endemic IUs in Myanmar , data on treatment coverage and the impact of MDA on infection prevalence have been lacking . Two papers ( Win et al . 2018 and Aye et al . 2018 ) recently summarised NPELF data on infection prevalence and treatment coverage surveys . [5 , 7] They report that whilst MDA rounds were interrupted in several districts , treatment coverage has been consistently high ( 60 . 0% to 98 . 5% ) , with significant reductions in prevalence in most areas between 2001 and 2016 . In Mandalay Region , mean mf prevalence in sentinel sites has decreased from 5 . 0% in 2003 to 1 . 2% in 2014 , although some districts have demonstrated persistently high levels of mf prevalence . This NPELF data provides a useful overview of Myanmar’s progress toward LF elimination , but the limitations of programmatic data are acknowledged . Reliance on repeated sampling of sentinel sites in historically high suspected prevalence areas could have over- or under estimated current prevalence and program impact given the focal distribution of LF , while programmatic data is of unknown reliability for assessing MDA coverage . [9] The program has indicated the need for in-depth studies in problem areas to independently verify Myanmar’s progress toward LF elimination . [5] Information on the burden of LF-related morbidity in Myanmar is incomplete . A 2004 NPELF questionnaire provides the only prevalence figures from within the country . [10] The survey assessed self-reported morbidity in 280 , 000 individuals in Sagaing Region and identified 520 cases of lymphoedema ( 0 . 19% ) and 827 cases of hydrocoele ( 0 . 59% ) . Such questionnaires , however , are not sensitive or specific . [11 , 12] Studies of Myanmar migrants living in Thailand provide the only other indication of the disease prevalence . [13 , 14] A meta-analysis of this data suggests that there is likely to be a more significant burden of hydrocoele ( 12 . 27% 95%CI 6 . 84–18 . 90 ) within the population . [3] Whilst these studies provide an indication of prevalence within the country , they are not representative and do not provide any information on lymphoedema . As a result , there remains no representative data on the prevalence of LF-related morbidity in Myanmar . In light of this , a cross-sectional survey in one of the country’s priority areas was conducted . The primary objective was to determine the prevalence of LF infection and morbidity . The secondary objectives were to assess participation in the MDA program and to explore risk factors for infection , morbidity and MDA non-participation . Prevalence data and MDA participation are reported here , whilst analysis of risk factors for infection and MDA non-participation will be reported separately .
Myanmar ( formerly Burma ) is a lower middle-income country in Southeast Asia with a population of 52 million people . [16] The country is administratively divided into a capital territory ( Naypyitaw ) , seven states and seven regions . These fifteen administrative areas are further divided into districts , townships , cities , towns , village tracts ( groups of adjacent villages ) and villages . A cross-sectional , population-based household survey was conducted between February and March 2015 in four townships of the Mandalay Region in central Myanmar: Amarapura , Patheingyi , Tada-U and Wundwin . Townships have a typical population of between 50 , 000 and 250 , 000 people . The NPELF identified these townships as priority areas because they were historically areas of high prevalence . The townships are located directly to the north ( Patheingyi ) and south ( Amarapura , Tada-U and Wundwin ) of Mandalay , which is Myanmar’s second largest city ( Fig 1 ) . Amarapura , Patheingyi and parts of Tada-U Township lie close to the Irrawaddy River and its tributaries , whilst the Samon River passes through Wundwin . Amarapura also lies adjacent to the large Taung Tha Man Lake . The Mandalay Region itself is situated in the low-lying and dry central plains of Myanmar . The area has two distinct seasons: the wet season from May to October ( temperature range 25 . 9–28 . 6 °C ) and dry season from November to April ( 19 . 1–22 . 8 °C ) . [17] The majority of rainfall occurs during the wet season ( average: 140 to 197mm per month ) . [17] As of 2014 , the combined population of the four townships was 869 , 730 ( Amarapura: 237 , 620; Patheingyi: 263 , 730; Tada-U: 138 , 620 and Wundwin: 229 , 760 ) , accounting for 1 . 56% of the total population of Myanmar . [18] The vast majority of this population lives rurally in villages ( 85% ) , where the predominant occupation is farming . [18] A two-stage random cluster sampling method was used to select study participants . The sample size calculation was based on an estimated prevalence of morbidity in adults older than 15 years of 5% . Using a desired precision of 3% , an estimated design effect of two , and allowing for a non-response rate of 10% we predicted a target sample size of 880 . With an average of five people per household and using a cluster size of 10 households , 234 households in 18 villages were required . To ensure the target sample size was met , we sampled 26 villages with an estimated total sample size of 260 households and 1300 individuals . The 26 villages were selected by systematically sampling with random start from a list of all villages in the four townships . [18] The populations of villages were not available at the time of selection . In Myanmar , villages are organised into groups of 10 households ( Sae Eain Su ) . All Sae Eain Su in selected villages were numbered on folded papers . The village leader then picked one from a hat . All individuals one year and older in households within the selected Sae Eain Su were then invited to participate . In the first two villages , one Sae Eain Su was selected per village , but the proportion of participants per household was lower than anticipated so two Sae Eain Su were then picked from each of the remaining 22 villages . To ensure maximum participation , absent households were re-visited a second time on the same day . Households that remained vacant or refused were not reselected . Data collection from consenting participants included three components: an infection survey , a morbidity survey and a risk factor questionnaire . Collected data was entered into an Excel database and analysed using STATA version 14 ( STATA Corporation , Texas , USA ) . Percentage prevalence estimates with 95% confidence intervals ( 95%CI ) were generated for infection and morbidity data . It was assumed that antigen negative individuals were also mf negative , and mf prevalence was calculated accordingly . Prevalence estimates were weighted for age and gender ( using the census population distribution ) , weighted for sampling probability ( using the number of Sae Eain Su and population of each village ) , and adjusted for clustering effect ( using the svy command in STATA ) . Infection and morbidity case estimates were generated using adjusted prevalence estimates with 2014 census population data . The prevalence of infection and morbidity prevalence was classified as low , medium , high or very high as outlined previously . [3] Maps of LF prevalence and MDA participation were created with open source base-layers from Geonode/MIMU ( Myanmar Information Management Unit ) and the World Borders Dataset of ThematicMapping . org using ArcGIS 10 . 5 ( ESRI , Redlands , CA ) . [26 , 27] Univariate analyses with χ2 and t-tests were used to compare variables . P-values less than 0 . 05 were considered significant . Odds-ratios ( OR ) were generated using univariate logistic regression . Incomplete data were excluded from analyses . The study was approved by The Human Research Ethics Committee , James Cook University ( approval H5849 ) and The Ethics Review Committee on Medical Research Involving Human Subjects , Department of Medical Research , Myanmar Ministry of Health . The Ministry of Health , Regional Health Director and village leaders provided permission for this study . Written informed consent was obtained from all participants or their guardians .
Fig 1 shows the location of the four townships within the Mandalay Region . Table 1 outlines the demographic characteristics of the source population and study sample . Two selected villages located in the mountains of eastern Patheingyi Township had to be excluded because of poor quality roads that are only accessible by motorcycle year-round . A total of 1014 individuals from 430 households in 24 villages participated in the study . Twenty households ( 4 . 4% ) were excluded from the study: 15 were absent and five declined to participate . The median number of inhabitants per household was 4 people ( interquartile range ( IQR ) : 3 , range: 1–12 ) with an average household monthly income of 30 , 000 Myanmar Kyat ( 18 . 7 USD ) per person ( IQR: 31 , 250 , range: 3 , 000 to 375 , 000 ) . The median age of participants was 36 years ( IQR: 30 , range: 1–86 ) with no significant difference in age between genders . There were significantly more females than males in the sample ( 63 . 7% vs 36 . 3% , p<0 . 001 ) . Table 2 outlines the prevalence of infection in the 1001 tested participants by age , gender and township . Fig 2 demonstrates the adjusted antigenaemia prevalence by age and gender . S1 Table shows crude and adjusted infection and morbidity prevalence estimates by village . The overall weighted and adjusted prevalence estimate for antigenaemia was 2 . 63% ( 95% confidence interval ( CI ) 1 . 71–4 . 04 ) . Using antigenaemia as a marker of active infection [28] , this represents an estimated 19 , 637 ( 95%CI 12 , 768–30 , 162 ) individuals with LF infection within the study area . The median age of infected individuals was 46 ( IQR: 23 , range: 8–81 years ) . Of the 145 individuals born since the commencement of MDA ( less than 12 years old ) , one was antigen positive . Six of the 1001 participants ( 0 . 60% ) were re-tested because of an equivocal initial result . Of the repeat tests , four were negative and two were positive . Antigenaemia prevalence increased with age ( adjusted ( a ) OR = 1 . 03 per year , 95%CI 1 . 01–1 . 05 , p = 0 . 008 ) and was higher in males ( aOR 2 . 34 , 95%CI 1 . 03–5 . 36 , p = 0 . 044 ) and in Amarapura Township ( aOR 9 . 96 , 95%CI 2 . 98–33 . 29 , p = 0 . 001 using Wundwin as reference ) . Microfilariae were found in 39 . 02% of antigen positive individuals , representing an overall adjusted mf prevalence estimate of 1 . 03% ( 95%CI 0 . 59–1 . 47% ) ( assuming all antigen negative persons were also mf negative ) . The median age of microfilaraemic individuals was 49 ( IQR: 20 . 5 , range: 27–81 ) . Blood slide examination identified only W . bancrofti . Geometric mean intensity of infection was 122 microfilariae per mL ( range: 17–2867 microfilariae per mL ) . Fig 3 demonstrates the distribution of infection across the sampled villages . Antigen positive individuals were identified in all townships and in 12 of the 24 villages ( 50% ) . Participants with mf were found in all three townships and five of the 10 villages where thick blood films were taken . The study identified a cluster of five villages in western Amarapura and northern Tada-U townships where the mean village-level antigenaemia prevalence was 13 . 32% ( range: 8 . 12–17 . 33% ) . By contrast , the mean prevalence outside the cluster was only 1 . 14% ( range: 0 . 00–5 . 91% ) . Table 3 shows the prevalence estimates for chronic LF morbidity and S1 Table gives a breakdown by village . Of the 824 participants ( 99% of those eligible ) aged 15 years or above who were examined for signs of clinical LF morbidity , none were experiencing acute dermatolymphangioadenitis at the time of assessment ( unadjusted 0% , 95%CI 0–0 . 45% ) . Eleven cases of bilateral lower limb oedema were found , however none were clinically considered to be LF-related lymphoedema ( unadjusted 0% , 95%CI 0–0 . 45% ) . As the prevalence was 0% , an adjusted prevalence estimate could not be generated . Fifteen of the 269 males who underwent ultrasound-assisted scrotal examination ( 92% of those eligible ) had an LF-related hydrocoele . Three cases of likely non-LF related hydrocoele were excluded: two participants stated they were present from birth , and one reported it was precipitated by trauma . After adjustment , the overall prevalence estimate for hydrocoele in males over 15 years was 2 . 78% ( 95%CI 1 . 23–6 . 15 ) representing approximately 8 , 505 cases in the study area . The median age of individuals suffering from LF-related hydrocoele was 55 ( IQR: 12 , range: 48–71 years ) . Nine of the participants with a hydrocoele ( 60% ) were asked about surgery , with two ( 22% ) having undergone previous hydrocoelectomy . Higher prevalence of hydrocoele was associated with increased age ( aOR 1 . 05 per year , 95%CI 1 . 03–1 . 08 , p = 0 . 000 ) and with residing in Amarapura Township ( aOR 11 . 11 ( 1 . 85–66 . 89 , p = 0 . 011 , Wundwin as reference ) . The presence of hydrocoele was not significantly associated with antigenaemia ( p = 0 . 088 ) or mf ( p = 0 . 076 ) . Fig 3 and S1 Table show the distribution of the hydrocoele burden at the village level . Cases of hydrocoele were identified across all townships and in 8 of the 24 villages ( 33% ) . Three villages in Amarapura Township ( all within the ‘high infection prevalence’ cluster ) demonstrated substantial prevalence of hydrocoele between 21 . 53 and 36 . 81% . Fig 4 depicts the severity of hydrocoele cases . Eleven cases were unilateral ( 73% ) and four bilateral ( 27% ) , representing a total of 19 hydrocoeles . Seventy-four percent were early stage ( I or II ) . One case of bilateral stage III hydrocoele also had scrotal lymphoedema ( elephantiasis ) . No stage IV–VI cases were observed . Fig 5 demonstrates the number of times participants reported being visited by the National LF program and taking medication during the MDA rounds . Ninety-five percent of households ( 95%CI 92 . 20–96 . 30% ) reported being visited by the MDA program at least once . Of these , the mean number of visits was 2 . 59 ( 95%CI 2 . 44–2 . 74 ) . Ninety percent of household members reported being present during the last MDA round in 2014 ( 95%CI 86 . 54–92 . 73% ) . Eighty percent ( 95%CI 73 . 88–84 . 60% ) of participants reported taking MDA medication at least once . Of those , the mean number of times MDA medication had been taken was 2 . 73 ( 95%CI 2 . 60–2 . 86 ) and 94 . 25% ( 95%CI 91 . 23–96 . 28% ) had taken it in the last year . Based on this , the calculated drug coverage ( i . e . proportion of the eligible population who consumed MDA medication ) for the study population during the 2014 MDA was 78 . 76% ( 95%CI 72 . 71–83 . 76% ) . Two hundred and twenty-nine participants ( 24% ) reported taking MDA medication on more occasions than their household head stated they were visited by the NPELF . When those participants were excluded , the mean frequency of medication consumption reduced to 2 . 37 . Similarly when the frequency of ingestion was capped by the number of household visits , the mean was 2 . 28 . Alternatively , if the number of household visits was raised to match the number of times medication was ingested , the mean number of household visits increased to 2 . 94 . Self-reported MDA participation was lower in males ( aOR: 0 . 62 95%CI 0 . 43–0 . 91 , p = 0 . 017 ) , among antigen positive individuals ( aOR 0 . 27 , 95%CI 0 . 11–0 . 66 , p = 0 . 006 ) and residents of Amarapura ( aOR 0 . 55 , 95%CI 0 . 33–0 . 93 , p = 0 . 027 , with Wundwin as reference ) and Tada-U ( aOR 0 . 54 , 95%CI 0 . 33–0 . 91 , p = 0 . 022 ) townships . There was no association between having ever taken MDA medication and age ( p = 0 . 739 ) . Fig 3 and S1 Table show the proportion of the population that had taken MDA medication at least once by village . In five villages ( 20 . 8% ) , less than 65% of the population had ever taken medication . Of these , three villages were in the ‘high infection prevalence’ cluster . Ten percent ( 95%CI 5 . 08–17 . 41 ) of participants had declined to participate in taking MDA medication on at least one occasion . The main reasons given were a fear of adverse drug effects ( 69 . 72% ) or a mistaken belief that hypertension was a contraindication to MDA medication ( 24 . 08% ) . Three percent of individuals reported adverse effects from MDA medication . The most common reported symptom was dizziness ( 95 . 65% ) .
This representative household cluster survey of LF prevalence and morbidity in four townships of the Mandalay Region , Myanmar has reported the persistent prevalence of LF infection and estimated the burden of LF-related morbidity following six rounds of MDA . It also demonstrated suboptimal treatment coverage and adherence across six rounds of MDA . Further , sequential MDA rounds with adequate coverage will be required to achieve LF elimination . With improved MDA and subsequent reductions in infection prevalence , the incidence of new hydrocoele cases should decline over time . The existing hydrocoele burden highlights the urgent need for a morbidity management program in Myanmar .
|
Lymphatic filariasis ( LF ) is a mosquito-transmitted worm infection that causes chronic and progressive swelling of the limbs ( lymphoedema ) and testis ( hydrocoele ) . Over time this swelling results in significant disfigurement and disability for sufferers . In 2000 , Myanmar commenced a National Program to Eliminate LF through the annual mass drug administration ( MDA ) of two de-worming medications: albendazole and diethylcarbamazine to all endemic districts . However there remains little reliable information on how well the elimination program is working , and how large the burden of disease is in the country . This study assessed the prevalence of LF and participation in the MDA program in 24 villages across four townships of Mandalay Region in central Myanmar . The study found ongoing filarial infection suggesting that the six rounds of MDA had not been sufficient to stop LF transmission in the area . It also identified a substantial burden of hydrocoele but no cases of lymphoedema . These results suggest that further rounds of effective MDA are required to halt LF transmission and highlight the urgent need for morbidity management programs in the country .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"geographical",
"locations",
"tropical",
"diseases",
"parasitic",
"diseases",
"health",
"care",
"filariasis",
"signs",
"and",
"symptoms",
"pharmaceutics",
"myanmar",
"drug",
"administration",
"neglected",
"tropical",
"diseases",
"medical",
"risk",
"factors",
"edema",
"lymphatic",
"filariasis",
"morbidity",
"public",
"and",
"occupational",
"health",
"epidemiology",
"people",
"and",
"places",
"helminth",
"infections",
"health",
"statistics",
"diagnostic",
"medicine",
"asia",
"drug",
"therapy"
] |
2018
|
The prevalence of lymphatic filariasis infection and disease following six rounds of mass drug administration in Mandalay Region, Myanmar
|
Since we still know very little about stem cells in their natural environment , it is useful to explore their dynamics through modelling and simulation , as well as experimentally . Most models of stem cell systems are based on deterministic differential equations that ignore the natural heterogeneity of stem cell populations . This is not appropriate at the level of individual cells and niches , when randomness is more likely to affect dynamics . In this paper , we introduce a fast stochastic method for simulating a metapopulation of stem cell niche lineages , that is , many sub-populations that together form a heterogeneous metapopulation , over time . By selecting the common limiting timestep , our method ensures that the entire metapopulation is simulated synchronously . This is important , as it allows us to introduce interactions between separate niche lineages , which would otherwise be impossible . We expand our method to enable the coupling of many lineages into niche groups , where differentiated cells are pooled within each niche group . Using this method , we explore the dynamics of the haematopoietic system from a demand control system perspective . We find that coupling together niche lineages allows the organism to regulate blood cell numbers as closely as possible to the homeostatic optimum . Furthermore , coupled lineages respond better than uncoupled ones to random perturbations , here the loss of some myeloid cells . This could imply that it is advantageous for an organism to connect together its niche lineages into groups . Our results suggest that a potential fruitful empirical direction will be to understand how stem cell descendants communicate with the niche and how cancer may arise as a result of a failure of such communication .
Stem cells offer exciting potential for regenerative therapy , with ultimate possibilities being the ability to regenerate limbs and heal genetic diseases [1] , [2] . Although studies have begun to address these issues , much work remains to be done [3] , [4] . Indeed , much of our knowledge of stem cells is derived from in vitro experiments , where the stem cells have been relocated from their native environment . For instance , in haematopoietic ( blood-producing ) stem cell experiments the stem cells are often isolated from a donor , expanded in vitro , and transplanted into a lethally irradiated host , with the question of interest being how the stem cells respond to this new environment ( e . g . , [5] ) . However , it is difficult to draw conclusions about the role and behaviour of stem cells in vivo , when experimentally we must investigate them in foreign environments [6] , [7] . Thus , theoretical models of stem cell systems are valuable tools , allowing us to think about stem cells in their native environments when this cannot yet be done experimentally . In vivo , stem cells are generally found in special microenvironments , or niches , which are defined by a complex set of biochemical and physical conditions that feed back on each other [2] , [8] . Niches play a critical role in the function and behaviour of stem cells [2] , [9] . For instance , experimentally changing certain niche attributes affects the dynamics of the stem cells inside them [10] . In addition , stem cells are often not single entities that exist independently of each other , but instead form an interacting population that includes stem cells and their more differentiated products , both within and outside the niche [11] , [12] . Moreover , even separate niches can affect each other , for instance through the effects of their daughter cells or migration ( e . g . , [13] ) . We focus on modelling the haematopoietic stem cell ( HSC ) system , for two reasons . Firstly , it is probably the most well-characterised stem cell system; secondly , it is representative of stem cell systems in general , incorporating their essential properties such as self-renewal , differentiation , multiple lineage choices and feedbacks to regulate cell populations [9] , [14] . This allows us to start thinking about heterogeneity and the introduction of population interactions in a comparatively simple setting [15] . It seems that there are a minimum of two distinct niche types in bone marrow , although their relationship to each other is not fully clear , nor has their connection to the different primitive cell types been unambiguously elucidated [16]–[20] . Spatially , the HSCs themselves are spread throughout the bone marrow ( as well as certain other organs , such as the liver and spleen ) , each in its own individual ‘facultative niche’ [17] , [21]–[24] . To be precise in our definition , henceforth we refer only to these facultative niches as ‘niches’ . Bone marrow thus contains an entire population of niches , with each niche containing small numbers of HSCs , and these HSCs can differentiate into blood cells , which eventually join the bloodstream . The HSC system operates by demand control [25]: there is a target level of differentiated blood cells , the homeostatic level , which is set by natural selection [15] , [26] , [27] , and which the organism attains by differentiation of the HSCs and blood progenitor cells into appropriate differentiated blood cell types [27] , [28] . This seems to be achieved by feedback from the differentiated progeny of the HSCs in the bloodstream [28]–[30] . In addition , there is also feedback from differentiated progeny that have not entered the bloodstream , but remain localised to the niche [12] . The HSC system must respond rapidly to perturbations such as wounding or infection , and even under normal conditions the blood cell turnover of an average human being is around one trillion cells per day [31] . Such enormous numbers mean that it is important to have a robust feedback mechanism for proper functioning of the system . The complex nature of the HSC system , with different blood cell types and feedbacks , as well as many spatially separate niches , means that it is difficult to model . In general , current models of stem cell dynamics involve either only one focal stem cell , or a homogeneous population of each cell type , and are modelled using ordinary differential equations ( ODEs ) [15] . Although such models can give useful results , it is important to include heterogeneity in the picture [32] . For example , there is considerable heterogeneity between individual stem cell clones [33] , [34]; this heterogeneity is also present within clonal cell lines [35] , [36] , and was even observed many years ago by Till et al . [5] , as well as by Suda et al . [37] . However , in the intervening decades the deterministic view of stem cell differentiation has taken hold with great success and has led towards understanding the feedback between differentiated and primitive cells [28] , [38] . More recently there has been a shift in emphasis , with stochastic models being used to examine the dynamics and the evolution of mutations in a stem cell population [39] , phenotypic equilibrium in a cancer cell population [40] , and the effects of different control mechanisms on stem cell populations [41] , [42] . Two of us have already proposed a population biology framework for stem cell dynamics , with the theme “stem cell biology is population biology” [15] , [27] . We used an ODE model of one niche lineage to show how evolution affects the decision of whether to differentiate into myeloid or lymphoid cells . In this paper , we expand on this framework by considering the stochastic dynamics of a heterogeneous metapopulation of niche lineages , comprised of stem , progenitor and differentiated blood cells . For simplicity , we restrict our study to intrinsic heterogeneity only ( that is heterogeneity arising in a clonal cell population in an identical environment ) . We take into account the further consideration that while the niches ( containing the primitive cells ) may be distinct , the blood cells are mixed in the bloodstream , and the niche lineages could be controlled by feedback from the entire bloodstream rather than just their own , possibly localised , descendants . Thus we couple together separate niche lineages , allowing them to interact with each other through their differentiated progeny . Our main aims in this paper are to 1 ) establish the stochastic framework , 2 ) investigate the dynamics of the stochastic system , 3 ) explore how coupling niche lineages together into niche groups affects the system dynamics , and 4 ) whether it has any effect on the response of the entire system to a perturbation . We first develop the stochastic modelling framework . Since stochastic simulations can be slow , we introduce a fast , approximate method for simulating an entire metapopulation of HSC niche lineages . We then describe how to take into account the interactions ( feedbacks ) from the differentiated blood cells on to the primitive cells in the niche ( stem and progenitor cells ) in our simulations . We simulate a metapopulation of lineages through time , which first settles to homeostasis and is then perturbed by reducing blood cell numbers . After the perturbation , there is a peak in blood cell numbers as the stem and progenitor cells replenish them . We investigate the effects of coupling niche lineages together: that is , what happens when the feedbacks are averaged across many niche lineages ( the number of niches averaged over is called the ‘niche group size’ ) . We find that 1 ) coupling niche lineages shifts the mean cell populations at steady state , and changes the shape of the cells’ distributions; 2 ) as more lineages are coupled together , the total blood cells in each coupled niche group approach the target steady state of the system; 3 ) different perturbation types elicit a different response from the system , and when blood cells are perturbed randomly , niche lineages coupled into larger groups respond better than smaller groups and uncoupled lineages . Taken together , these results imply that for the organism , connecting the individual niche lineages into larger niche groups is advantageous , both for optimal regulation of the overall system and for responding to random perturbations .
We begin with the model of the HSC system as developed by Mangel and Bonsall [27] , which characterises the stem cell niche and its products as a control system driven ultimately by demand from the organism ( Fig . 1 ) . The system consists of a HSC niche , containing stem and progenitor cells , and its fully differentiated progeny cells in the bloodstream . The demand from the organism occurs via changes in the levels of differentiated blood cells , which feed back this demand to the primitive ( stem and progenitor ) cells . Specifically , the model is comprised of the populations of stem cells ( S ) , multipotent progenitor cells ( MPP ) , common lymphoid and common myeloid progenitor cells ( and , respectively ) and their fully differentiated products , lymphoid and myeloid blood cells ( and , respectively ) . Although there are many differentiated blood cell types ( see , for example , [14] ) , here we classify them as myeloid and lymphoid types for the sake of simplicity . Thus our model has six state variables , to correspond to the population of each cell type , with certain transitions allowed between the states: self-renewal via either symmetric or asymmetric division; ( symmetric ) differentiation; multiplication or differentiation into or , i . e . either the lymphoid or myeloid route , with relative probabilities and , respectively ( see below ) ; and differentiation into or , respectively; in addition , all cell types can die . In [27] , these transitions are written down as a set of ODEs ( also given in Supporting Text S1 , Section 1 ) , which give the rate of change of each state in time as a function of the current state . Here , we use the stochastic version of this model , given by formulae for each transition between the states , which occur probabilistically ( Table 1 ) . The model also incorporates four different feedbacks from the blood cells and on to the and cells . Three of these , and , take the form ( 1 ) where their respective parameters are defined in Table 2 . These inhibit the activity of and when blood cell levels are high . Specifically , inhibits all activity ( both self-renewal and differentiation ) , inhibits symmetric differentiation only and inhibits all activity . The form of Eq . ( 1 ) is based on earlier studies [28] , [38] , and conforms to the assumptions that: 1 ) numbers of both blood cell types have an effect on and activity , 2 ) their effects are additive , 3 ) the strength is different for and cells , and 4 ) when numbers of either fall , the activity of and increases again . Note that feedbacks always take values on . The last feedback is perhaps the most interesting , and is one aspect that differentiates this model from previous work . We refer to it as the Multipotent Progenitor Commitment Response , or MPCR [27] . This feedback determines the probability of an cell differentiating into either the lymphoid or myeloid routes . The idea behind this is that when blood cell numbers are not at their homeostatic levels ( defined as a specific target value of ) , the MPCR aims to shift the production of new blood cells to the appropriate type . We model the MPCR as ( 2 ) where and are positive parameters . When either or ( states that are not reached in practice by the deterministic model , but do occur in the stochastic model ) this causes a problem in Eq . ( 2 ) , so in this event we simply treat or , respectively , for the purposes of evaluating ; this has the advantage of affecting the value of by only a small amount whilst keeping the MPCR pressure towards the correct cell type . We set the MPCR parameters and to give a target homeostatic blood cell ratio , which here is to loosely correspond to that in humans . To do this , we note that is defined as the probability of an differentiating to a , i . e . at homeostasis we have on average . From this , we can also specify steady states using the blood cell numbers , i . e . as , provided that the differentiation and death rates are identical for both and , as well as and ( however , we examine the general case and use a parameter setup where the death rates of and are not equal , but the only consequence is that the homeostatic state will not be exactly equal to for the chosen ; we explain this issue further in Supporting Text S1 , Section 2 ) . Now , at homeostasis we have . We then substitute these values into Eq . ( 2 ) , choose a value for and so calculate the corresponding . We can do this for different combinations of and , thus varying the strength of the response whilst retaining the same target cell ratio . Although many combinations of and can give the same homeostatic ratio of , they strongly affect the sensitivity of the MPCR to changes in cell numbers and its response to perturbations . In [27] , we used this model to examine the behaviour of the haematopoietic system from an evolutionary perspective . Treating it as a demand control system , where the demand comes from the entire organism , we showed that there is varying selection on organisms with different MPCR parameters and . Different organisms can thus evolve a range of parameters as their environments vary , and this affects the dynamics of their haematopoietic system as well as its response to perturbations . This implies that it is important to take into account the evolutionary background of an organism when examining the dynamics of the haematopoietic system , and stem cell systems in general . This is consistent with the idea that stem cells are units of evolution [43] , [44] . The system of ODEs for the deterministic HSC model ( Supporting Text S1 , Section 1 and Ref . [27] ) can be considered the continuously-conditioned average of the stochastic system [45] . If these ODEs were linear , we could say that they represent the mean of the stochastic system ( that is , the initially-conditioned average: see [45] ) ; however , as they are non-linear due to the feedback functions , we cannot tell a priori the relationship between the deterministic and stochastic solutions ( although having said this , initial explorations of a much simpler stem cell system found the ODE solution to be reasonably close to the stochastic mean in the case of a single lineage with feedbacks [15] ) . In general , ODE models are not able to account for the full range of dynamics of highly stochastic systems , and in extreme cases can even give results that are unrepresentative of the full behaviour of the system [46] , [47] . The stochastic formulation of the ODE model also has six states and fourteen transitions between the states . However , rather than occurring at deterministic rates , these transitions now occur with particular propensities at each step of the simulation . The stochastic simulation algorithm ( SSA ) , developed by Gillespie [48] , allows us to simulate such a system in a statistically exact way . We first describe it in general terms and then discuss its application to the HSC system . In general , we consider a set of types of transitions between kinds of cells . We track cell populations through time with the state vector , where represents the number of cells of type at time and denotes the matrix transpose . We let denote the cell type index and denote the transition index; boldface font represents a vector of size . The SSA is a simple and powerful method , and essentially consists of finding , at each step , the time until the next transition and which transition occurs . To do this , we define the vector of propensity functions , where is the probability of transition occurring in an infinitesimal time , and where represents terms of higher order in ( for further details about the importance of this term , see [49] ) . In addition , we have a stoichiometric matrix of size , which represents how each transition affects the numbers of cells . Knowledge of , and is all that we need in order to simulate the time dependence of the HSC system . The time until the next transition , , is sampled from an exponential random variable with parameter , where This implies that the probability of no transition in the next is , which can be expanded as a Taylor series to . Given that a transition occurs , the probability that it has index is Once these two have been chosen , the state vector is updated as ( 3 ) where is the index of the transition that occurred and The SSA was initially developed to simulate the interactions of different chemical species in a dilute gas , and has since been extended to dilute solutions [50] . Both of these scenarios assume that the system is macroscopically well-stirred and homogeneous . The usual mass-action form of its propensity functions are directly based on these assumptions . In order to use the SSA with the HSC system , which does not necessarily obey either assumption , we adopt instead a phenomenological approach to definining the propensity functions , as is the custom when constructing ODE population models . In effect , we simply convert the transition rates of the ODE system into transition propensities . The form of the propensities depends on our assumptions regarding the processes involved: thus here , the propensities are dependent upon a rate constant , the population of the transitioning cell type , and in the case of stem and progenitor cells , also the feedbacks that we have assumed exist ( Table 1 ) . Note that the propensities give the probability of a reaction occurring per unit time , and therefore are not required to remain on . For our HSC model simulations , we define the state vector as . The SSA framework of the previous section is both simple and statistically exact , meaning that a histogram built up of an infinite number of simulations is identical to the true histogram of the system . However , especially for systems with larger populations ( generally , hundreds or thousands of cells , or more ) , faster transitions or those whose transition rates have a complicated form , it can become slow . For such systems , if computational time is an issue , it is more appropriate to use an approximate method . A common example of such a method is the -leap method [51] , which evaluates many transitions in one ( larger ) step , thereby speeding up computation . The -leap update formula also takes the form in Eq . ( 3 ) , but rather than a single transition , now the number of transitions occurring in each channel over each step , represented by , is given by ( 4 ) i . e . it is a Poisson random number with mean . This approach can greatly speed up computation , although it incurs a loss in accuracy . The stepsize can be varied , and is commonly chosen to be sufficiently small to achieve reasonable accuracy but sufficiently large to increase the computational speed . A simple way of doing this is to bound the change in each cell population over one step , , by a small fraction of . Since is a random variable , in practice this means bounding its mean and standard deviation . can then be chosen to be consistent with these bounds . For the simulations in this paper , we have used a simple version of this scheme ( set out in detail in [52] , specifically , Eqs . ( 32 ) and ( 33 ) ) , without any consideration of reaction criticality . Several similar methods have been proposed with higher efficiency or accuracy ( for example , [53]–[55] ) . Since we introduce additional complexity by simulating an entire metapopulation of lineages and coupling them , here we have chosen to use a simple stepsize-adapting scheme . In order to simulate a large number of niche lineages , we expand the Gillespie SSA/-leap approach from just one sub-simulation ( i . e . , lineage ) to many . By including interaction terms between each individual niche lineage , we can easily simulate an entire interacting heterogeneous metapopulation of niche lineages . The heterogeneity results only from intrinsic noise , that is , noise arising from random thermal fluctuations , which is present even in genetically identical populations in the same environment [35] . Our method almost resembles a compartment-based model , which consists of many discrete spatial compartments , each of which is assumed to be homogeneous inside . However , as details of the spatial aspects of stem cell niches are still emerging , we chose not to explicitly equate each sub-simulation with a discrete spatial compartment; rather , each sub-simulation represents a niche lineage whose physical locations are not taken into account . We take advantage of the native matrix structures of the Matlab programming language , with the state vector of each niche lineage forming one column of the overall state matrix . Thus , if there are separate niche lineages , instead of an state vector , we now manipulate an state matrix . This approach is conceptually simple , easily allows for the introduction of coupling and interactions , and is especially fast ( as Matlab is optimised for matrix calculations , calculating each step of the SSA scheme on a matrix rather than a vector has little effect on the speed , whereas doing the same for each niche lineage in turn would be very much slower ) . This state matrix approach could easily be implemented in other programming languages , and although it would not necessarily result in a large computational speedup ( for instance , this is likely to be the case in the popular programming language C ) , we argue that it is favourable even for its inherent simplicity alone . Since each sub-simulation of the SSA chooses timesteps randomly , the metapopulation of niche lineages would not be simulated in time synchronously , akin to a running race where some runners are ahead and some lag behind . Since we want to simulate an interacting , coupled metapopulation , all lineages must stay in step otherwise the interactions would effectively be averaging over time . The solution is to switch to the -leap method from the previous section , use it to choose a suitable timestep and evolve every niche lineage over this timestep . It is important to note that this does not bias our results in any way: we are only selecting a common timestep for all the lineages , but the reactions that occur in each lineage are then chosen according to the true Markov process . To explain this , let us go back to basics: the evolution of each lineage is governed by a Markov jump process [56] , which is approximated by the -leap method . If we wanted to simulate a population of niche lineages using a standard -leap , we would run repeat simulations of a single lineage . This could be done with either a fixed or an adaptive timestep , and we would sample the Markov process ( carry out the -leap update ) at the time points given by those timesteps . However , the process itself is independent of the times at which we sample it ( although , of course , the same cannot be said for the solution of our approximate -leap method , which approaches the true Markov process as the timesteps decrease ) . Thus we are free to sample the Markov process at whatever time points we choose , provided we remember the condition on our approximate solution . Now , a reasonable part of the computational time of a leaping method is taken up with the overhead of calculating the timestep adaptively . By simulating the metapopulation simultaneously , our method allows us to choose just one timestep for all niche lineages , reducing the total overhead . The only disadvantage is that if one lineage contains unusually large populations , this would pose as a bottleneck on the common stepsize . We must thus find the common limiting timestep from the whole metapopulation . First , the propensities of each transition in each niche lineage are calculated . Then , we find the lineage with the largest , that is the sum of the propensities . Now , we simply continue with the stepsize selection as if we were only simulating a single lineage , and its propensities were those of the selected one . Once the stepsize has been chosen , the entire metapopulation is evolved over that step using Eqs . ( 3 ) and ( 4 ) . We describe this more precisely in Algorithm 1 . Each HSC niche does not exist in isolation in the bone marrow; in fact HSCs often circulate around the bone marrow and bloodstream [57] , [58] . Differentiated blood cells are also , in general , ejected from the niche and enter the bloodstream , although certain differentiated cell types can remain localised to the niche [12] . Thus , cells from each niche lineage are mixed to various degrees after they have fully differentiated and leave the niche . To investigate the dynamics of coupling together separate niche lineages , we introduce the implementation of the coupling . We assume that there is no interaction between cells that are not fully differentiated ( that is , any cell type except for and ) . The coupling comes into effect only through the feedback functions of the and cells on to and cells ( although it should be noted that our computational method can handle any form of coupling ) . To capture this , we create ‘niche groups’ , where the feedbacks on the stem and progenitor cells in each niche lineage depend on the total levels of in the entire niche group of that lineage . In practice , this means that the blood cells in each lineage of a niche group are replaced in the feedback equations by the total in that niche group ( whilst normalising the parameters by the niche group size ) . The propensities for each niche lineage are then calculated as described in the previous section and the populations of each niche lineage updated separately ( Algorithm 2 ) . To aid in visualising this , we give an example using a population of four niche lineages coupled into niche groups of size two , i . e . ( Fig . 2 ) . When the lineages are coupled , the feedbacks are taken over the total , in the respective niche group . Then , denoting by the population of from niche lineage , and similarly for , the feedbacks of the first two niche lineages would be , and the last two would be . This is the case for all feedback functions , including the MPCR . The factor of one half is necessary to normalise the steady states to be directly comparable , regardless of niche group size .
We begin by evaluating the performance of our computational method . Although it is not exact , the -leap is in general a much faster simulation method than the SSA . The error parameter ( introduced in the Fast Stochastic Simulation section ) indicates the amount of error we allow into the leaping approximation . Common values for are of the order of 0 . 01 , meaning roughly that the timestep selected allows at most a 1% change in the population of the rarest cell type; a value of typically corresponds to high accuracy and to low accuracy , but this can vary . We ran simulations of a metapopulation of uncoupled niche lineages with the vectorised -leap method described in Algorithm 1 for a wide range of values of , as well as with a vectorised SSA , and recorded the average runtimes on a standard desktop computer . The SSA can be regarded as finding the exact solution ( for uncoupled niche lineages only — it loses this exactness when the lineages are coupled , see Vectorised -leap section ) . Therefore we compared the probability density functions ( PDFs ) returned by the -leap to the exact PDF given by the SSA to get an idea of how the errors of the -leap simulations changed as the error parameter was varied . The simulation runtimes are listed in Table 3 , as are the total errors of the -leap results . We calculated these by taking the -distance between the weight of each bin ( that is , probability density multiplied by bin width ) of the -leap PDFs and that of the SSA . The runtimes decrease as the error parameters increase , with the SSA taking the longest , as expected . The self-distance of two different SSA simulations is relatively large ( Table 3 , top row ) , indicating that the differences in errors between the -leap with may be due to Monte Carlo error . This means that the vectorised -leap with these error parameters is about as accurate as the SSA . With , however , the -leap does become substantially less accurate . Accordingly , in the rest of our simulations , we used Table 3 shows that the vectorised -leap is indeed faster than the SSA , significantly so when . However , even with , the -leap finds remarkably accurate solutions . This is compounded with the fact that the SSA should not be used to simulate coupled niche lineages , as each lineage proceeds at its own pace . These factors mean that approximate , fast methods that can sample the state matrix synchronously are most ideal for simulating larger , interacting systems such as our HSC system . We then ran simulations of the HSC system on metapopulations of uncoupled and coupled niche lineages for each set of parameters , using our vectorised -leap method from above with . In order to investigate the coupling between different lineages , this was grouped into sub-populations ( for example , 200 sub-populations of niche groups of size 100 ) . The model is not parametrised using any specific data: the parameters in Table 2 are a canonical parameter set , chosen to elucidate general principles rather than make specific biological predictions . Due to the number of parameters , a thorough parameter sweep or sensitivity analysis was beyond the scope of this paper; however , manual experimentation using several parameter sets showed relative robustness in the system dynamics ( for instance , see Supporting Text S1 , Section 3 ) . In one or two cases , we observed consistent oscillations in cell populations , qualitatively similar to Ref . [59]; here , we have used parameters that settle down to homeostatic cell populations . Between and seconds , transitions do not occur faster , as it may seem from some of the plots; not all transitions are recorded , and we have sampled the ones in this time period more often to give an accurate picture of the system dynamics after a perturbation . We elucidate the basic dynamics of the model in Fig . 3 , which shows a stochastic simulation of a single niche lineage along with the ODE model for comparison . We started all our simulations in the state , i . e . with one and no other cells . All cell populations experience an initial surge , which then dies down to a steady state . At seconds , we perturbed the cells by removing 75% of them ( indicated by yellow dashed line; ODE model not perturbed ) . The and surge just after the are depleted , but there seems to be little response from the and cells . Significantly , there is also little response from cells . After around 1000 seconds the cells return to their pre-perturbation numbers , and all three cell types then settle back to their steady states . We set the MPCR parameters to reach homeostasis at the ratio ( corresponding to ) . However , as the death rates of and were not equal , we did not expect to observe this exact homeostatic ratio; indeed , Fig . 3 shows that the homeostatic state of the model using this particular parameter space is around , corresponding to from Eq . ( 2 ) ( see HSC Model section and Supporting Text S1 , Section 2 ) . The ODE model roughly follows the stochastic simulations , with both indicating similar homeostatic states . In Fig . 4A , B , C we show the time evolution of six separate simulations each , of both uncoupled and coupled ( niche group size 100 ) niche lineages . The first thing we notice is that the cells in some lineages die out ( Figs . 4A and S1 ) , but the rest of the lineage keeps functioning ( Fig . S1 ) . Over one quarter of all lineages had lost their by seconds , and this number went up to over one half by the end of the simulations . Only in a handful of these cases did the entire lineage die out; the rest were maintained by the cells . Next , the total numbers per niche group ( , normalised by niche group size; Fig . 4D ) are close but not identical for uncoupled and coupled niche lineages . This is supported by Fig . 4F , where colour indicates numbers and which shows 100 trajectories each of uncoupled and coupled niche groups . The numbers are consistent for all niche groups , and there is also little difference between uncoupled and coupled numbers . In contrast , Fig . 4E highlights the differences between per individual lineage seen in Fig . 4C: uncoupled lineage numbers fluctuate in an uncorrelated way over time and all lineages behave in a similar way , whereas those of coupled lineages show a distinct correlation over their own trajectories , as well as considerable variation between individual niche lineages . Fig . S1 demonstrates that this also happens , to varying degrees , for the other cell types . It is difficult to tell whether this is also the case for , where stochastic fluctuations are large compared to cell numbers , but Fig . S2 helps to clarify the issue: the steady states of the uncoupled and coupled are also fairly close but not identical ( Fig . S2A , C ) , and in Fig . S2B we can make out the distinct lines made by the coupled lineage levels , implying their fluctuations are correlated compared to the uncoupled lineages . To sum up so far , Figs . 4 , S1 and S2 tell us that 1 ) although there is a large surge in numbers , there is a smaller relative response in numbers of ; 2 ) there is also a large surge in numbers to replenish the lost , which corresponds to a modest drop in and numbers followed by a small surge to return to their steady states; 3 ) cell populations in individual uncoupled niche lineages fluctuate considerably with time , whereas those of coupled niche lineages less so; 4 ) however , cell numbers between individual coupled lineages are much more varied than those of uncoupled lineages , which are all roughly similar . Next , we look more closely at the response of the system to perturbations . We examine three types of perturbation: even perturbations ( 37 . 5% reduction of from every niche lineage ) , uneven perturbations ( 75% reduction of from every second lineage only ) , and random , or more precisely , probabilistic , where each lineage has a 50% chance that its are reduced by 75% . The perturbations were chosen to cause , on average , an identical change in cell numbers across the entire population of niche lineages , that is the removal of 37 . 5% of the entire population of . The actual values of 37 . 5% and 75% are illustrative in nature , rather than realistic examples of blood loss from injury . The response of the system to perturbations is given by two main indicators: return time to homeostatic levels , and overshoot/oscillation size , defined as the difference between the maximum of the post-perturbation spike in cell numbers ( and feedbacks ) and their steady states . Return time , much like the homeostatic levels of the system , is dictated by the model parameters . Moreover , it is difficult to accurately measure , as even in homeostasis , there is a continuous turnover of cells , leading to fluctuations in the cell numbers . We did not find a substantial difference in return time between uncoupled and coupled niche lineages for any type of perturbation , and the ODE model and the mean of the stochastic system closely matched in this respect .
Most of the results above were concerned with linking together separate niche lineages into groups . A large niche group size indicates that the feedback from the blood cells ( ) to the primitive cells ( , ) is regulated by a large fraction of the overall blood cell numbers in the organism . We found that as niche group size was increased , the mean levels of , moved closer to the ODE model solutions . This is not a huge surprise: summing the blood cells in each niche group and normalising is equivalent to averaging over niche groups; the larger the niche group , therefore , the less the noise in total cell numbers per niche group , and the closer the system is to the ODE model . This is also a possible explanation for the lower variance of cell distributions in coupled niche groups . This reduction in noise can be useful for biological systems , for which noise is often detrimental . However , the question remained of whether it was the uncoupled lineages or the coupled ones ( and the ODEs ) that better achieved the target cell populations . From the control system perspective that we have taken , good control is defined as regulation of the cell populations to the target ratio . Given the interactions of the MPCR parameters and this ratio in setting the cell steady states ( see Supporting Text S1 , Section 3 ) , it was the ODE solutions , and therefore the coupled niche lineages , that followed the target cell levels more closely than the uncoupled ones . Thus , it seems that on a systemic level , it is advantageous to connect together niche lineages . This hints at some intriguing possibilities for understanding the emergence of tissues , which are interacting populations of single cells . The difference between the overshoots for the three perturbation types can be understood as follows . The even perturbation should result in a similar overshoot from both uncoupled and coupled niche lineages , since it affects all niches equally . This is roughly consistent with our results for , but it is unclear why the overshoot of the uncoupled is considerably lower . The uneven perturbation affects uncoupled and coupled lineages differently , with coupled niches having smaller overshoot , but there is no variation with niche group size . Because it is a regular perturbation , coupling lineages ( into even-sized groups ) reduces the niche group overshoot , and it does not change with niche group size as in every case 37 . 5% of the cells in each niche group are lost . However , random perturbations elicited yet another response . With smaller groups or individual lineages , it is more likely that the entire niche group is perturbed , resulting in a larger overshoot . At the extreme ends of the scale , one could conceivably have one niche group with all niche lineages perturbed , and another with none . As niche group size is increased the chances of this decrease and the percentage of total niche group that is lost tends asymptotically to 37 . 5% , with the overshoot declining to the same levels as for an uneven perturbation . This shows that in environments with even perturbations , it may be advantageous to not couple niche lineages – however such environments are unlikely to occur in nature . In contrast , in natural environments with random perturbations , coupling niche lineages results in a more favourable response . This overshoot of blood cells following a perturbation is an important aspect of our model . There has been little work on this , although experimental studies have found that some types of T-cells are reconstituted very quickly and exceed normal levels , possibly supporting our results [60] , [61] . We do not know of similar results for other blood cell types . An interesting result from our simulations is the large variation we see in cell populations of coupled lineages between different lineages in the same niche group , and the relatively low variation over time of the populations in each lineage . This indicates that the activity of the primitive cells of each lineage varies , with some inactive/less active and others continuously differentiating to produce more cells , in order to achieve the correct homeostatic cell levels , somewhat akin to the HSC subsets found by Sieburg et al . [62] . Although we have not explicitly considered it here , our model also naturally captures the cycling behaviour of HSCs , with periods of quiescence and activity in each lineage [63] . In addition , after a perturbation , our model finds a response from both stem and progenitor cells . This is in agreement with studies finding stem cell activation after injury ( e . g . , [29] ) , but also supports the suggestion that at least part of the response is from progenitor cells [64] . Our results indicate that , in order to regulate blood cell populations tightly and for a less severe response following random perturbations , it is advantageous to the organism to couple haematopoietic lineages together via the feedbacks from blood cells on to primitive cells . There are three biologically-viable possibilities for the nature of this feedback mechanism: lineage-dependent feedback , where the primitive cells in one lineage can only sense numbers of their own differentiated progeny; local feedback , where the primitive cells can sense blood cells of any lineage in proximity to them; global feedback , where all primitive cells can sense all blood cells in the organism . Lineage-dependent feedback would require a biochemical mechanism in which niche lineages ( or niche groups ) can identify signals from their descendants and respond to the demand control from those cells , but not others in the blood; this could imply an epigenetic process . Indeed , studies have found that stem cell daughters of HSCs have a similar lifetime to their parents [34] , and such an epigenetic mechanism could also exist in non-primitive progeny to regulate their feedback . Local feedback implies a spatial constraint on the feedbacks; although this has already been found to exist in the case of certain HSC progeny as well as other niche cells [12] , it may not be a universal mechanism for the haematopoietic system because most blood cells enter the bloodstream rather than localising around the niche . However , in other stem cell systems , it is quite a plausible mode of feedback [65] . Finally , global feedback would require the HSCs to sense every blood cell in the bloodstream . Since it is likely that the feedbacks from the blood cells occur via growth factors [28] , which naturally have a limit on their range of action , it does not seem likely that the HSC system incorporates global feedbacks from all blood cells . More likely is some combination of the above mechanisms . Looking for groups of epigenetic markers shared by HSCs , progenitor cells and differentiated blood cells could be a useful avenue for further experimental work . Finally , as evidenced by the dynamics of our model , the feedbacks are essential for achieving homeostatic cell rates [28] . Although we have not explored this issue further , our results also support the idea that cancers may be a failure of the signalling mechanism and the associated feedback control [66] . In ODE models , we can only account for a single , or at best an identical set of deterministic niche lineages , so that the interactions between a heterogeneous metapopulation of lineages is underexplored theoretically . This is important for two reasons: first , the dynamics of the entire system cannot be determined just by looking at its parts , and second , we can take a much broader point of view by looking at an entire population [26] . Indeed , Huang [32] suggests that this is one of three as-yet-neglected perspectives that should be adopted in stem cell modelling . For example , maintaining homeostasis at the population level can be achieved by several possible strategies [64]; only looking at a single stem cell restricts consideration to just one strategy , asymmetric division , which does not reveal the full picture . A stochastic treatment is needed to be able to incorporate population-level strategies such as a combination of both asymmetric and symmetric division and differentiation . Our work also links with the idea of a potential landscape of cell states [67] ( although here , the axes of the landscape represent not , say , expression levels of a protein , but numbers of cells in each sub-population ) : one simulation represents a niche lineage moving along the landscape and falling into a stable state ( the homeostatic state for that lineage ) , and many simulations , as we have done , could allow us to reconstruct the potential landscape by randomly generating trajectories until we can see its full shape . Thus Monte Carlo simulations offer a computational way to explore the potential landscape . In this paper , we first introduced a fast method of simulating an entire metapopulation of interacting niche lineages ( or cells or biochemical species ) synchronously through time . This is based on a version of the -leap method [51] and then generalised to the metapopulation level . It compares favourably with the popular stochastic simulation algorithm method [48] , both in terms of speed and accuracy – when interactions are to be included , the stochastic simulation algorithm averages them over time , as each member of the population proceeds through time at a different pace . The computational method we have proposed here can be combined with many stochastic simulation schemes in order to allow one to quickly and easily simulate whole metapopulations . Naturally , it is not limited to cell metapopulations , and can be used in any context where we would otherwise use Gillespie's standard SSA to simulate biochemical populations without tracking individual particles . For instance , with no interactions specified , it can be used to simultaneously run many repeat simulations of the same chemical reaction system ( by regarding each sub-simulation as an independent repeat simulation ) , in order to find the full distribution of possible states , arising from intrinsic noise , at some time . However , it is especially useful when we are interested in interacting populations/metapopulations; for instance , this is often the case in ecological systems . It could also be used in condensed matter and chemical physics and in any biochemical context with spatial homogeneity . Finally , it is a very short logical step away from a spatial stochastic model made up of separate compartments ( e . g . , [68] , [69] ) , and this is one obvious extension . We used this method to build upon the haematopoietic stem cell model introduced in [27] , to simulate a heterogeneous metapopulation of haematopoietic stem cell lineages in time . Using this model , we considered the coupling of individual niche lineages into niche groups . We found that the more niche lineages are coupled , the more closely the mean blood cell numbers approached the target cell ratio . Moreover , when perturbations affected each lineage randomly , as would be the case in a natural environment , a larger number of niche lineages being coupled leads to a smaller overshoot in cell numbers , implying a more ideal response . This suggests that it is advantageous for an organism to couple haematopoietic lineages in order to better regulate homeostasis in the haematopoietic system , as well as respond better to natural perturbations . Our work leads naturally on to questions about linking cells into whole tissues [65]; for instance , an obvious question is whether these are evolutionarily favourable compared to single niche lineages ( or cells ) . One advantage might be the ability of larger systems to ‘average out’ excessive noise , as is the case with our coupled niche groups . So far , there are few studies investigating whole populations of stem cells , and even fewer on the consequences of linking them into tissues . It is well-known that HSCs routinely leave the niche and migrate in the bloodstream [19] , [57] , [70] . Using our current model , an easy modification is to allow for this migration into and out of the niches ( which might mitigate the instances of all stem cells in one lineage dying out , as we observed ) . Another extension of our work would be to introduce environmental or even genetic heterogeneity into the picture . Then it becomes possible to investigate the effects of mutations , for instance by introducing niche lineages with different parameters , in a similar way to evolutionary invasion analysis .
|
Stem cells portend great potential for advances in medicine . However , these advances require detailed understanding of the dynamics of stem cells . In vitro studies are now routine and challenge our preconceptions about stem cell biology , but the dynamics of stem cells in vivo remain poorly understood . Thus , there is a real need for novel computational frameworks for general understanding and predictions about experiments on stem cells in their native environments . By implementing a stochastic model of stem cell dynamics , generically based on the bone marrow system , in a novel , fast and computationally efficient way , we show how different couplings of stem cell niche lineages lead to different predictions about homeostatic control . Understanding the demand control of stem cell systems is essential to both predicting in vivo stem cell dynamics and also how its breakdown may lead to the development of cancers of the blood system .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"systems",
"biology",
"developmental",
"biology",
"computer",
"and",
"information",
"sciences",
"computer",
"modeling",
"mathematics",
"stem",
"cells",
"animal",
"cells",
"cell",
"biology",
"theoretical",
"biology",
"applied",
"mathematics",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"population",
"biology",
"computational",
"biology",
"computerized",
"simulations",
"physical",
"sciences"
] |
2014
|
Stochastic Dynamics of Interacting Haematopoietic Stem Cell Niche Lineages
|
The development of new diagnostics is an important tool in the fight against disease . Latent Class Analysis ( LCA ) is used to estimate the sensitivity and specificity of tests in the absence of a gold standard . The main field diagnostic for Schistosoma mansoni infection , Kato-Katz ( KK ) , is not very sensitive at low infection intensities . A point-of-care circulating cathodic antigen ( CCA ) test has been shown to be more sensitive than KK . However , CCA can return an ambiguous ‘trace’ result between ‘positive’ and ‘negative’ , and much debate has focused on interpretation of traces results . We show how LCA can be extended to include ambiguous trace results and analyse S . mansoni studies from both Côte d’Ivoire ( CdI ) and Uganda . We compare the diagnostic performance of KK and CCA and the observed results by each test to the estimated infection prevalence in the population . Prevalence by KK was higher in CdI ( 13 . 4% ) than in Uganda ( 6 . 1% ) , but prevalence by CCA was similar between countries , both when trace was assumed to be negative ( CCAtn: 11 . 7% in CdI and 9 . 7% in Uganda ) and positive ( CCAtp: 20 . 1% in CdI and 22 . 5% in Uganda ) . The estimated sensitivity of CCA was more consistent between countries than the estimated sensitivity of KK , and estimated infection prevalence did not significantly differ between CdI ( 20 . 5% ) and Uganda ( 19 . 1% ) . The prevalence by CCA with trace as positive did not differ significantly from estimates of infection prevalence in either country , whereas both KK and CCA with trace as negative significantly underestimated infection prevalence in both countries . Incorporation of ambiguous results into an LCA enables the effect of different treatment thresholds to be directly assessed and is applicable in many fields . Our results showed that CCA with trace as positive most accurately estimated infection prevalence .
It is estimated that 237 million individuals require treatment for schistosomiasis [1] . Endemic in 56 countries spanning over Africa , the Middle East , South America , and the West Indies , Schistosoma mansoni is the most geographically widespread schistosome species . Nevertheless , despite a growing body of studies looking at distribution ( e . g . [2–4] ) , we do not have a true representation of the number of people infected with S . mansoni as there is no definitive ‘gold standard’ field diagnostic test . Accurate prevalence estimates of those infected are important , as even low infection intensities have associated morbidity [5] . The current recommended control method for schistosomiasis is preventive chemotherapy ( PC ) of at-risk populations , where all school-aged children ( SAC ) and , where appropriate , at-risk adults in the community are treated . Frequency of treatment and who receives the drugs are dependent on the prevalence of schistosomiasis in the local area [6] , as determined by the parasitological diagnostic test Kato-Katz , where eggs are detected in faecal samples examined microscopically [7 , 8] . Kato-Katz has low sensitivity in those with low infection intensities and in areas of low prevalence , as egg output varies both within and between days [9] and infection can easily be missed [10] . However , Kato-Katz is highly specific , as an S . mansoni egg is easily identifiable to a trained technician [11] . A rapid diagnostic test , Circulating Cathodic Antigen ( CCA ) , has recently been endorsed by the World Health Organisation ( WHO ) for use in mapping and programme impact evaluation [12] . CCA uses a urine sample to test for S . mansoni infection and consequently is much simpler to use than Kato-Katz . However , the presence of an ambiguous result between negative and positive , known as a ‘trace’ result , complicates the interpretation of a CCA test . There is no consensus in the literature on whether trace should be considered as positive or negative ( e . g . trace as negative as found to be closest to Kato-Katz:[13] , trace assumed to be negative: [14] , trace assumed to be positive: [11] , [15] ) . Additionally , it is not clear how to interpret prevalence estimates from CCA tests and whether or not they are reflective of infection prevalence in the population . Latent Class Analysis ( LCA ) estimates the sensitivity and specificity of diagnostic tests and the prevalence in the population in the absence of a gold standard , and has been applied in a wide range of fields including human soil transmitted helminthiases [16] , malaria [17] , and veterinary biology [18] . Analysis of LCA can be within a frequentist or Bayesian framework , with the Bayesian approach having several advantages . Firstly , the distribution of additional parameters such as Positive and Negative Predictive Values ( PPV and NPV respectively ) are easily calculated from the posterior distributions obtained from the LCA , which enables the results to be more easily interpreted . Secondly , Bayesian analysis enables straightforward comparison of estimated infection prevalence and the prevalence by each test to assess how well the test performs in estimating prevalence rather than infection status of each individual . Finally , the use of posterior distributions enables straightforward testing of differences between countries and between tests . The aim of this study is to robustly analyse CCA data from two countries , Côte d’Ivoire and Uganda , to determine the effects of considering CCA trace as negative or positive . We particularly focus on assessing the performance of CCA and Kato-Katz at measuring ‘infection prevalence’ , which is estimated from the LCA , and is the main use of S . mansoni diagnostics in control programs .
Ethical approval for both surveys , including the consent process , was obtained from Imperial College Research Ethics Committee ( ICREC_8_2_2 ) as well as from the appropriate country: Comité National d’Ethique de la Recherche ( CNER; ref: 086/MSHP/CNER-kp ) in Côte d’Ivoire and Uganda National Council for Science and Technology ( UNCST; ref: HS1993 ) in Uganda . The surveys were undertaken as part of the national schistosomiasis control programmes in each country , overseen and approved by the relevant Ministries of Health . As participants were under 18 years of age , written consent was required by a parent or informed guardian . Head-teachers in each school acted as the informed guardian as literacy levels in many areas are low . The head-teacher was informed fully about the study and requested to provide informed consent for field teams to collect urine and stool samples from pupils . Only children who consented orally both before and after selection in the presence of a witness ( head-teacher ) took part in the survey . Additionally , all children gave urine and stool samples freely following selection , and there were no consequences if a child did not return their samples . All data were entered and analysed anonymously .
A total of 3 , 035 and 693 children were included in the analysis for Côte d’Ivoire and Uganda , respectively ( Table 1 ) . Prevalence by Kato-Katz was higher in Côte d’Ivoire ( 13 . 4% ) than in Uganda ( 6 . 1% ) , with mean intensity of infection , among all children , being over seven-fold higher in Côte d’Ivoire ( 26 . 8 eggs per gram ( epg ) ) than in Uganda ( 3 . 4 epg; Table 1 ) . However , prevalence by CCA was similar in both countries , both when trace was assumed to be negative ( CCAtn prevalence 11 . 7% in Côte d’Ivoire and 9 . 7% in Uganda ) and when trace was assumed to be positive ( CCAtp prevalence 20 . 1% in Côte d’Ivoire and 22 . 5% in Uganda; Table 1 ) . A completed STARD checklist and participant flow diagram are available in S3 and S4 Supporting Informations respectively and the raw data used for analyses are available in S5 ( Côte d’Ivoire ) and S6 ( Uganda ) Supporting Informations . In both countries , just over 75% of pupils were negative on both tests , and 6 . 5% of children in Côte d’Ivoire and 3 . 8% of children in Uganda were positive on both tests ( Table 1 , Fig 1 ) . CCA sometimes failed to detect infection where eggs were found by Kato-Katz: 4 . 7% of pupils in Côte d’Ivoire and 1 . 4% of pupils in Uganda had a negative CCA result but were positive by Kato-Katz . Similarly , Kato-Katz sometimes failed to detect infections that were positive ( not trace ) by CCA: 5 . 2% and 5 . 9% of pupils in Côte d’Ivoire and Uganda , respectively , tested positive for CCA but negative by Kato-Katz . Tables of CCA results split by Kato-Katz infection category are available in S7 Supporting Information . Estimates of parameters obtained from the Bayesian LCA ( sensitivity and specificity of each test and prevalence ) with the associated BCI are available in Table 2 , and Fig 2 shows the posterior distributions of sensitivity and specificity of each test . Table 2 also shows parameters calculated from the posterior distributions of the estimated parameters . Sensitivity of Kato-Katz in Côte d’Ivoire ( 59 . 9% ) was significantly higher than in Uganda ( 32 . 3%; Table 1 , Fig 2 ) , although , there was no evidence of sensitivity estimates of CCAtn and CCAtp differing between the countries ( CCAtn = 49 . 3% and 50 . 1%; CCAtp = 63 . 0% and 69 . 7% in Côte d’Ivoire and Uganda respectively ) . In both countries , the sensitivity of CCAtp was significantly higher than the sensitivity of CCAtn . However , the countries differed with respect to the patterns of differences between CCA and Kato-Katz . In Côte d’Ivoire , the sensitivity of CCAtn was significantly lower than the sensitivity of Kato-Katz , but the sensitivity of CCAtp was not significantly different from the sensitivity of Kato-Katz . In contrast , in Uganda , the sensitivities of CCAtn and CCAtp were both higher than the sensitivity of Kato-Katz . The estimated specificity of Kato-Katz did not differ significantly across the two countries ( 99 . 0% in Côte d’Ivoire vs . 99 . 3% in Uganda; Table 1 , Fig 2 ) . There was no evidence that the specificity of CCAtn differed between the countries ( Côte d’Ivoire = 98 . 0% , Uganda = 98 . 8% ) , but the estimated specificity of CCAtp in Côte d’Ivoire ( 91 . 0% ) was marginally , non-significantly , higher than in Uganda ( 87 . 6%; Table 2 , Fig 2 ) . In both countries , the estimated specificity of CCAtp was significantly less than both the estimated specificity of CCAtn and the estimated specificity of Kato-Katz . In Côte d’Ivoire , the estimated specificity of CCAtn was marginally , non-significantly , less than the estimated specificity of Kato-Katz and there was no evidence in Uganda that the estimated specificity of CCAtn and Kato-Katz differed . Infection prevalence was estimated to be 20 . 5% in Côte d’Ivoire ( Table 1; Fig 3 ) and 19 . 4% in Uganda . Indeed , there was no evidence that the estimates of infection prevalence differed between the countries ( Table 2; Fig 3 ) . Both Kato-Katz and CCAtn substantially underestimated infection prevalence ( Table 2; Fig 3 ) . In contrast , there was no significant difference between test and infection prevalence by CCAtp in either country , with the CCAtp prevalence falling within 0 . 4 percentage points of the infection prevalence estimate in Côte d’Ivoire , and within 3 . 1 percentage points of the infection prevalence estimate in Uganda .
The aim of this study was to understand the implications considering CCA trace as negative or positive using data from two countries , Côte d’Ivoire and Uganda . We particularly focused on assessing the performance of CCA and Kato-Katz at measuring infection prevalence , as this is the main purpose of S . mansoni diagnostics in the control programs . We found that the sensitivities and specificities of CCA were much more consistent between countries than Kato-Katz and that estimates of prevalence by CCA with trace as positive was not significantly different from infection prevalence in either country . We discuss below possible reasons and implications for these results . We found that trace values treated as positive ( CCAtp ) had significantly and substantially higher sensitivity and lower specificity than when treated as negative ( CCAtn ) in both countries . However , the sensitivity estimates indicated that CCAtp still did not detect a substantial proportion of infections ( at least 30% in both countries ) , and this was supported by the raw data where 35% and 24% of the children that were Kato-Katz positive in Côte d’Ivoire and Uganda , respectively were negative by CCA . Although some of this may be due to misidentifying of samples , it seems extraordinary that this could explain the entire pattern . We used LCA to analyse the data , as there is no ‘gold standard’ test for S . mansoni infection . We put a strong prior distribution ( 95% certain greater than 80% , with mode at 95% ) on the specificity of Kato-Katz which may partly explain why estimated specificity of Kato-Katz ( 99% ) was high and did not differ between countries . However , the lower Bayesian 95% confidence limit of Kato-Katz specificity was greater than the mode of the prior distribution suggesting that the data were not indicating lower specificity of the Kato-Katz than the prior . Consequently , the use of Bayesian analysis enabled us to assess the appropriateness of our assumptions , which would not be possible in a frequentist framework . We elected to analyse trace positive and trace negative results within a single model . This method of analysis can be applied to any diagnostic where the result is not binary , where the results are in some way graded , and where alternative cut-off points can be assumed positive . The analysis is simple to implement and extend , with the key being that increasing the number of people testing positive on a sliding scale increases the sensitivity but decreases the specificity of the test . The preferred sensitivity and specificity balance of a diagnostic will depend on a number of different factors such as the properties of other diagnostics in use , the expected prevalence in the test population , and cost considerations . An alternative way to approach the analysis would be to analyse CCA trace negative and trace positive results separately . However , this risks the model returning logically impossible sensitivity estimates lower for trace positive than trace negative , or specificity estimates higher for trace positive than trace negative . Fitting trace positive and negative within a single model prevents this and also avoids having to interpret multiple estimates for sensitivities and specificities of other tests , and also for infection prevalence . We analysed the data in a Bayesian framework , where the use of prior distributions can help overcome issues around degrees of freedom while still letting the data indicate if the assumptions made are not valid , as opposed to the absolute assumptions that can be required in a frequentist framework . Bayesian analysis also outputs the full distribution of each parameter through the iterations saved by the model . Consequently , distributions of additional variables calculated from the parameters are simple to obtain; we used this technique both to test for significance of differences between terms and between studies and to compare test prevalence and infection prevalence for each test . Comparing between studies , we found that the estimated sensitivity and specificity of both CCAtp and CCAtn did not differ between the countries , in line with previous results that found the performance of CCA to be consistent before and after treatment [13] . However , the estimated sensitivity of Kato-Katz was much higher in Côte d’Ivoire , where the mean infection intensity was also much higher , than in Uganda . Lower sensitivity of Kato-Katz at low infection intensities is a well-known issue [10] and it is possible that this is the reason for our findings . A number of previous papers have assumed Kato-Katz to be the gold standard , including in a recent Cochrane review [27] . It is clear from these , and many other results , that it is not appropriate for Kato-Katz to be considered a gold standard , particularly at low intensities . We echo researchers in this [28] and other [29] fields in concluding that LCA is the most appropriate tool for assessing test sensitivity and specificity in the absence of a gold standard . Our work adds to a growing body of literature using LCA to assess the performance of CCA [20 , 30–35] . Our estimates of sensitivity for both CCA and Kato-Katz were on the lower end of those observed in these other studies , and our specificity estimates for CCA were relatively high . These studies together seem to be reflecting the general pattern of Kato-Katz sensitivity being strongly associated with infection intensities , with CCA perhaps being more consistent between environments [36] . However , this study is the first , to our knowledge , to assess the performance of CCA with respect to its main use in control programs of estimating prevalence within the study population . Comparison of test and estimated infection prevalence suggested that the prevalence measured by CCAtp was not significantly different from infection prevalence in either country and that both Kato-Katz and CCAtn significantly underestimated infection prevalence in both countries . Consequently , our results imply that CCAtp is revealing substantial numbers of children infected with S . mansoni that were not detectable by Kato-Katz [37] . Although it could be argued that the infected children that are being missed by Kato-Katz are those with the lowest infection intensities , even low levels of schistosomiasis have an associated morbidity burden [38] . Interestingly , estimated infection prevalence did not differ between Côte d’Ivoire and Uganda , despite Kato-Katz prevalence being more than twice as high in Côte d’Ivoire than in Uganda . The difference in Kato-Katz prevalence was presumably due to repeated rounds of treatment in Uganda leading to lower average infection intensities than in Côte d’Ivoire , which are better detected by CCA than Kato-Katz . However , in the absence of historical Kato-Katz and CCA data from the schools in Uganda , it is not possible to assess whether repeated rounds of treatment in Uganda has also been associated with a concurrent decrease in estimated prevalence . The main weakness of this study is the reliance on only two tests . Additional tests would be expected to increase the robustness of the study through increasing degrees of freedom , and there is clearly a need for additional studies incorporating more tests . However , we tried to mitigate for this weakness by using LCA , and also by incorporating covariances between the tests , which is expected to lead to more robust estimates than simply assuming the properties of different tests to be independent [39] . Additionally , the sample sizes in Côte d’Ivoire were over 4-fold higher than in Uganda , which is likely reflected in the larger confidence intervals around estimates from Uganda . As CCA becomes a more commonly used field tool , we expect sample sizes available for analyses to also increase . We demonstrated here how ambiguous trace results can be incorporated into LCA using S . mansoni data from Côte d’Ivoire and Uganda , enabling direct comparison of test properties when trace was considered both as negative and positive , and avoiding having to make assumptions as to the nature of trace results . Our results suggested CCA with trace as positive was most reflective of infection prevalence and that both Kato-Katz and CCA with trace as negative substantially underestimated infection prevalence . Consequently , we conclude that CCA is an appropriate tool for field testing for S . mansoni in control programmes .
|
Schistosomiasis is a debilitating disease affecting over 200 million people worldwide , mainly in developing countries . Treatment for schistosomiasis is straightforward and involves treating all school-age children , and sometimes also adults , in areas where schistosomiasis is known to occur . However , detecting intestinal schistosomiasis ( Schistosoma mansoni ) using Kato-Katz , the most common technique , can fail to detect infection when it is present as Kato-Katz relies on finding eggs within a small stool sample . A new field diagnostic , CCA , is promising as it is simple to use and seems to detect more infections that Kato-Katz . However , assessing the performance of CCA is difficult as we cannot be certain whether or not those positive by CCA but negative by Kato-Katz are truly infected . Additionally , CCA can often return a trace result between negative and positive which is difficult to interpret . We assess the performance of Kato-Katz and CCA in both Côte d’Ivoire and Uganda . We showed that CCA did indeed detect more infections than Kato-Katz , and that CCA accurately estimated the proportion of people truly infected in the population , when trace readings were considered positive . CCA is consequently an important and valuable tool in the fight against schistosomiasis .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"schistosoma",
"invertebrates",
"schistosoma",
"mansoni",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"education",
"helminths",
"sociology",
"geographical",
"locations",
"tropical",
"diseases",
"uganda",
"social",
"sciences",
"parasitic",
"diseases",
"animals",
"pediatrics",
"urine",
"neglected",
"tropical",
"diseases",
"pediatric",
"infections",
"africa",
"science",
"and",
"technology",
"workforce",
"schools",
"people",
"and",
"places",
"technicians",
"helminth",
"infections",
"schistosomiasis",
"professions",
"eukaryota",
"diagnostic",
"medicine",
"anatomy",
"science",
"policy",
"careers",
"in",
"research",
"physiology",
"biology",
"and",
"life",
"sciences",
"population",
"groupings",
"organisms"
] |
2017
|
Interpreting ambiguous ‘trace’ results in Schistosoma mansoni CCA Tests: Estimating sensitivity and specificity of ambiguous results with no gold standard
|
We investigated the roles played by the cysteine proteases cathepsin B and cathepsin L ( brucipain ) in the pathogenesis of Trypansoma brucei brucei in both an in vivo mouse model and an in vitro model of the blood–brain barrier . Doxycycline induction of RNAi targeting cathepsin B led to parasite clearance from the bloodstream and prevent a lethal infection in the mice . In contrast , all mice infected with T . brucei containing the uninduced Trypanosoma brucei cathepsin B ( TbCatB ) RNA construct died by day 13 . Induction of RNAi against brucipain did not cure mice from infection; however , 50% of these mice survived 60 days longer than uninduced controls . The ability of T . b . brucei to cross an in vitro model of the human blood–brain barrier was also reduced by brucipain RNAi induction . Taken together , the data suggest that while TbCatB is the more likely target for the development of new chemotherapy , a possible role for brucipain is in facilitating parasite entry into the brain .
Subspecies of Trypanosoma brucei are the causative agents of human African trypanosomiasis . In vitro studies utilizing both small molecule cysteine protease inhibitors and RNA interference ( RNAi ) have implicated the Clan CA ( papain ) family of cysteine proteases as critical to the successful lifecycle of Trypanosoma brucei brucei ( T . b . brucei ) [1] , [2] . In vivo studies have demonstrated that cysteine protease inhibitors prolong the lives of mice infected with lethal inocula of trypanosomes [1] , [3] . There are two distinct Clan CA cysteine proteases identified in the T . brucei genome . Brucipain ( aka trypanopain-Tb , rhodesain ) is a cathepsin L-like protease responsible for the bulk of protease activity in the organism [2] . Trypanosoma brucei cathepsin B ( TbCatB ) is a more recently characterized protease that is upregulated in the bloodstream stage of the parasite [2] . In in vitro studies , RNAi of TbCatB produced swelling of the endosome compartment analogous to that seen with class-specific cysteine protease inhibitors [1] , [2] and led to arrest of trypanosome replication and death . In contrast , knockdown of brucipain by RNAi in vitro produced no detectable phenotypic changes . However , it was hypothesized that this enzyme might play a role in the degradation of mistargeted glycosylphosphatidylinisotol ( GPI ) anchored proteins , VSG turnover , disruption of the blood–brain barrier , or degradation of host immunoglobulin [4] , [5] While RNAi with cultured parasites can provide important insights into the role of a specific gene product in parasite replication and viability , a role in pathogenesis , as proposed for brucipain , can only be validated in vivo . We show that introduction of RNAi from a tetracycline-inducible promoter can be achieved in vivo in a mouse model of T . b . brucei infection , and show that transcriptional silencing of either of these two proteases alters the course of T . b . brucei infection [6] .
Bloodstream T . brucei strain 90-13 was electroporated with plasmids containing either brucipain ( TbRho ) , TbCatB , or GFP transgenes [2] . The plasmid used , pZJM , allows transfected organisms to be induced to produce RNAi in the presence of tetracycline . The brucipain RNAi construct used for this study is one of three partial open reading frames ( ORF ) of brucipain used to down regulate its message in vitro . R1 encodes a cDNA that corresponding to the first 597 nucleotides of brucipain ORF . R2 encodes a cDNA encoding the middle 400 nucleotides of the brucipain ORF and R3 encodes a partial cDNA encoding the last 300 nucleotides of the brucipain ORF . Each of these constructs were capable of efficiently and specifically silencing the mRNA of brucipain in vitro . The same T . b . brucei clones expressing the R1 construct used in a previous study [2] . The TbCatB transgene has been described in detail previously [2] . To generate the GFP transgene , the gene encoding GFP ( 714 nucleotides ) was amplified from the pHD-HX-GFP vector [7] . Methods for electroporation and selection of stable transformants have been described [2] . Bloodstream form ( BSF ) 90-13 cells expressing T7 RNA polymerase and tetracycline repressor protein were maintained in HMI-9 medium [8] . Five BALB/c mice per group ( 6–8 weeks old ) were infected by intraperitoneal injection with 600 parasites carrying pZJMTbRho , pZJMTbCatB , or pZJMGFP plasmids or with control 90-13 parasites . To rule out any direct effects of doxycycline on the course of trypanosome infection , two additional groups of mice were infected with the parental T . b . brucei strain 90-13 . One group was given doxycycline-containing food ( 200-mg/Kg , Bioserv Corporation , San Diego , CA ) and water containing 1 mg/ml doxycycline hyclate ( Sigma-Aldrich ) , the second group was given standard food and water . Six other groups of mice were infected with T . b . brucei containing an RNAi-producing plasmid for brucipain ( pZJMTbRho ) , cathepsin B ( pZJMTbCatB ) , or GFP ( pZJMGFP ) . Three control ( uninduced ) groups were given standard food and water , and another three groups were given doxycycline containing food and water . The two groups infected with pZJMGFP served as a control for a gene that is not found in the trypanosome . Mice were monitored every other day for weight loss , general appearance , and behavior . Experiments were carried out in accordance with protocols approved by the Institutional Animal Care and Use Committee ( IACUC ) at UCSF . We used a human brain microvascular endothelial cell ( BMEC ) line whose phenotypic expression was stabilized by immortalization with pSVT , a pBR322-based plasmid containing the DNA sequence encoding the simian virus 40 large-T antigen [9] . Similar to the primary human BMEC cell line ( XIII ) from which they were derived , the transfected human BMECs are positive for FVIII-Rag , carbonic anhydrase IV , and Ulex europeus agglutinin I; take up acetylated low-density lipoprotein; and express gamma glutamyl transpeptidase [9] , [10] . Human BMECs were cultured at 37°C in medium 199 ( GIBCO ) supplemented with 20% heat-inactivated fetal bovine serum and 1× Glutamax ( GIBCO ) in a humidified environment of 95% air , 5% CO2 . The cells were grown to confluence on 6 . 5-mm-diameter collagen-coated Costar Transwell inserts with a pore size of 3 . 0 m until transendothelial electrical resistance ( TEER ) measurements exceeded 25 cm2 [11] . For the transmigration study , the parasites were added to the top of the human BMEC-containing inserts . The cultures were incubated with and without tetracycline ( 100 ng/ml ) in triplicate at 37°C , and the number of parasites present at the bottom chamber were determined by counting aliquots in the Neubauer chamber . Gene transcripts for brucipain were quantified in freshly isolated T . b . brucei from mice infected with pZJMTbRho at five days post infection . Blood was separated in a DEAE-sepharose column as previously described [12] . Total RNA extraction from T . b . brucei was performed using the TRIzol reagent ( Invitrogen , Carlsbad , CA ) . RT-PCR , the one-step RT-PCR kit ( Invitrogen , Carlsbad , CA ) , and gene-specific primers forward 5′-ATACGCAACGTTTGGTGTGA-3′ and reverse 5′CCTTCGATGTTGCCGATAGT -3′ were used to amplify brucipain . The relative amount of gene transcripts was calculated using methods previously described [13] . Parasites were purified from mice infected with parental 90-13 or pZJMTbRho . As reported previously [12] , T . b . brucei from infected mice were harvested by centrifugation , washed once in PBS-containing 1% glucose , and resuspended in lysis buffer ( 1 . 0% Triton X-100 , 10 mM Tris pH 7 . 5 , 25 mM KCl , 150 mM NaCl , 1 mM MgCl2 , 0 . 2 mM EDTA , 1 mM dithiothreitol , 20% glycerol ) . The lysates were incubated on ice for 20 min and cleared by centrifugation at 16 , 000 g for 15 min at +4°C . Protein concentration of was determined by the Bradford assay ( Bio-Rad ) . Ten µg of trypanosome lysate was resolved by 15% SDS-PAGE and transferred a to polyvinylidene difluoride ( PVDF ) membrane . After transferring and blocking , the PVDF membranes were incubated with rabbit anti-brucipain antiserum ( 1∶2500 dilution ) or anti-TbcatB 1∶2000 [14] for 1 h and washed three times for five min with TBST ( 10 mM Tris , pH 7 . 4 , 150 mM NaCl , 0 . 4% Tween 20 ) . After the third wash , horseradish peroxidase-conjugated donkey anti-rabbit IgG ( 1∶1 , 000 dilution ) was added to the blots for 1 h . The blots were washed again in the same buffer three times for five min and examined by ECL ( Amersham Biosciences ) . Equal amounts of trypanosome lysate ( 10 µg ) were labeled with 125I-DCG-04 [15] in the presence of 2 mM dithiothreitol for 45 min at room temperature and subjected to SDS-PAGE . Quantification of labeled enzymes was determined by Phosphoimager analysis ( Molecular Dynamics ) . Data were analyzed using the Mann-Whitney nonparametric test to determine the statistical difference in spleen weight in induced versus un-induced infected mice . Chi-square analysis was performed to determine the significant difference in survival .
The goal of these experiments was to validate the in vitro effects of RNAi on TbcatB in an in vivo disease model of African trypanosomiasis , and to explore a potential role of brucipain as a virulence factor . For safety reasons we conducted the knockdown experiment in the human non-infective strain T . b . brucei which has been traditionally grown and studied in mice . Doxycycline by itself produced no significant alteration ( +/−1 day ) in the course of T . b . brucei 90-13 infections ( Fig . 1A ) . Equivalent levels of parasitemia and splenomegaly were observed in mice whether or not they were maintained on a doxycycline-containing diet ( not shown ) . The in vivo induction of RNAi against brucipain in T . b . brucei did not cure infection , but extended the survival of three out of five mice beyond 60 days ( Fig . 1B ) the experiment was repeated twice with the same result . All mice infected with trypanosomes having the brucipain transcript knockdown had parasitemia and splenomegaly equivalent to that seen in control mice at the time of their sacrifice ( not shown ) . Splenomegaly ( quantified by spleen weight ) is a convenient gross pathological marker of disease burden [16] . Analysis of mRNA levels in trypanosomes isolated from infected mice confirmed 60% reduction in the level of brucipain mRNA ( Fig . 2A ) . The level of cathepsin B mRNA was not affected by RNAi induction against brucipain in pZJMTbRho induced parasites ( Fig . 2B ) . Active site labeling of brucipain in trypanosomes purified from mouse blood confirmed 60% reduction in brucipain protease activity ( Fig . 3C ) . Endogenous activity levels of brucipain and cathepsin B , quantified by DCG-04 labeling of purified parasites from mice infected with 90-13 strain , confirmed that brucipain was more abundant than cathepsin B ( Fig . 3D ) , consistent with previously published data [2] , [14] . A control cell line with an insert of GFP was generated to investigate the role of RNAi plasmid construct itself on the parasites in vivo . No difference was seen in mouse pathology or in brucipain or cathepsin B levels with GFP-induced parasites ( data not shown ) . In vivo induction of TbCatB RNAi resulted in survival of all five mice for up to two months post infection , after which time the experiment was terminated ( Fig . 1C ) . Un-induced mice began to die 13 days after infection . No trypanosomes were detected in the blood of mice infected with pZJMTbCatB trypanosomes after induction of RNAi with doxycycline . These mice also had normal spleen weights compared to un-induced controls ( Fig . 1D ) . Control mice with no doxycycline died between day 11 and 15 post infection . The last day on which untreated mice died from the trypanosome infection may vary depending on the exact parasite inoculum received and other host defense and host metabolic factors ( Fig . 1A vs 1B ) . The demonstration that doxycycline induction of RNAi can be achieved in parasites within an animal model of infection is an important technological advance that should encourage the use of this approach by other investigators . The failure of parasites to establish infection with TbCatB RNAi might have been predicted from in vitro assays . However demonstration in an in vivo model of infection is a significant and necessary validation of the key role of TbCatB in infection . The effect of reducing transcripts for the cathepsin L-like trypanosome protease ( brucipain ) on the progression of the infection was not predicted from in vitro assays . The effect of brucipain RNAi suggests that the cathepsin L protease might play a role as a virulence factor in in vivo infections , where host tissue tropism and the host immune response add new layers of complexity . In conclusion , gene-specific RNAi can be induced in bloodstream parasites in an experimental model of trypanosome infection . Induction of RNAi targeting TbCatB transcripts in vivo correlates with the results observed in previous in vitro RNAi experiments [1] , [2] . In the mouse model of infection , RNAi of TbCatB rescued mice from a lethal T . b . brucei infection , resulting in no splenomegaly and no detectable parasites in blood . While induction of RNAi against brucipain in two independent experiments did not cure mice of their infection , it did significantly prolong the survival of five out of ten mice . Since RNAi led to a 60% reduction of brucipain activity ( Fig . 3C ) , it is still possible that a 100% knockdown might uncover a more direct role for brucipain in parasite viability; brucipain knockouts are being pursued as strategy to more clearly delineate the role of brucipain . However , even the modest RNAi knockdown achieved for TbCatB ( quantified in [2] ) had a profound negative effect on parasite viability both in vitro [5] and in vivo , suggesting that T . brucei cathepsin B is the more likely target for protease inhibitors as chemotherapy against human African trypanosomiasis [17] . While the residual brucipain activity seen after RNAi induction might be responsible for disease progression in two of the mice shown in ( Fig . 1B ) , an alternative conclusion is that brucipain plays a specific role in Trypanosoma pathogenesis in vivo , but not in parasite viability per se . Nikolskaia et al . [5] showed that a cysteine protease inhibitor , known to target brucipain , blocked the ability of African trypanosomes to cross a model of the blood–brain barrier ( BBB ) [5] . Using this in vitro model of the blood–brain barrier , we confirmed that brucipain is required for African trypanosomes to effectively cross the brain endothelial barriers . Without tetracycline 3 . 54E+04±1 . 41E+03 ( mean±SEM ) of the initial brucipain RNAi trypanosome ( pZJMTbRho–tet ) inoculum crossed the endothelial cell barrier ( 1–2% ) ( Fig . 4 ) . This is comparable to those noted for T . b . brucei 427 and TREU 927 in previously published reports [5] , [11] . However when brucpain RNAi was induced by tetracycline , the number of parasites migrating across the barrier was reduced by 50% ( 1 . 10E+03±6 . 35E+02 ) , ( p = 0 . 003 ) . The human BMEC transendothelial electrical resistance ( TEER ) at the end of the experiment was 30 . 4±1 . 2 ohms ( p = 0 . 00002 ) , indicating that barrier integrity was maintained for all T . b . brucei treatment conditions . To rule out any effect of tetracycline on the in vitro BBB model other than induces RNAi , trypanosomes ( pZJMTbRho ) were pretreated with tetracycline , but the antibiotic was then removed and the parasites incubated with human BMEC overnight . The number of parasites crossing the BMEC was the same as control ( with tetracycline ) , demonstrating that tetracycline has no effect on endothelial cells ( data not shown ) . Experiments were repeated twice with the same result . In summary , the data show that knockdown of brucipain transcripts by RNAi led to reduced protease activity but no effect on parasitemia or splenomegaly . However the prolonged survival of some of the infected mice might be due to inability of the parasite to efficiently enter the CNS .
|
African trypanosomiasis , or sleeping sickness , is caused by the single-cell parasite Trypanosoma brucei ( T . brucei ) . Two parasite-derived enzyme proteins have been hypothesized to play an important role in the viability of the parasite or its ability to produce disease in the human host . Utilizing RNA interference that blocks the production of these proteins in the parasite , we show that elimination of parasite cathepsin B cures infection in mice . RNAi of the second enzyme protein , brucipain , results in the prolongation of life of half the infected mice , but does not cure . Further experiments carried out in a culture system show that brucipain facilitates the migration of parasites across a model of the blood–brain barrier . This suggests that while brucipain is not necessary for the viability of the organisms , it may play a role in infection by allowing parasites to reach the central nervous system and produce the severe second stage of sleeping sickness .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results/Discussion"
] |
[
"infectious",
"diseases/neglected",
"tropical",
"diseases"
] |
2008
|
RNA Interference of Trypanosoma brucei Cathepsin B and L Affects Disease Progression in a Mouse Model
|
In human leishmaniasis Th1/Th2 dichotomy similar to murine model is not clearly defined and surrogate marker ( s ) of protection is not yet known . In this study , Th1/Th2 cytokines ( IL-5 , IL-10 , IL-13 and IFN-γ ) profile induced by purified CD4+/CD8+ T cells in response to Leishmania antigens were assessed at transcript and protein levels in 14 volunteers with a history of self-healing cutaneous leishmaniasis ( HCL ) and compared with 18 healthy control volunteers . CD4+/CD8+/CD14+ cells were purified from peripheral blood using magnetic beads; CD4+/CD8+ T cells were co-cultured with autologous CD14+ monocytes in the presence of soluble Leishmania antigens ( SLA ) . Stimulation of either CD4+ T cells or CD8+ T cells of HCL volunteers with SLA induced a significantly ( P<0 . 05 ) higher IFN-γ production compared with the cells of controls . Upregulation of IFN-γ gene expression in CD4+ cells ( P<0 . 001 ) and CD8+ cells ( P = 0 . 006 ) of HCL volunteers was significantly more than that of controls . Significantly ( P<0 . 05 ) higher fold-expression of IFN-γ gene was seen in CD4+ cells than in CD8+ cells . In HCL volunteers a significantly ( P = 0 . 014 ) higher number of CD4+ cells were positive for intracellular IFN-γ production than CD8+ cells . Collectively , the volunteers have shown maintenance of specific long-term immune responses characterized by a strong reaction to leishmanin skin test and IFN-γ production . The dominant IFN-γ response was the result of expansion of both CD4+ and CD8+ T cells . The results suggested that immune response in protected individuals with a history of zoonotic cutaneous leishmaniasis ( ZCL ) due to L . major is mediated not only through the expansion of antigen-specific IFN-γ producing CD4+ Th1 cells , but also through IFN-γ producing CD8+ T cells .
Leishmaniasis is expanding both by increasing the incidence rate in endemic foci and extending the disease to new regions [1] , [2] . Control measures against leishmaniasis are not fully effective , chemotherapy is not always successful , and drug resistant is emerging [3]–[5] . Although theoretically development of an effective vaccine against leishmaniasis is feasible but yet there is no vaccine available against any form of leishmaniasis [6] , [7] . CD4+ T cells upon activation differentiate into functional effector Th1 and/or Th2 subsets and the outcome of Leishmania major infection in murine model is dependent upon the type of immune response generated: in most strains of mice L . major infection induces a Th1 type of response associated with a high level of IFN-γ , low level of IL-4 , and similar to human cutaneous leishmaniasis the lesion ( s ) heals spontaneously and the animals are protected against further infection; whereas L . major infection in BALB/c mice induces a Th2 response and a high level of IL-4 and low level of IFN-γ , as a result the disease is fetal [8] , [9] . The mechanism ( s ) of protection in human leishmaniasis is not well characterized; however , the role of T lymphocytes and Th1/Th2 cytokine profile are extensively studied [10]–[16] . In human leishmaniasis , peripheral blood mononuclear cells ( PBMC ) are routinely collected from patients with different clinical pictures of cutaneous leishmaniasis ( CL ) for immunological investigations . Results from the majority of these studies showed that PBMC of healing or cured cases of CL produce significant amount of IFN-γ in response to Leishmania antigens [17] , [18] . There is evidence demonstrating CD4+ T cells collected from patients with CL or mucocutaneous leishmaniasis ( ML ) or individuals with history of CL produced a high level of IFN-γ in response to Leishmania antigens which is an indication of a Th1 like response [10] , [11]; Conversely , T cells from patients with diffuse CL ( DCL ) failed to express IL-2 receptor and did not produce IFN-γ in response to Leishmania antigens , whereas IL-4 mRNA markedly increased in DCL lesions [17] , [18] . A clear Th1/Th2 dichotomy similar to murine model is not yet defined in human leishmaniasis [19] . There are reports which showed that CD8+ T cells play a role in controlling intracellular pathogens including protozoal and viral infections . CD8+ T cells are shown to confer a significant role in protection against acute and chronic form of Toxoplasma gondii infection [20] . In early stage of murine toxoplasmosis , CD8+ T cells hamper parasite dissemination by either direct lysis of infected cells or through release of cytokines . During chronic infection CD8+ T cells limit Toxoplasma cyst formation in tissues [21] , [22] . Immunity against malarial sporozoites is mediated partially by neutralizing antibodies , but largely depends on antigen specific CD8+ T cells , thus vaccines are designed based on induction of infection-blocking CD8+ T cells [23] , [24] . CD8+ T cells are also important in the control of HIV infection [25]–[27] . During HIV infection , CD8+ T cells recognize infected cells through an MHC-I dependent process and viral infected cells are lysed by secretion of perforin and granzymes [28] . Most patients chronically infected with HIV show CD8+ T cell response against HIV virus , but the response is not enough to successfully control viral replication [25]–[27] . In Listeria monocytogenes infection , both CD4+ and CD8+ T cells contribute in induction of protection , but the major bactericidal role is attributed to CD8+ T cells [29] . In experimental models of leishmaniasis , CD8+ T cells , in cooperation with CD4+ T cells , appear to be involved in the induction of host immunity against both primary infection and reinfection of Leishmania parasite [30]–[32] . In L . major infected CD8+ depleted BALB/c mice , during lesion healing the frequency of IFN-γ producing CD4+ T cells and the amount of IFN-γ are diminished resulted in a higher parasite burden [30] . In a study performed on C57BL/6 mice , infection with low dose of L . major induces a transient Th2 type response and then shifts to a Th1 response associated with healing . Induction of this Th1 type of response partly depends on the activation of IFN-γ producing CD8+ T cells and in the absence of CD8+ T cells , the Th2 response is sustained [33] . In mouse model of both genetically resistant and susceptible ( that were rendered resistant ) backgrounds , CD8+ T cells have been demonstrated to produce IFN-γ and contribute to the rapid healing of secondary lesions which develop after primary challenge with L . major [31] . There are reports from New World leishmaniasis which showed that CD8+ T cells are involved in healing process of CL due to L . braziliensis [34]–[37] . However , to our knowledge there is no data available about the possible role of CD8+ T cells and their cytokines in CL due to L . major . In leishmaniasis , most of the data generated so far is drawn from PBMCs culture without separation of T cell subtypes [10] , [12] , [13]which makes it difficult to judge the role of Th1/Th2 CD4+ cells and CD8+ T cells . In the current study two major lymphocyte subtypes , CD4+ and CD8+ T cells , were purified from individuals with history of self-healing CL and cytokine pattern were analyzed at transcript and protein levels in response to Leishmania antigens .
The study was approved by Ethical Committee of Tehran University of Medical Sciences . Potential candidates were invited and those who were willing to participate and sign a written informed consent were recruited . Fourteen volunteers with history of self-healing CL ( HCL ) caused by L . major and with leishmanin skin test ( LST ) more than zero and as control 18 healthy volunteers from non-endemic area with no response to LST were included . HCL volunteers were selected among the previous Center's patient who received no treatment for the CL lesion ( s ) and the lesion ( s ) cured spontaneously within one year of onset . The causative agent of every CL patient was previously identified as L . major using PCR method . Leishmania major ( MRHO/IR/75/ER ) was cultured on NNN medium and passaged on RPMI 1640 ( Gibco Invitrogen , Carlsbad , CA , USA ) supplemented with 10% fetal calf serum ( FCS ) . Promastigotes were harvested at day 5 , washed 3 times with PBS ( pH 7 . 2 ) and used for preparation of soluble Leishmania antigen ( SLA ) as previously described [14] . Briefly , protease inhibitor cocktail enzyme ( Sigma , St . Louis , MO , USA ) was added at 100 µl per 1×109 promastigotes , and then the parasites were freeze-thawed 10 times followed by sonication at 4°C with two 20-sec blasts . Parasite suspension was centrifuged at 30 , 000×g for 20 min , the supernatant was collected and re-centrifuged at 100 , 000×g for 4 hours . SLA protein concentration was measured using Bradford method [38] . Finally the supernatant was sterilized using 0 . 22 µm membrane filter , aliquoted and stored at −20°C until use . Twenty mL of blood sample was collected from each volunteer and Peripheral Blood Mononuclear Cells ( PBMCs ) were isolated using Ficoll–Hypaque ( Sigma , St . Louis , MO , USA ) density gradient centrifugation . CD4+ and CD8+ lymphocytes isolation was performed using magnetic beads system ( StemCell Technologies Inc . , Vancouver , BC , Canada ) by positive selection using anti-CD4 or anti-CD8 coated nanoparticles . Briefly , cell suspension was prepared at a concentration of 1×107 cells/ml in a 5 ml tube in isolation buffer containing PBS plus 2% ( v/v ) FBS and 1 mM EDTA . EasySep CD4/CD8 cocktail Abs was added at 10 µl/ml cells , mixed well and incubated at room temperature ( RT ) for 15 min . Magnetic nanoparticles were added at 5 µl/ml cells and incubated for 10 min at RT . The cell suspension was brought to 2 . 5 ml by adding buffer and the tube was placed into the magnet for 5 min , then the supernatant was discarded . The desired cells were remained bound inside the tube . The steps of placing tube into the magnet were repeated three times . Monocytes ( CD14+ ) were isolated from autologous PBMC by negative selection according to the manufacturer's instruction ( StemCell Technologies Inc . , Vancouver , BC , Canada ) . Briefly , cell suspension was prepared at a concentration of 5×106 cells/ml in isolation buffer . EasySep monocyte enrichment cocktail Abs was added at 5 µl/ml cells , mixed well and incubated at 4°C for 10 min . Magnetic microparticles were added at 5 µl/ml cells for 5 min at 4°C . The cell suspension was brought to 2 . 5 ml by adding buffer and the tube was placed into the magnet , for 2 . 5 min at RT . The desired unbound fraction was transferred into a new tube . The purity of the yielded lymphocytes or monocytes was more than 95% by flow cytometry analysis using specific conjugated mAb ( Fig . 1 ) . The contamination of CD8+ T cells with NK cells was less than 9% using α-CD56 mAb . Monocytes were co-cultured with sorted lymphocytes as antigen presenting cells ( APCs ) following mitomycin C ( Merk , Darmstadt , Germany ) treatment at a final concentration of 10 µg/ml for 30 min at 37°C with 5% CO2 . The cells were cultured in RPMI 1640 media supplemented with 10% heat-inactivated human AB Rh+ serum , 10 mM/L Hepes , 2 mM L-glutamine , 100 U/ml penicillin G and 100 µg/ml streptomycin ( Gibco Invitrogen , Carlsbad , CA , USA ) . CD4+ or CD8+ lymphocytes were adjusted to 0 . 5–1×106 cells/ml mixed with 1∶10 of autologous monocytes and were cultured in U-bottomed 96-well plates ( Nunc , Roskilde , Denmark ) in the presence of either 10 µg/ml PHA or 50 µg/ml of SLA in a final volume of 200 µl . Plates were incubated at 37°C with 5% CO2 in humidified atmosphere for 72 hrs . Culture supernatants were collected at 72 hours , the level of IL-5 , IL-10 , IL-13 and IFN-γ were titrated in culture supernatants using ELISA method ( Mabtech , Stockholm , Sweden ) . Briefly , the plates were coated with anti-IFN-γ/IL-5/IL-10/IL-13 mAb in PBS , pH 7 . 4 , and incubated at 4°C over night . After blocking the wells using buffer containing PBS plus 0 . 05% ( v/v ) Tween 20 and 0 . 1% ( w/v ) BSA , supernatants were added to each well . Biotin-labeled mAb in incubation buffer was added to each well and as enzyme streptavidin-HRP was used . The reaction was developed using 3 , 3′ , 5 , 5′-tetramethylbenzidine ( TMB ) substrate and stopped with 0 . 5M H2SO4 solution . The plates were washed after each step of incubation using PBS+0 . 05% ( v/v ) Tween20 . The plates were read at 450 nm using a reader ( BioTek , Winooski , VT , USA ) . The mean optical densities ( ODs ) of triplicate cultures were compared with the standard curves prepared using recombinant IL-5 , IL-10 , IL-13 and IFN-γ . The cytokine levels represent the differences between the ODs of test and background wells . The detection limit of the assays was 4 pg/ml for IL-5 and 0 . 5 pg/ml for IL-10 , 5 pg/ml for IL-13 and 2 pg/ml for IFN-γ . After SLA stimulation , part of the cells was used for ICS assay . Cells were adjusted at 5×105 per ml and stimulated with PMA ( Sigma , St . Louis , MO , USA ) 50 ng/ml plus Ionomycin calcium ( Sigma ) 500 ng/ml and incubated at 37°C , 5% CO2 for 5–6 hrs . Monensin ( Sigma ) was added at 25 µM/ml during the last 4–5 hrs of culture for blocking . Cells were harvested and washed 2 times with PBS ( pH 7 . 2 ) plus 0 . 1% bovine serum albumin ( BSA ) . The cells were permeabilized using BD Cytofix/Cytoperm kit according to the manufacturer's instruction ( BD Biosciences , San Jose , CA , USA ) . In the final step , cells were stained with FITC-conjugated mouse anti-human IFN-γ and PE-conjugated rat anti-human IL-2 ( BD Biosciences , San Jose , CA , USA ) . Cells were washed ×2 with perm/wash buffer and resuspended in PBS ( pH 7 . 2 ) plus 1% BSA . Cells were analyzed using Partec flow cytometer ( DAKO cytomation , Glostrup , Denmark ) while isotype matched negative controls were used to set the threshold of autofluorescence . A minimum of 50 , 000 events were acquired for each sample . FACS data analysis was performed using FloMax ( DAKO cytomation ) software . At the time of supernatants collection ( at day 3 ) , the SLA stimulated cells were harvested and used for RNA extraction . The procedure began with reverse transcription of mRNA to cDNA . The cDNA was then used as template for Real-time PCR using specific primer of each cytokine . Solutions were treated and glassware was filled with 0 . 1% ( v/v ) diethylpyrocarbonate ( DEPC ) ( Merck , Darmstadt , Germany ) in H2O . The cell pellet was resuspended in cold PBS ( pH 7 . 2 ) and lysed by addition of 0 . 2 mL of Trizol ( Sigma , St . Louis , MO , USA ) per 1×106 cells . RNA was solublized through pipetting and incubated at RT for 5 min . Then 0 . 2 ml chloroform was added per 1 ml of homogenate , the suspension was shook vigorously and kept on ice for 5 min followed by centrifugation for 15 min , 12 , 000×g at 4°C . The upper phase was collected and added to an equal volume of isopropanol and incubated at 4°C over night . Then the cell suspension was centrifuged at 12 , 000×g , 4°C for 15 min , the supernatant was discarded and the RNA pellet was washed with 1 ml 75% ethanol at 7 , 500×g for 8 min . At the end , the pellet was drayed briefly and dissolved in DEPC treated water . The purity of RNA samples was assessed by the ratio of ODs at 260/280 nm using UV spectrophotometry . Reverse transcription was carried out using RevertAid M-MuLV enzyme ( Fermentas life sciences , York , UK ) in a 30 µL reaction mixture . Briefly , 1 µL ( 0 . 5 µg ) of oligo dT18 primer was added to about 3 µg of total RNA , mixed and incubated at 70°C for 5–10 min . The tube was placed on ice for a few minutes , centrifuged briefly and added with: 4 µL of 5× reaction buffer , 1 µL 10 mM dNTPs , 20 u RNase inhibitor ( RiboLock; Fermentas life sciences ) and DEPC treated water up to 30 µL . Tube was incubated at 37°C for 5 min , the contents were mixed gently and then added with 100 U of enzyme . The reaction mixture was incubated at 42°C for 60 min . The enzyme was inactivated by heating at 70°C for 10 min . and chilled on ice . In a Real-time PCR MicroAmp optical 96-well reaction plate ( Applied Biosystems , Foster City , CA , USA ) for 25 µl reaction mixture the followings in each optical well were prepared: 12 . 5 µl QuantiTect SYBR Green I ( Qiagen , Hilden , Germany ) , 3 µl cDNA , 2 µl primer pair mix ( 0 . 5 µM each ) , 7 . 5 µl dH2O ( For sequences of primer pairs see Table 1 ) . For normalizing the difference in the amount of inputting cDNAs , the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) was used as the internal standard . Samples and internal standard were amplified in separate wells of the plate . Two-step thermal profile as a PCR program was set up on the SDS software ( version 1 . 3 . 1 ) of Applied Biosystems 7 , 500 machine ( Applied Biosystems , Foster City , CA , USA ) . The dissociation curve on the instrument software was set as follows:Software generates reports including amplification plots and dissociation curves . Any bimodal dissociation curve or abnormal amplification plots were checked to see if there is an indication of different Tms and nonspecific products . The Ct ( threshold cycle ) of each sample was used in gene relative expression calculation . The 2−ΔΔCt method was used to calculate relative changes in the gene expressions:While To have a valid calculation of 2−ΔΔCt , the amplification efficiencies of the target and reference ( GAPDH ) genes must be approximately equal . For this purpose , the Ct values variation with cDNA template dilutions was checked . A pooled cDNA preparation was diluted over a 10-fold range and PCR was performed for each dilution using specific primers . A plot of the log cDNA dilution versus CT was prepared . The slope of each line was obtained from regression equation and the efficiencies of the target and reference genes were calculated using the equation: Efficiency ( E ) = [10 ( 1/slop ) ]−1 Non-parametric tests of Mann-Whitney , Kruskal-Wallis and Dunn's post-test for paired comparisons were used for statistical analysis of the data using SPSS version 11 . 5 ( SPSS Inc . , Chicago , IL , USA ) and GraphPad Prism version 5 . 01 ( GraphPad Software Inc . , La Jolla , CA , USA ) softwares . Nonparametric tests were chosen because the samples did not follow a Gaussian distribution . P value of <0 . 05 considered to be significant .
Using ELISA method , cytokine profile ( IFN-γ , IL-10 , IL-13 ) were measured on supernatants collected at 72 hrs of SLA stimulated PBMC or CD4+ , CD8+ T cells culture . The amount of IFN-γ level was significantly higher in PBMC culture of HCL volunteers compared with that of healthy controls ( P<0 . 05 ) . Results of purified T cell culture showed that stimulated CD4+ T cells from HCL volunteers induced a significantly higher IFN-γ production compared with cells from healthy controls ( P<0 . 05 ) ( Fig . 2A ) . Similarly , stimulated CD8+ T cells from HCL volunteers induced a significantly higher IFN-γ production compared with the cells from healthy controls ( P<0 . 05 ) ( Fig . 2B ) . The levels of IL-10 and IL-13 were not significantly different in either CD4+ or CD8+ T cells between HCL volunteers and healthy controls ( Fig . 2A and B ) . IL-5 level was not detectable in culture supernatants of either CD4+ or CD8+ T cells . The relative quantities of the target genes were normalized against the relative quantities of the internal standard ( GAPDH ) . Ct values of amplified templates of antigenic stimulated T cells were used for calculation of different cytokine gene expressions using 2−ΔΔCT method . The expression amount was compared with unstimulated cells of culture and relative fold-expression was reported . Result for each donor was calculated and then data was pooled and presented as a mean of HCL volunteers against healthy controls . Results showed that the upregulation of IFN-γ gene expression in CD4+ cells from HCL volunteers was significantly higher than that of healthy controls ( P<0 . 001 ) ( Fig . 3A ) . Similarly , fold-expression changes of IFN-γ gene was significantly higher in CD8+ cells from HCL volunteers compared to the cells from healthy controls ( P = 0 . 006 ) ( Fig . 3B ) . Comparing CD4+ and CD8+ T cells , the significantly higher fold-expression of IFN-γ gene was seen in CD4+ cells than CD8+ cells of HCL volunteers . In both CD4+ and CD8+ T cell cultures , the changes in the gene expression of IL-5 , IL-10 and IL-13 were not significantly different between HCL volunteers and healthy controls . By amplification of serially diluted pooled cDNA , the amplification efficiency of the target ( cytokines ) compared to reference ( GAPDH ) genes was examined using SYBR Green detection . Using the equation pointed out in methods , efficiency of GAPDH was 96% while that of targets were between 91% and 95% . Seventy two hrs after SLA stimulation of CD4+/CD8+ T cells , part of the cells were harvested and stimulated with PMA plus Ionomycin for 5–6 hrs , stained for intracellular IFN-γ with conjugated mAbs and the frequency of positive cells was analyzed using flow cytometry . In CD8+ cells compartment , antibody to CD56 marker allowed to exclude IFN-γ positive populations of natural killer cell sources ( Fig . 4 A and B ) . Results of analysis of cells from HCL volunteers and healthy controls were pooled separately and presented as median number of intracellular IFN-γ positive CD4+ and CD8+ T cells ( Fig . 4C ) . Based on this analysis , HCL volunteers showed that a significantly higher number of CD4+ T cells were positive for intracellular IFN-γ production than CD8+ cells ( P = 0 . 014 ) .
Resistance and susceptibility to L . major infection in murine model depend upon induction of Th1 or Th2 response , respectively [8] , [9] , [39] . Recovery from CL usually is accompanied with long lasting protection and strong immune response generation indicated by in vivo LST and in vitro lymphocyte response to Leishmania antigens [10] , [12] , [13] , [19] , [40] , yet in human leishmaniasis the surrogate marker ( s ) of protection is not well defined . Most of the studies performed on human immune response against leishmaniasis is carried out on crude PBMCs without purifying CD4+ T cell and CD8+ population and there is no report to show a clear-cut CD4+ Th1/Th2 response [34] . In the current study cytokines patterns of CD4+ Th1/Th2 and CD8+ T cells in volunteers recovered from CL is evaluated at the transcript and protein levels . The results of soluble Leishmania antigens ( SLA ) stimulated cells showed that pure CD4+ T cells induced a significantly higher IFN-γ production in HCL volunteers compared to that of healthy controls , IL-5 as a Th2 type cytokine was not detectable and IL-10 and IL-13 levels were not significantly different in culture of T cells from HCL volunteers compared with that of healthy controls . At the same time , the level of cytokines' mRNA expression was evaluated to detect T cell cytokine responses to Leishmania stimulation at transcript level , after several experiments to explore the optimum stimulation time points for mRNA analysis . Simultaneous analysis of cytokines gene expression showed a strong up-regulation of Th1 cytokine IFN-γ mRNA in CD4+ T cells . In line with the results of secreted proteins , level of Th2 cytokines transcripts including IL-5 , IL-10 and IL-13 showed no significant increase in HCL volunteers compared with healthy control . The up-regulation of IFN-γ transcripts expression in Leishmania stimulated CD4+ T cells consistent with IFN-γ secretion is an indication of Th1 type of response in HCL volunteers parallel with no Th2 response indicating by low level of IL-5 , IL-10 and IL-13 cytokines . Analysis of cytokine secretion and transcript expression of SLA stimulated CD8+ T cells showed also a significantly higher IFN-γ production in HCL volunteers compared to the healthy control volunteers . Similar to CD4+ T cells , the levels of IL-10 and IL-13 were not significantly different in CD8+ T cell culture between HCL volunteers and healthy controls . When CD4+ and CD8+ T cell response was compared , the level of IFN-γ secretion in SLA stimulated cells was not significantly different between CD4+ and CD8+ cells , but real-time PCR analysis revealed that expression level of IFN-γ mRNA was higher in CD4+ T cells than CD8+ T cells . To confirm the real-time PCR results , part of CD4+/CD8+ sorted T cells were harvested following SLA stimulation and intracellular production of IFN-γ was assessed using flow cytometry . Results of intracellular cytokine staining ( ICS ) in HCL volunteers confirmed that a significantly higher number of CD4+ T cells produced intracellular IFN-γ compared with CD8+ T cells ( median = 15% vs . 8% ) . Based on the results it seems that the source of IFN-γ production is both CD4+ Th1 cells and CD8+ cells in individuals with history of CL . The role of CD8+ T cells in human Leishmania infection is not well known and existed reports are controversial . In a study performed on Sudanese individuals it was suggested that IFN-γ production is associated with CD4+ T cells rather than CD8+ T cells in individuals with history of CL due to L . major [40] . A report from New World leishmaniasis showed that in both asymptomatic and antimonial treated CL individuals caused by L . braziliensis , a higher proportions of CD4+ than CD8+ T cells was present [34] . In another report the authors showed that after treatment of CL due to L . braziliensis , the frequency of CD4+ and CD8+ T cells was the same with approximately constant production of IFN-γ [41] . On the other hand , some clinical studies reported high numbers of Leishmania specific CD8+ T cells in the lesions and peripheral blood during acute phase and healing process in L . major or L . braziliensis CL patients [42] , [43] . In mice , the requirement of CD8+ T cells for the control of L . major infection is shown to be partly dependent on the procedure of challenge: β2-microglobuine or CD8+ deficient C57BL/6 mice when challenged with high dose of L . major have the ability to cure the lesion , which indicates that CD8+ T cells are not necessary for the control of primary [44] infection , while in the intradermal challenge with low dose ( 100 metacyclic promastigotes into the ear dermis ) the outcome of primary L . major infection in anti-CD8 Ab treated or CD8 deficient mice was dependent on the CD8+ T cells [45] . The role of CD8+ T cells was studied in CBA and anti-CD4 mAb treated BALB/c mice healed from L . major infection . The cured mice were re-challenged with L . major in the contralateral footpad and lymph nodes cells were depleted of CD4+ T cells and stimulated in vitro . The remaining CD8+ T cells produced a significant amount of IFN-γ [31] . It is believed that in the resolution of the primary Leishmania infection and induction of protection in murine model CD8+ T cells play an important role [31] . In the present study , following the isolation of CD4+/CD8+ T cells , Th1/Th2 cytokines were titrated on culture supernatant of in vitro restimulated T cells to check the type of immune response elicited against Leishmania antigen . The main cytokine produced was found to be IFN-γ in the volunteers' T cells . IFN-γ eliminates intracellular pathogens primarily through macrophage activation . Macrophages upon activation produce nitric oxide ( NO ) which inhibits growth of intracellular pathogens . It is shown that during active lesion of CL due to L . major the proliferative response and IFN-γ production of PBMC was increased [12] , [46] and T cells from healed CL produced a significantly higher level of IFN-γ but a low level of IL-10 than the cells from controls [14] , [15] , [40] . Similarly , studies in L . braziliensis infection demonstrated a Th1/Th2 mixed response in early stage of active CL [11] , [42] and then a sustained Th1 response with elevated level of IFN-γ and down-regulation of IL-4 and IL-10 production were seen apparently associated with healing [11] . Likewise , the presence of high level of IFN-γ in the skin lesions of CL patients support the role of IFN-γ in healing process [46] . Using RT-PCR , the cytokine patterns of skin lesions of CL patients showed predominance IFN-γ , and low levels of IL-5 and IL-10 [17] . The contribution of IFN-γ to the recall of immunological memory against L . major reinfection was assessed in mice . The neutralization of IFN-γ at the time of reinfection reduced the specific DTH response , showing the involvement of IFN-γ in the recall of memory response to L . major [31] . Similarly in intracellular infection with T . gondii , it is shown that CD8+ T cells confers resistance against acute infection [20] and IFN-γ producing CD8+ T cells play a significant role in controlling chronic T . gondii infection and inhibits encephalitis in mouse model [21] , [22] . In the current study , even though using real-time PCR the expression level of IFN-γ transcripts in CD8+ cells was less than CD4+ cells , but interestingly a significant amount of IFN-γ was produced by CD8+ T cells in cell culture and around 5–12% of CD8+ cells was positive for IFN-γ secretion by ICS assay . It is concluded that CD8+ T cells contribute along with CD4+ Th1 cells in IFN-γ production in individuals with history of CL . Despite the limited reports of CD4+ Th1 cells as the main source of IFN-γ production in CL patients [47] , [48] in most studies of CD4+ Th1/Th2 paradigm in human CL , PBMCs rather than purified T cells were used , hence the role of IFN-γ producing CD8+ T cells should not be ruled out when reporting a “Th1” type response in PBMC culture . The strong lymphoproliferative and IFN-γ response in self healing CL caused by L . braziliensis is previously shown [49] , [50] . In the current study , HCL volunteers with spontaneous healing during 1 . 5–5 months were recruited . Individuals with history of self healing CL are presumed to be protected against further Leishmania infection . The blood samples were collected a few months to years after cure of CL lesions . The strong LST response ( mean LST = 10 . 7±7 . 5 mm ) and IFN-γ production is an indication of sustaining cell mediated immune response . This sustaining protective immunity is mediated not only through the expansion of antigen-specific IFN-γ producing CD4+ Th1 cells , but also through IFN-γ producing CD8+ T cells . The question that which one of these T cell subsets plays a more important role in IFN-γ production at the initiation of exposure to sand fly bite needs to be explored .
|
Cutaneous leishmaniasis ( CL ) is usually a self-healing skin lesion caused by different species of Leishmania parasite . Resistance and susceptibility of mice to Leishmania major infection is associated with two types of CD4+ T lymphocytes development: Th1 type response with production of cytokine IFN-γ is associated with resistance , whereas Th2 type response with production of cytokines IL-4 and IL-5 is associated with susceptibility . A clear Th1/Th2 dichotomy similar to murine model is not defined in human leishmaniasis and we need as much information as possible to define marker ( s ) of protection . We purified CD4+/CD8+ T cells , stimulated them with Leishmania antigens and analysed gene and protein expression of Th1/Th2 cytokines in volunteers with a history of self-healing CL who are presumed to be protected against further Leishmania infection . We have seen significant upregulation of IFN-γ gene expression and high IFN-γ production in the Leishmania stimulated CD4+ T cells and CD8+ T cells . We concluded that both antigen-specific IFN-γ producing CD4+ Th1 cells and IFN-γ producing CD8+ T cells contribute to the long term protection in individuals with a history of CL . This proves the importance of CD8+ T cells as a source of IFN-γ in Th1-like immune responses .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"immunology/immune",
"response"
] |
2010
|
CD8+ T Cells as a Source of IFN-γ Production in Human Cutaneous Leishmaniasis
|
Embryonic stem ( ES ) cells are pluripotent cells derived from the inner cell mass of the mammalian blastocyst . Cellular differentiation entails loss of pluripotency and gain of lineage-specific characteristics . However , the molecular controls that govern the differentiation process remain poorly understood . We have characterized small RNA expression profiles in differentiating ES cells as a model for early mammalian development . High-throughput 454 pyro-sequencing was performed on 19–30 nt RNAs isolated from undifferentiated male and female ES cells , as well as day 2 and 5 differentiating derivatives . A discrete subset of microRNAs ( miRNAs ) largely dominated the small RNA repertoire , and the dynamics of their accumulation could be readily used to discriminate pluripotency from early differentiation events . Unsupervised partitioning around meloids ( PAM ) analysis revealed that differentiating ES cell miRNAs can be divided into three expression clusters with highly contrasted accumulation patterns . PAM analysis afforded an unprecedented level of definition in the temporal fluctuations of individual members of several miRNA genomic clusters . Notably , this unravelled highly complex post-transcriptional regulations of the key pluripotency miR-290 locus , and helped identify miR-293 as a clear outlier within this cluster . Accordingly , the miR-293 seed sequence and its predicted cellular targets differed drastically from those of the other abundant cluster members , suggesting that previous conclusions drawn from whole miR-290 over-expression need to be reconsidered . Our analysis in ES cells also uncovered a striking male-specific enrichment of the miR-302 family , which share the same seed sequence with most miR-290 family members . Accordingly , a miR-302 representative was strongly enriched in embryonic germ cells derived from primordial germ cells of male but not female mouse embryos . Identifying the chromatin remodelling and E2F-dependent transcription repressors Ari4a and Arid4b as additional targets of miR-302 and miR-290 supports and possibly expands a model integrating possible overlapping functions of the two miRNA families in mouse cell totipotency during early development . This study demonstrates that small RNA sampling throughout early ES cell differentiation enables the definition of statistically significant expression patterns for most cellular miRNAs . We have further shown that the transience of some of these miRNA patterns provides highly discriminative markers of particular ES cell states during their differentiation , an approach that might be broadly applicable to the study of early mammalian development .
ES cells are pluripotent cells derived from the inner cell mass of the mammalian blastocyst . Depending on culture conditions , these cells can differentiate into various cell types [1] . Cellular differentiation entails loss of pluripotency and gain of lineage-specific characteristics . However , the molecular controls that govern the differentiation process are poorly understood . During differentiation , lineage-specific transcription factors activate the expression of specific sets of genes to form hierarchical transcription networks [2] , while repressors and epigenetic modifications restrict pluripotency and help to define developmental potential [3] . Nevertheless , the precise molecular pathways involved remain unclear . Over the past two decades , several important studies have implicated regulatory non-coding RNAs in the control of gene expression during development [4] , [5] . In particular , a large body of work in several organisms has demonstrated that transcriptional regulation is controlled not only by protein factors , but also by small endogenous RNA molecules of ∼19–23 nucleotides ( nt ) in length called microRNAs [6] . miRNAs serve as regulators of gene expression by partially binding to complementary sites on their target transcripts . Modes of miRNA action include endonucleolytic cleavage of target mRNA , accelerated mRNA decay or repression of translation [7]–[9] . In animals , several hundred miRNAs have been identified , that regulate diverse biological processes ranging from cell metabolism to cell differentiation and growth , apoptosis , cancer and immune responses [10] , [11] . Moreover , it has been shown that many miRNAs are characterized by highly specific spatial and temporal expression patterns supporting their role in such processes [12] . The biogenesis of miRNAs involves nuclear processing of a long primary transcript ( pri-miRNA ) into a stem-loop structured pre-miRNA by the RNase III Drosha . The pre-miRNA is then exported to the cytoplasm and further matured by the RNase III Dicer into a siRNA-like duplex . The single-stranded mature miRNA is then asymmetrically transferred into an Argonaute-containing miRNA effector complex , while the passenger strand , or microRNA* ( miR* ) is degraded [13]–[16] . Several recent reports underscore important roles for miRNAs in preventing differentiation of ES cells , most notably through the activity of the pluripotency miR-290 cluster [17]–[20] . However , the dynamics of small RNA accumulation during early ES cell differentiation , particularly at early stages , has not been investigated so far . Additionally it remains unclear whether expression of some small RNAs can be regulated in a sex-specific manner and could thereby contribute to poorly understood processes such as X chromosome inactivation in females ( for a review , see [21] ) . To address these issues , we have analysed mouse ES cell small RNA populations through the generation and sequencing of small RNA libraries isolated from both male and females cell lines at day 0 , 2 and 5 of differentiation . We describe here the results of this analysis , focusing on the most abundant class of isolated small RNAs , the miRNAs .
In order to examine the small RNA profiles of mouse ES cells during early differentiation , we generated six libraries using RNA isolated from male ( E14 , XY1 ) and female ( PGK , XX1 ) ES cells , either in the undifferentiated state ( D0 ) , or after 2 or 5 days of differentiation ( D2 , D5 , respectively ) . These ES cell lines are cultured under feeder-free conditions in the undifferentiated state , thus avoiding any contamination of fibroblast cell-derived small RNAs . We induced differentiation by LIF withdrawal and cell dilution ( see Materials and Methods ) , which differs from most previous studies [17] , [22] , [23] , where differentiation had been induced by retinoic acid ( RA ) that preferentially promotes neuronal differentiation . By initiating a slower differentiation process than with RA treatment , we hoped to be able to monitor the diversity of early events during the acquisition of cell identity . We confirmed the early differentiation status of the cells throughout time by monitorig the expression of key differentiation and cell fate-specific markers ( Figure S1 ) . Libraries were prepared using the ∼19-to-30 nucleotides ( nt ) fraction of total RNA from ES cells and differentiating cells ( see Materials and Methods ) . This fraction was chosen , as it represents the size range of known small RNAs families in mammals [24] . To obtain a comprehensive picture of the corresponding small RNA profiles , 50 , 000-100 , 000 reads per library were produced using the 454 pyro-sequencing technology . For all six libraries , initial analysis of the cloned populations revealed that the vast majority of small RNAs present in these samples were 22–23 nt in length ( Figure S2A ) . Bioinformatics analyses employing available small RNA databases ( see Materials and Methods ) showed that most small RNAs corresponded to known miRNAs , while the remaining , less abundant classes corresponded to ribosomal RNA ( rRNAs ) , transfer RNA ( tRNAs ) , and other non-coding ( nc ) RNAs . An additional small RNA class mapped to repeated ( repeat ) and non-repeated ( genome ) genomic DNA , while a final fraction corresponded to unclassified species , with no obvious matches to the mouse genome ( not annotated ) ( Figure S2B ) . The overall distribution of small RNAs did not vary significantly between male and female ES cell differentiation . However , the cloned miRNA fraction significantly increased during differentiation ( 45% and 65% of cloned sequences at D0 and D5 , respectively; Figure S2B ) , suggesting a highly dynamic regulation of this population of small RNAs during early differentiation . Given the critical role of miRNAs in regulating differentiation events during development [25] , we decided to focus on this specific class of small RNAs . Libraries were queried against precursor sequences of all known microRNAs , as deposited in miRbase ( release 10 . 1 , December 2007 ) . The complete set of cloned miRNAs is presented in Dataset S1 ( Sheet 1 ) , alongside the cloning frequency relative to the total number of miRNAs in each library . The results were filtered according to a two-component Gaussian mixture model ( Figure S3 and Materials and Methods ) defining a lower threshold frequency of 0 . 05% , below which cloned miRNAs were considered as background and removed from the dataset , apart from a group of 17 novel miRNAs* not previously cloned or deposited on miRbase release 12 . 0 ( designated by ‘SL’ in Dataset S1 , Sheet 2 ) . The identification of these miRNA* , which are transient intermediates in the miRNA biogenesis pathway , demonstrated that our cloning/sequencing approach was of adequate depth . In order to assign quantitative miRNA measurements to different mouse chromosomes we used an adaptation of the VAMP software ( Visualisation and Analysis of Molecular Profiles; Figure S4 ) [26] . For the analysis of miRNA expression patterns during differentiation , we were able to treat male and female cell samples as expression replicates , as the Pearson and Spearman correlation coefficients between miRNA quantification and sequencing frequencies indicated that male and female cell libraries are highly comparable at each differentiation time point ( p-values<2 . 2e-16; Figure S5 and Materials and Methods ) . A lower ( albeit significant ) Pearson score at D5 is likely due to the fact that sex-enrich miRNA are fully expressed at this stage and create differences between male and female samples ( see later in the text ) . The lower Spearman test score obtained at D0 is likely explained by the fact that few miRNA clusters are expressed in totipotent cells ( Dataset S1 , Sheet 1; see below ) . In our samples only a discrete set of miRNAs dominates the small RNA repertoire of undifferentiated and early differentiating ES cells ( miRNAs with cloning frequencies >1%; Figure 1A ) , which agrees with the results of previous studies [17] , [22] , [27] . However , unlike in previous work attributing an ES-cell specific status to miR-21 [22] , this miRNA was cloned at a low frequency in all our samples ( 1 , 4% in female at D5 , <1% in all other samples ) . In fact its levels even appear to increase slightly during early differentiation at D2 ( Figure 1A; Dataset S1 ) . Additionally , miR-15b and miR-16 , which were among the most abundant small RNAs identified by Calabrese et al . in undifferentiated mouse ES cells , were either not ( miR-15b ) or moderately ( miR-16 ) represented at D0 in our experiments . In fact , the levels of miR-16 increased steadily throughout differentiation and it was one of the most abundant miRNA in male and female cells at D5 ( Figure 1A ) , clearly indicating that miR-16 is not a reliable indicator of ES cell pluripotency . Furthermore , elevated ( >1% ) expression of the oncogenic miR-17_92 cluster was only apparent in our D2 ( 7% ) and D5 ( 10 . 5% ) samples ( Figure 1A ) , but not in undifferentiated ES cells , unlike in the previous study where members of this cluster were cloned at a frequency of 11% from undifferentiated cells [22] . Most of these discrepancies can be readily explained by the fact that the ES cell lines and differentiation conditions used in these previous studies were different to ours . The fact that we did not use RA as a differentiating agent , which is known to induce expression of specific sets of microRNAs targeting Nanog , Oct4 ( also known as Pou5f1 ) and Sox2 coding sequences [28] , could account , for instance , for the low accumulation of specific miRNAs in our samples , including miR-15b . Our differentiation protocol also induced slow differentiation of specific lineage found in mouse embryo as presented in Figure S1 with little variation between the two ES male and female ES cell lines tested . Secondly , our ES cells were grown using feeder-free conditions , while other studies employed embryonic fibroblasts feeder cells for the culture of ES cells , and we note that miR-21 is one of the most abundant miRNAs found in fibroblasts [27] . Furthermore , a distinct analysis of SOLiD-based small RNA sequencing data obtained from an independent XY ES cell line grown on feeders indeed showed a significant enrichment ( 5% ) of cloned miR-21 at D0 ( C . C , N . S , E . B and H . E; data not shown ) . Therefore , the presence of miR-21 is unlikely to be a genuine feature of undifferentiated mouse ES cells . In some of the previous studies , the use of feeder cells also imposed a lag phase of about 24 h following their removal by adsorption . Such a procedure may induce the very earliest steps of ES cell differentiation . Indeed , we note that the miRNA profile of our D2 samples is remarkably similar to that reported by Calabrese and colleagues , raising the possibility that their analysis involved a mixture of very early differentiating and undifferentiated ES cells , rather than bona fide undifferentiated cells . This would also explain why miR-22 -which we cloned at a high frequency specifically in both male and female undifferentiated samples- was not overrepresented in their study . Our high resolution time-course analysis of the most abundant miRNAs from two distinct mouse ES cell lines has enabled to discriminate pluripotency patterns from early differentiation patterns . Notably , the miR-290_295 cluster , miR-127 and miR-22 contribute collectively to more than 65% of all cellular miRNAs of undifferentiated ES cells , and their respective abundance consistently decreases during early differentiation . Similar figures were also obtained in two additional , independent analyses employing the SOLEXA and SOLiD deep sequencing technologies ( data not shown ) . We propose , therefore , that these molecules represent reliable small RNA markers of pluripotency . We further distinguish abundant miRNAs present in undifferentiated cells , the expression of which increases during differentiation ( exemplified by miR-16 ) , from those that are initially only moderately or poorly expressed at D0 , but are highly abundant by D2 and D5 of differentiation ( Figure 1A ) . The latter includes the well-characterized miR-17_92 cluster , which targets several tumor suppressors and is enriched in many types of cancer ( for a review , see [29] ) , but also members of the miR-27 family , which suppresses expression of the breast cancer marker CYP1B1 [30] . Such co-expression of pro- as well as anti-oncogenic miRNAs might ensure that processes favoring cell proliferation versus acquisition of cell identity are appropriately balanced in early differentiating ES cells . There is currently no predictable correlation between the level of miRNA accumulation and their efficiency in suppressing gene expression . On the one hand , only a fraction of miRNAs that accumulate at saturating levels ( eg the miR-290 cluster ) might effectively recruit miRNPs for target regulation . On the other hand , moderately expressed miRNAs might be present at sufficient levels to suppress low-abundance transcripts . To address this issue objectively , we used a clustering analysis based on partitioning ( Partitioning Around Medoids , PAM ( Kaufman & Rousseeuw , 1990; see Materials and Methods and Figure 1B ) that groups all miRNAs cloned at a statistically significant frequency ( >0 . 05% ) into classes with correlated expression profiles . miRNAs in differentiating ES cells can be divided into three highly contrasted expression clusters designated A , B and C ( Figure 1B ) , which can be further refined into 10 sub-groups , detailed in Figure S6 ( See also Materials and Methods for details ) . miRNAs in class A are present in undifferentiated ES cells ( D0 ) , but progressively decrease in abundance as differentiation proceeds . miRNAs grouped into class B show an inverse pattern to cluster A , i . e . an increase in abundance during differentiation . Class C , on the other hand , is characterized by a peak of expression at Day 2 , followed by a decrease at Day 5 . We believe that these expression patterns are of biological relevance , because many miRNAs known to be neighbours along the genome ( and thus likely to be co-regulated ) were grouped together within the same PAM classes ( Figure 1B , Figure S6 ) . Further strengthening this idea , use of an in silico boostraping procedure employing a randomized miRNA expression dataset indicated that the observed PAM clustering in the real data set cannot occur by chance ( Figure S7 , Materials and Methods ) . Moreover , we experimentally verified that representative members of the 3 major PAM classes ( chosen not to be in the same miRNA genomic clusters ) do indeed exhibit the predicted expression profile ( Figure S8 ) . Therefore , analysis of discrete expression subclasses ( as presented in Figure S6 ) should provide an important handle with which to link distinct miRNA gene families functionally , since their coordinated expression likely entails the targeting of related functions within common cellular pathways . This analysis also reveals for the first time that miRNA temporal expression patterns can be extremely narrow , as illustrated with members of PAM class C , epitomized here by miR-182 and miR-27a ( Figure 1A–1B and Figure S8 ) . The unexpected peak of expression at Day 2 for this class of miRNAs could not have been appreciated in previous , single time-point analyses . This type of transient expression pattern must be linked to early loss of pluripotency and/or initiation of lineage-specific expression pathways and such miRNAs merit future investigation for their functions in early mammalian development . Additionally , they represent new and useful markers for early ES cell early differentiation . Our time course analysis during differentiation provided us with a unique opportunity to dissect the specific regulation of individual members of genomically clustered miRNA genes . We first focused our attention on the highly expressed miR-290 cluster located on chromosome 7 , a potent marker of mouse ES cell pluripotency ( Figure 1A–1B ) [17] , [22] . The mouse miR-290 cluster was recently shown to target several key cell cycle regulators and transcriptional repressors to enable rapid G1-S transition and maintenance of DNA de-methylation , two defining features of stem cells [19] , [31] . Interestingly , both studies involved the rescue of either Dicer−/− or Dgcr8−/− ES cell defects through ectopic expression of the entire miR-290 cluster or some of its highly abundant members [20] , [32] . These and other studies thus point to the effects of the miR-290 cluster being due to a single , coordinated expression unit with functionally redundant products . In agreement with these previous studies , separate time-course analyses of each member revealed that only 4 miRNAs , which share the same AAAGUGC 5′ seed sequence ( miR-291a-3p , miR-292-3p , miR-294 , miR-295 , Figure 2A , blue ) , likely contribute significantly to the global trend of miR-290 cluster expression ( reduced throughout differentiation , PAM class A; Figure 2A , grey ) . Nonetheless , PAM classification of individual miRNAs revealed sharp differences in temporal expression between some members of the cluster , and , more unexpectedly , between the mature and presumed passenger strand sequences of the most abundant miRNAs , of which three out of four were grouped into PAM class C rather than PAM class A ( Figure 2B ) . Moreover , these 3 presumptive miRNA* and the class C miR-290-5p all have the same 5′ seed sequence ACUCAAA ( C/A ) , a feature also shared by the class A miR-292-5p ( Figure 2A , yellow boxes ) . This common characteristic , together with their cloning frequencies being well above background ( e . g . miR-295* ) , suggests that several of these small RNA might be specifically engaged into a common regulatory pathway at D2 of differentiation . In agreement with this hypothesis , widespread functional recruitment of miRNA* has been inferred in Drosophilid , and several specific cases were recently experimentally validated in vitro and in vivo [33] . Perhaps even more compelling , the analysis of the miR-290 cluster also revealed an unexpected expression profile for the highly abundant miR-293 . This miRNA shows the opposite pattern to all of the other highly expressed members of the miR-290 cluster ( increase throughout differentiation , PAM class B , Figure 1B ) , suggesting drastically distinct targets and cellular functions for this specific miRNA . Indeed , a gene ontology ( GO ) analysis ( http://www . mirz . unibas . ch/ElMMo2/ ) of predicted target transcripts revealed a consensus set of regulated cellular functions for miR-291a-3p , miR-292-3p , miR-294 and miR-295 , but not for miR-293 ( Dataset S2 for the EIMMo target prediction software and Dataset S3 for the Pictar target prediction software ) . Accordingly , a seed inspection uncovered a completely different sequence for miR-293 ( Figure 2A , red ) , thereby confirming its singular status within the miRNA-290 cluster . A recent study also showed that several miRNA of the miR-290 cluster could individually help reprogramming mouse fibroblasts into induced pluripotent cells [34] . However , this could not be achieved with miR-293 , indicating different functions for this specific miRNA . The miR-290 cluster comprises two pri-miRNA giving rise to six pre-miRNAs , of which pre-miR-293 , premir-294 and premiR-295 are produced from the same primary transcript [17] , [18] . Thus , the most likely explanation to the result obtained in our analysis is a specific , post-transcriptional regulation of pre-miR-293 or mature miR-293 . In any case , these results show that the contribution of the miR-290 cluster to pluripotency cannot be interpreted in terms of a single , coordinated expression unit with redundant products . Collectively , these data reinforce the growing view that miRNA genes undergo extensive post-transcriptional regulation through mechanisms that selectively affect pri-miRNA processing and/or pre-miRNA stabilization [35] , notwithstanding possible effects on mature miRNAs , as recently suggested in plants [36] . These refinements in gene expression suggest that the regulatory potential and versatility of miRNAs is likely much broader than initially anticipated . Given its abundance , single chromosomal location and well-defined composition , studies of the miR-290 cluster in the ES cell-based system described here could help addressing these important issues further . The second most highly expressed miRNA cluster in undifferentiated ES cells is located on chromosome 12 and contains a total of 26 members , as annotated in miRbase ( Figure 2C ) . Among these , a short cluster of maternally expressed miRNAs genes ( miR-431 , miR-433 , miR-127 , miR-434 and miR-136 ) is transcribed and processed from an antisense gene to the paternally expressed Retrotransposon-like 1 ( Rtl1 ) gene , the recently characterized protein product of which appears to be indispensable for mouse foetal development [37] . Due to their perfect complementarity to Rtl1 , the above miRNAs have been proposed to mediate trans-allelic RNAi at the Rtl1 locus in a variety of mouse embryonic tissues , based on Northern and 5′ RACE analyses {Seitz , 2003 #930; Davis , 2005 #918} . However , their respective contribution to Rtl1 silencing has not been addressed . Our time-course analysis revealed that of these five miRNAs , only miR-127 , miR-434 and miR-136 , are likely to contribute to Rtl1 silencing in ES cells ( Figure 2D ) because their cloning frequencies largely exceeds that of the other members of the cluster , which accumulate at background level ( Figure 2C , Dataset S1 ) . The three miRNAs can be further distinguished based on their expression profiles and respective cloning frequencies , with miR-127 contributing alone 10% of all cloned miRNAs at D0 . Unlike miR-434 and miR-136 ( PAM class C ) , miR-127 expression gradually decreases during the differentiation process ( PAM class A ) , a pattern inversely correlated to that of Rtl1 expression in ES cells , as assayed by Quantitative Reverse-transcriptase PCR ( Q-RT-PCR; Figure 2D ) . We conclude that miR-127 is likely the major contributor of Rtl1 silencing in differentiating mouse ES cells . Thus , in contrast to the complex situation described above for the miR-290 cluster , in this particular case the PAM analysis of temporal miRNA/target variations shows that the effect of a miRNA gene cluster can be probably equated to that of a single miRNA member within it . One goal of this study was to identify putative sex-specific small RNAs , including miRNAs . To this end , we reanalyzed separately the data in male and female ES cell samples , in order to set apart outlier miRNAs the distribution of which diverged significantly from the median obtained upon analysis of both sexes ( Figure S5 ) . Few miRNAs were found to be differentially expressed between sexes ( using MA plot transformation see Materials and Methods ) ; and for those that were , their level tended to increase with differentiation ( Figure 3A ) . Although some of these variations in expression between male and female ES cells could be attributed to differences in the rates of differentiation between the ES cell lines used , the greatest and most striking difference between the two sexes was observed at D5 with the miR-302 genomic cluster . For this cluster , all five members ( including some miRNA* sequences; Figure 3A , highlighted in green ) were cloned at a significant and much higher frequency in male ( >10% of all miRNAs cloned ) as opposed to female D5 differentiated ES cells ( <0 . 5% of all miRNAs cloned , Figure 3B; Dataset S1 ) . This suggests a strong and specific transcriptional enhancement of the miR-302 gene in male cells . Northern analyses of total RNA confirmed that the levels of miR-302d ( one of the most frequently cloned representatives of the miR-302 cluster , Figure 3B , upper panel ) are at least 20 fold higher in D5 male cells than in D5 female cells ( Figure 3B , lower panel ) . Further support for a male-specific enrichment of the miR-302 family came from Q-RT-PCR analyses showing that XPGKO ES cells ( lacking a Y or a X chromosome ) displayed only a very minor increase in miR-302d content , similar to female XX cells at D5 ( Figure 3D ) . We could also rule out possible cell line-specific effects , because Q-RT-PCR analyses of mature miRNAs from two independent male ( E14; XY1 and HM1; XY2 ) as well as two independent female ( PGK; XX1 and LF2; XX2 ) cell lines gave similar results ( Figure 3C ) . This Q-RT-PCR analysis also revealed that in differentiating female samples , miR-302d could also be detected , albeit at much lower expression levels . To examine further this male-specific differentiation miR-302 expression pattern , RNA from ES cells that had undergone differentiation for 10 days was also examined . This revealed that the increase in miR-302 observed at D5 in males is transient because it had decreased by D10 ( Figure 3C ) . This highly dynamic and apparently sex-specific pattern suggested that the miR-302 family might have a role during a narrow window of male development . To investigate this possibility in vivo , we monitored mirR-302d expression in various tissues of adult mice ( Figure 3E ) . However , expression of this cluster was at or below detection limit of Q-RT PCR analysis in all sampled tissues ( Figure 3E ) . Based on the strong enrichment of miR-302 at Day 5 of early differentiation and its male ES cell-specificity , we thus compared its accumulation in dissected gonads of male and female embryos from 13 . 5 to 10 . 5 dpc ( Figure 3F ) . No sex-related difference could be detected at these time points , although miR-302d expression was clearly higher at 10 . 5 dpc in both males and females , and reached background levels at 13 . 5 dpc ( Figure 3F ) . These results thus suggested that any putative sex-specific embryonic expression of miR-302 should occur at earlier stages , before colonization of the gonads by primordial germ cells ( PGC ) , which are also known to form during early ES cell differentiation ( for a review , see [38] ) . To address this issue , we analyzed miR-302 expression in male and female pluripotent embryonic germ cells ( EG ) derived from PGCs that had been isolated at various stages of embryogenesis ( 8 . 5 , and 9 . 5 dpc ) . Indeed , while it was at background levels in female samples , a strong male-specific enrichment of miR-302d was observed in EG cell lines at both stages , and particularly at 9 . 5 dpc ( Figure 3G ) . Our ES cell differentiation analysis thus uncovers the first example of sex-specific regulation of a mammalian miRNA . A recent study indicates that members of the miR-290 and of the oncogenic miR-17_92 cluster are among the most abundant miRNAs found in proliferating mouse PGCs [39] . Although this study involved a mixture of male and female embryos and thus , could not identify the sex-specific enrichment of miR-302 members , this profile resembles that of D2 and D5 female ES cells in our early differentiation system . Interestingly , all members of the miR-302 cluster share the same AAGUGC ( U/C ) 5′ seed sequence with the highly expressed members of the miR-290 cluster ( with the notable exception , of course , of miR-293; Figure 2A–2B ) . It has thus been speculated that the two clusters carry out similar functions , particularly in totipotency , as shown in human ES cells [40] . This idea was recently given some experimental support by the demonstration that certain proliferation defects of Dgcr8−/− mouse ES cells are rescued to a similar extent through ectopic expression of individual members of either the mouse miR-302 or miR-290 clusters [32] . These observations together with the results of the present study thus predicted that some shared targets of the miR-302 and miR-290 clusters should be specifically downregulated in male embryonic stem and germ cells , in which the miR-302 family accumulates at much higher levels than in female cells . To identify cellular targets of the miR-302 family , we used two different algorithms: Pic-Tar [41] and the EIMMo microRNA target prediction server ( http://www . mirz . unibas . ch/ElMMo2/ ) . These softwares search mRNA 3′-UTRs for the presence of conserved 7-mers matching the seed region of queried miRNAs . Strikingly , both algorithms identified the 3′-UTR of the Arid4a and Arid4b paralogous genes as first ranking candidates ( Dataset S2 and Dataset S3 ) . Other high-scoring candidates included a set of genes that had been previously validated as targets for miR-290 in mice . Both human ARID4 paralogs were previously known as retinoblastoma-binding protein 1 ( RBBP1 or RBP1 ) [42] . They serve as adapters to recruit the mSin3A-Histone deacetylase ( HDAC ) to E2F-dependent promoters undergoing transcriptional repression by Rb [43] . We decided to focus on the Arid4b gene , whose 3′-UTR contains three evolutionary conserved matches for the seed shared by miR-302 and miR-290 ( Figure 4A–4B ) . Western Blot analysis at D0 , D2 and D5 revealed a steady accumulation profile for the Arid4b protein throughout early differentiation of female mouse ES cells ( Figure 4C , left panel ) , in which miR-302 levels remained low ( Figure 3B–3C ) . In contrast , there was a progressive loss of Arid4b accumulation in male cells ( Figure 4C , left panel ) , a pattern inversely correlated to that of miR-302d levels ( Figure 3B–3C ) . Q-RT-PCR analyses showed that slight changes in Arid4b mRNA levels were observed throughout differentiation of male cells . However , these slight changes could not account for the strong reduction in Arid4b protein levels in the same samples , thereby indicating an effect at the protein level consistent with a seed/3′-UTR type of regulation by the miR-302 family . To validate the predicted interaction between miR-302 and Arid4b , we used a dual luciferase-based reporter assay . The entire 3′-UTR of the endogenous Arid4b mRNA was inserted dowsntream of the open reading frame of a Renilla luciferase reporter gene . Expression of the resulting construct was then measured in transfected human HEK-293 cells , which are devoid of miR-302 ( data not shown ) . The authentic miR/miR* duplex of miR-302d was then chemically synthesized and transfected into cells together with the reporter plasmid . An unrelated siRNA duplex was transfected in parallel , as a negative control . We also used a miR-291a-3p duplex as a representative of the miR-290 family , of which several members are also predicted to target Arid4b owing to seed identity with miR-302 ( Figure 4D , Dataset S2 and Dataset S3 ) . 24 h post-transfection , a similar decrease in Renilla luciferase reporter gene activity was observed with both miR-302d and miR-291a-3p treatments , but not upon transfection of the control siRNA duplex ( Figure 4D ) , indicating a sequence-specific effect . Together with our ES cell time-course analysis ( Figure 3B and 3D; Figure 4C ) these results strongly suggest ( i ) that Arid4b is a common mRNA target of miR-302 and miR-290 family members and ( ii ) that the differences in Arid4b levels between male and female D2 and D5 ES cells are due to the male-specific accumulation of miR-302 . By extension , it can be inferred that many common targets of miR-302 and miR-290 are likely to undergo similar differences in expression in tissues or cell types showing sex-related polymorphism of miR-302 expression , including EGs ( Figure 3G ) and , presumably , PGCs . One possible explanation to this male-specific pattern is that it might be generated by the subpopulation of male ES cells that goes down the germline differentiation pathway . Thus , it could reflect an important male-specific genetic program , possibly normally initiated in male PGCs . However , testing this idea in ES cells will first require the isolation of germline cells within the population , which , though technically challenging , represent an interesting perspective of the present work . Members of the miR-302 and miR-290 clusters can individually rescue the proliferation defects of mouse Dgcr8−/− ES cells [32] . In this context , identification of Arid4 as a novel target of both miR-290 and miR-302 is entirely consistent with the established role of ARID4 as an Rb-mediated repressor of E2F-dependent transcription , which is mandatory for the G1-S phase transition in the cell cycle [44] . These data can now be assembled into a comprehensive , albeit still speculative model , integrating the possible overlapping functions of miR-290 and miR-302 in mouse cell totipotency during early development ( Figure S9 ) . This study demonstrates that small RNA sampling throughout early ES cell differentiation enables the definition of expression patterns for most cellular miRNAs . We have further shown that the transience of some of these miRNA patterns provides highly discriminative markers of particular ES cell states during their differentiation , an approach that might be broadly applicable to the study of early mammalian development . Our study also underscores the benefit of unsupervised classification analyses in deciphering complex regulations of miRNAs cistrons , notably by uncovering outlier members within miRNA clusters , as shown here with the surprising findings made with miR-293 . The analysis finally unravelled a puzzling enrichment of miR-302 expression during male ES cell differentiation as well as in male embryonic germ cells , suggesting a contribution of this miR family to male germline determination . We are currently in the process of generating miR-302 conditional-deletion ES cells in order to produce miR-302-deficient mice . Analysis of these animals , and notably of their germlines , might provide important insights into sex-specific miRNA regulations .
Female PGK ( XX1 ) and LF2 ( XX2 ) ES cell lines , male E14 ( XY1 ) and HM1 ( XY2 ) cell lines , and the XPGKO cell line ( from Dr Neil Brockdorff laboratory ) were cultured in Dulbecco's Modified Eagle Media ( DMEM ) ( Invitrogen ) , containing 15% FCS ( Bio West ) , 1000 U/ml LIF ( Chemicon ) , 0 . 1 mM 2-mercaptoethanol ( Invitrogen ) , 0 . 05 mg/ml of streptomycin ( Invitrogen ) and 50 U/ml of penicillin ( Invitrogen ) on a gelatin-coated support in the absence of feeder cells . The 4 EG cell lines ( have been derived in Dr Stephane Viville laboratory ) were cultivated with feeder cells under the same conditions . Differentiation was induced spontaneously into LIF-free DMEM , 10% FCS medium , at day 2 , 5 and 10 of differentiation . The culture medium was changed daily . All cells were grown at 37°C in 8% CO2 . Total cellular RNA from XX1 and XY1 cell lines at D0 , 2 , 5 was prepared using Trizol reagent ( MRC Molecular Research Center ) following the manufacturer's instructions . Small RNA cloning was performed as described in [45] using 200 µg of total RNA per library . Libraries were sequenced using the 454 technology ( http://www . 454 . com ) . Sequences were annotated with blast ( word size = 7/no filter ) using the following databases as references . Genomic sequences were retrieved from release mm9 of the mouse genome from the UCSC Genome Browser database ( NCBI build 37 , July 2007 ) . tRNA , rRNA , and other non-coding RNA sequences were extracted from release 158 of Genbank ( February 15 , 2007 ) , microRNA precursor sequences were extracted from miRbase ( release 10 . 1 , December 2007 ) . The results were filtered to authorize 0 , 1 or 2 mismatches per small RNA sequence , to take into account polylorphism and sequencing errors . Northern blot analysis was as described [46] . 30 µg of total RNA were used per lane . Hybridization probes corresponded to 5′ 32P-radiolabelled oligodeoxynucleotide complementary to the miR-302d sequence or to part of the U6 snRNA sequence ( used as loading control ) . Blots were analyzed and quantified by phosphorimaging ( FLA7000 scanner; Fuji ) . Real-time PCR reagents for miRNAs and control U6 snRNA were from Qiagen . For RT reactions , 1 µg total RNA was reverse transcribed using the miScript Reverse Transcription Kit ( Qiagen ) following the manufacturer's instructions . Following the RT reactions , cDNA products were diluted five times in distilled water , and 2 µl of the diluted cDNAs was used for PCR using QuantiTect SYBR Green PCR Master Mix and miScript Universal Primer ( Qiagen ) . The PCR reaction was conducted at 95°C for 10 min , followed by 40 cycles at 95°C for 15 s and 60°C for 30 s on a LightCycler 480 real-time PCR machine ( Roche ) . Real-time PCR for mRNAs was performed as described in [47] using the Rrm2 as a reporter . Differences between samples and controls were calculated based on the 2-ΔΔCP method . Each Real-time PCR reaction was carried out in triplicates using samples from three independent differentiation events of the four ES cell lines ( PGK , E14 , LF2 , HM1 ) . For Figures S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 , and S9 , Q-RTPCR analyses ( miRNA and mRNA ) involved two independent differentiation events of pools of female ( PGK , XX1 and LF2 , XX2 ) and pools of male ( E14 , XY1 and HM1 , XY2 ) cell lines , respectively . Western blotting was performed using standard procedures . The Brcaa1 ( Arid4b; ab36962 ) antibody was purchased from Abcam , ( Cambridge , UK ) . The 3′ UTR of Arid4b ( 614 nt ) was amplified from DNA extracted from E14 ( XY1 ) ES cells using attB-containing primers ( internal primers for the first PCR: Fwd-5′-AAAAAGCAGGCT CCATCAATGTCCAGTGCATC-3′ , Rev-5′-AGAAAGCTGGGTTTTGGTTACCAGGATGATGTCT-3′ and external primers for the second PCR: Fwd-5′-GGGGACAAGTTTGTACAAAAAAGCAGGCT-3′ , Rev-5′-GGGGACCACTTTGTACAAGAAAGCTGGGT-3′ ) . The PCR fragment was cloned into the attB-site of pDONR/Zeo ( Invitrogen ) , checked for orientation , sequenced and cloned into the psiCHECK-2 vector ( Promega ) using Gateway cloning Technology . The resulting plasmid was named psiCHECK-Arid4b . For reporter assays , HEK-293 cells were transiently transfected with psiCHECK-Arid4b together with miR-302d , miR-291-3p and control siRNA ( as indicated in Figure 4D ) using lipofectamine 2000 ( Invitrogen ) . Reporter assays were performed 24 h post-transfection using the Dual-luciferase-assay-system ( Promega ) , normalized for transfection efficiency by Renilla-luciferase , also present in psiCHECK-Arid4b . Each experiment was done in triplicate and reproduced twice independently . The distribution of miRNA frequency was fitted with a Gaussian mixture model . A Bayesian Information Criterion allowed selection of two components . The first component is truncated at zero and corresponds to miRNAs with low counts ( interpreted as background ) . The second component corresponds to miRNAs showing statistically significant expression level . These two components indicated that a reasonable threshold for the “background” expression is T = 0 . 05% . The dataset was filtered accordingly and all miRNAs with a frequency lower than T were removed from the analysis . The similarity of miRNA profiles between the two sexes was tested via a correlation analysis at each time point ( 0 , 2 and 5 days ) . The Pearson correlation assesses the linear relationship between two variables ( here miRNA expression in male XY1 and in female XX1 ) . The Spearman correlation is equivalent to the Pearson correlation , but uses miRNA expression rank as variable . The Pearson correlation therefore assesses similarities in miRNA ranking between male and female samples . An unsupervised clustering approach was used to group into cluster those miRNAs showing similar expression profiles throughout days 0 , 2 and 5 of differentiation . Because miRNA expression data in both sexes are highly correlated , they were used as replicates and averaged for clustering . A partitioning analysis ( PAM – partitioning around medoids ) was then performed using the Pearson correlation as a measure of similarity . PAM classifies objects in a given number k of groups , each of them being represented by a medoid miRNA indicated in red in Figure 1B . The number of clusters k was chosen as to maximize the average silhouette width of the classes . The silhouette is a measure of the quality of the clustering , based on the difference between the average distance of a given miRNA to all other objects of its class , and the distance between this miRNA to the closest one outside its class . Calculating the silhouette scores for k = 3 to k = 10 , showed that k = 3 achieves the best score . These three classes correspond to a peak of expression at day 0 , 2 and 5 respectively . We also tested clustering at a higher resolution ( 5<k< = 10 ) . In this case , the division into 10 clusters had the best silhouette score and these 10 classes correspond to an exact subdivision of the 3 major classes . One thousand randomized datasets were generated from the experimental miRNA expression set ( by bootstrapping expression profiles ) , and the PAM clustering was performed for k = 3 . For each dataset , the number of pairs of miRNAs from the same genomic cluster present in the same PAM class was counted . This provided an estimate of the distribution of the number of pairs of neighbour miRNAs in a randomized dataset . The results were then compared to those obtained from the experimental dataset ( indicated by a red dot in Figure S7 ) . P-values show that PAM clustering in the experimental data set is very unlikely to occur by chance . In order to unravel sex specific miRNAs , we have to take into account that variability of occurrence is smaller for the low-expressed miRNAs than for the high-expressed ones . As a consequence , data were transformed as follows , and represented as MA plot . M , the ratio of the miRNAs profiles M : log2 ( Dxx ) −log2 ( Dxy ) A , the average miRNAs level , A = ( log2 ( Dxx ) +log2 ( Dxy ) ) /2 The data were split into bins of similar intensities . The number of bins was set to 5 in order to retrieve enough miRNAs so as to estimate the distribution . Outlier miRNAs were identified in each bin by estimating the variance of the bin after discarding the miRNA and then estimating the probability for the miRNA of being an outlier using a Gaussian distribution . All miRNAs with a p-value lower than 5% were judged as significant and considered as being differentially expressed . No multiple testing corrections were applied .
|
The discovery of the first microRNA ( lin-4 ) in C . elegans in 1993 and the increasing realization that small RNAs are at the heart of many biological processes have led to a revolution in our thinking about development and disease . In animals , several hundred microRNAs ( miRNAs ) have been identified that regulate diverse biological processes ranging from cell metabolism to cell differentiation and growth , apoptosis , and cancer . Moreover , it has been shown that many miRNAs are characterized by highly specific spatial and temporal expression patterns supporting their role in such processes . However , the dynamics of small RNA patterns in male and female embryonic stem ( ES ) cells in the course of early differentiation has not been investigated so far . Our work represents the first study of this kind . Notably , we have identified new classes of miRNAs that show extremely defined temporal profiles during ES cell differentiation , as well as sex-specificity . Our results are of broad interest and importance because they raise the power of ES cells in defining the repertoire of small RNAs and their dynamics in mammals , and underline the importance of integrating miRNA expression patterns into the transcription factor networks and epigenomic maps defined in ES cells in order to provide a better understanding of the control of pluripotency and lineage commitment .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] |
[
"developmental",
"biology/cell",
"differentiation",
"developmental",
"biology/germ",
"cells",
"developmental",
"biology/stem",
"cells",
"genetics",
"and",
"genomics/bioinformatics"
] |
2009
|
Highly Dynamic and Sex-Specific Expression of microRNAs During Early ES Cell Differentiation
|
The human T-cell leukemia virus type 1 ( HTLV-1 ) Tax protein hijacks the host ubiquitin machinery to activate IκB kinases ( IKKs ) and NF-κB and promote cell survival; however , the key ubiquitinated factors downstream of Tax involved in cell transformation are unknown . Using mass spectrometry , we undertook an unbiased proteome-wide quantitative survey of cellular proteins modified by ubiquitin in the presence of Tax or a Tax mutant impaired in IKK activation . Tax induced the ubiquitination of 22 cellular proteins , including the anti-apoptotic BCL-2 family member MCL-1 , in an IKK-dependent manner . Tax was found to promote the nondegradative lysine 63 ( K63 ) -linked polyubiquitination of MCL-1 that was dependent on the E3 ubiquitin ligase TRAF6 and the IKK complex . Tax interacted with and activated TRAF6 , and triggered its mitochondrial localization , where it conjugated four carboxyl-terminal lysine residues of MCL-1 with K63-linked polyubiquitin chains , which stabilized and protected MCL-1 from genotoxic stress-induced degradation . TRAF6 and MCL-1 played essential roles in the survival of HTLV-1 transformed cells and the immortalization of primary T cells by HTLV-1 . Therefore , K63-linked polyubiquitination represents a novel regulatory mechanism controlling MCL-1 stability that has been usurped by a viral oncogene to precipitate cell survival and transformation .
Human T-cell leukemia virus 1 ( HTLV-1 ) infects approximately 20 million people worldwide and is the etiological agent of adult T-cell leukemia ( ATL ) , an aggressive CD4+CD25+ malignancy that occurs in a small percentage of infected individuals after a long latent period [1] . HTLV-1 infection is also associated with a host of inflammatory diseases including HTLV-1-associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) . HTLV-1 Tax is a key regulatory protein essential for viral gene expression by recruiting CREB/ATF transcription factors to the viral long terminal repeats ( LTRs ) [2] . Tax also plays a central role in cell transformation by HTLV-1 and is sufficient to immortalize primary human T lymphocytes [3] . Furthermore , transgenic mice expressing Tax in the T-cell compartment develop leukemia and lymphoma with clinical and pathological features resembling ATL [4] . Tax stimulates the proliferation , survival and immortalization of T cells by inactivating tumor suppressors , promoting cell cycle progression and activating anti-apoptotic pathways [5] . One of the principal cellular pathways targeted by Tax and essential for Tax-mediated transformation is NF-κB [6] . NF-κB is composed of heterodimeric DNA binding proteins containing RelA , c-Rel , RelB , p50 and p52 [7] . NF-κB is held inactive in the cytoplasm by members of the IκB family , all of which contain ankyrin repeat domains . In the canonical NF-κB pathway , a wide variety of stimuli including proinflammatory cytokines and stress signals converge on the IκB kinase ( IKK ) complex consisting of the catalytic subunits IKKα and IKKβ and the regulatory subunit IKKγ ( also known as NEMO ) [8] . IKKβ phosphorylates the NF-κB inhibitor IκBα to trigger its ubiquitination and degradation by the proteasome thus allowing NF-κB to translocate to the nucleus and activate anti-apoptotic and pro-inflammatory target genes [9] . In the noncanonical NF-κB pathway , specific tumor necrosis factor receptor ( TNFR ) superfamily members including BAFF , lymphotoxin-β and CD40 induce the proteasome-dependent processing of the p100 ( NF-κB2 ) precursor to yield p52 , which heterodimerizes with RelB to activate a distinct gene program [10] . Tax constitutively activates both canonical and noncanonical NF-κB pathways , in part by interacting with NEMO [11] , [12] . HTLV-1 molecular clones bearing Tax mutants defective for NF-κB activation are impaired in T-cell immortalization [6] . HTLV-1 transformed cell lines and primary ATL cells all exhibit constitutive NF-κB activation that is integral for the survival of these virally transformed lymphocytes [13] . Since Tax expressing cells are vigorously targeted for elimination by cytotoxic T cells , the majority ( ∼60% ) of ATL tumors exhibit downregulated or lost Tax expression , either by mutations within Tax or deletion or methylation of the 5′ LTR [14] . NF-κB activation is tightly regulated by post-translational modifications , with ubiquitin playing a prominent role in both canonical and noncanonical NF-κB pathways . Ubiquitin ( Ub ) is conjugated to a lysine residue in a substrate by the sequential process of three enzymes: Ub-activating enzyme ( E1 ) , Ub-conjugating enzyme ( E2 ) , and Ub ligase ( E3 ) [15] . Ubiquitin contains seven lysine residues ( K6 , 11 , 27 , 29 , 33 , 48 , 63 ) , each of which can support the elongation of polyubiquitin ( polyUb ) chains [16] . K48-linked polyUb chains direct substrates to the proteasome for degradation , whereas K63-linked polyUb chains mainly serve nondegradative roles including receptor trafficking , DNA damage repair , kinase activation and signal transduction [17] . In the NF-κB pathway , K63-linked polyUb chains conjugated to adaptor proteins ( e . g . RIP1 ) provide platforms for the recruitment of TAK1 ( TGF-β activating kinase 1 ) and IKK kinase complexes that signal downstream NF-κB activation [18] . Dysregulation of the ubiquitination machinery , in particular the E3 ligases that direct a protein to proteasomal degradation , may lead to uncontrolled cell growth and tumorigenesis [19] . HTLV-1 Tax is conjugated with K63-linked polyUb chains which plays an essential role in NEMO binding and Tax activation of IKK and NF-κB [20] , [21] . The E2 enzyme Ubc13 is required for Tax ubiquitination [20] , however the E3 enzyme that ubiquitinates Tax has yet to be identified . Although Tax hijacks ubiquitin for NF-κB activation , whether Tax utilizes the ubiquitin machinery for distinct events in the transformation process remains unknown . We postulated that ubiquitination events downstream of Tax might play important roles in HTLV-1-induced cellular transformation; therefore , a proteome-wide quantitative survey of cellular proteins modified by ubiquitin in the presence of Tax was undertaken . This endeavor led to the identification of the anti-apoptotic myeloid cell leukemia-1 ( MCL-1 ) protein as a key target that was ubiquitinated in a TRAF6-dependent manner downstream of Tax . Tax interacted with TRAF6 and facilitated its mitochondrial localization where it conjugated MCL-1 with K63-linked polyubiquitination chains to enhance MCL-1 stability and protection from genotoxic stress-induced degradation .
To identify cellular proteins ubiquitinated in a Tax-dependent manner Jurkat cells expressing tetracycline-inducible Tax and TaxM22 , a mutant ( T130A , L131S ) impaired in NEMO binding and NF-κB activation [11] , [22] , were subjected to the UbiScan ubiquitination proteomics platform . Ubiquitinated peptides were enriched from cell lysates using the ubiquitin branch ( “K-ε-GG” ) antibody and subjected to liquid chromatography-tandem mass spectrometry ( LC-MS/MS ) analysis . Tax expression was confirmed by immunoblotting in Tax and TaxM22-induced cells ( Figure 1A ) . A total of 136 proteins ( total of 204 Ub sites ) were identified with ubiquitination profiles modulated in response to Tax expression ( Table S1 ) . Furthermore , 22 of these candidates , including the anti-apoptotic BCL-2 family member MCL-1 , were ubiquitinated in a Tax and IKK-dependent manner ( by Tax but not TaxM22 ) ( Figure 1B ) . MCL-1 is localized in the mitochondria and directly binds and antagonizes pro-apoptotic BCL-2 family members BAX and BAK , as well as BH3 only family members BIM , BID , NOXA and PUMA , to restrain cytochrome C release from the mitochondria and inhibit apoptosis [23] . MCL-1 has an amino ( N ) -terminal extended PEST ( rich in proline ( P ) , glutamic acid ( E ) , serine ( S ) and threonine residues ( T ) ) domain that regulates its stability [24] . MCL-1 is a highly labile protein with a half-life of 2–4 hours , and its expression is rapidly diminished in response to apoptotic stimuli , mainly through the ubiquitin-proteasome degradation pathway [25]–[29] . To determine the type of polyUb chains conjugated onto MCL-1 in response to Tax expression , we performed an ubiquitination assay with HA-tagged wild-type ( WT ) , K48-only , or K63-only Ub plasmids . MCL-1 was conjugated predominantly with K63-linked polyUb , and with a lesser extent K48-linked polyUb when co-expressed with Tax ( Figure S1 ) . This result was confirmed with endogenous ubiquitin chains by immunoblotting with a K63-Ub linkage-specific antibody in Tax-transfected 293T cells , Tax-inducible Jurkat Tet/On-Tax cells , and the HTLV-1 transformed T-cell lines MT-2 , SLB-1 and TL-OM1 ( Figures 1C–E ) . TL-OM1 is an ATL cell line lacking Tax expression [13] . MT-2 and SLB-1 cells express Tax; in addition MT-2 cells also harbor an envelope ( Env ) /Tax fusion ( Figure 1E ) . MCL-1 K63-linked polyubiquitination correlated with Tax expression , with MT-2 cells exhibiting higher levels of MCL-1 ubiquitination compared to SLB-1 and TL-OM1 ( Figure 1E ) . To further confirm that the K63-linked polyubiquitination observed was specific to MCL-1 , and not an artifactual result caused by an associated protein , a ubiquitination assay was conducted with transfected His-tagged MCL-1 . His-MCL-1 was precipitated from lysates with Ni-NTA agarose beads and washed with buffer containing 8 M urea to eliminate any MCL-1 associated proteins . Consistent with earlier results , Tax specifically induced the K63-linked polyubiquitination of His-MCL-1 ( Figure S2 ) . Among K63-Ub specific E3 ligases , TRAF6 autoubiquitination appears to be enhanced by Tax expression [30] . Consistent with this report , our in vitro ubiquitination assays revealed that TRAF6 was ubiquitinated when incubated with purified Tax protein ( Figure S3 ) . However , TRAF6C70A , a RING domain point mutant impaired in E3 ligase activity , was not ubiquitinated in response to Tax ( Figure S3 ) , indicating that Tax triggers the enzymatic activity and autoubiquitination of TRAF6 . Moreover , Tax lost its ability to induce MCL-1 K63-linked polyubiquitination in cells expressing two distinct TRAF6 short hairpin RNAs ( shRNAs ) ( Figure 1F ) , indicating that Tax activation of TRAF6 is critical for MCL-1 ubiquitination . By inspection of the Tax protein sequence , we identified in silico a putative TRAF6 binding motif [31] in the carboxyl ( C ) -terminal tail of Tax ( Figure 1G ) . Interestingly , the putative TRAF6 binding motif lies within a domain essential for Tax transformation and is immediately adjacent to the PDZ binding motif ( PBM ) of Tax [32] . Indeed , co-immunoprecipitation ( co-IP ) analysis revealed that Tax specifically interacted with TRAF6 and not other TRAF proteins , but the interaction with TRAF6 was substantially diminished when the conserved glutamic acid residue in the TRAF6 binding motif was substituted with alanine ( TaxE345A ) ( Figure 1H and I ) . Conversely , TRAF6 did not interact with Tax when the TRAF-C domain of TRAF6 , a region that regulates oligomerization of TRAF6 and interaction with upstream signaling molecules , was deleted ( Figure 1J ) . Together , these results indicate that Tax induces the K63-linked polyubiquitination of MCL-1 and interacts with TRAF6 via a specific C-terminal motif . We next examined the effect of Tax mutants impaired in IKK activation ( M22 ) or TRAF6 binding ( E345A ) in the activation of TRAF6 by conducting TRAF6 auto-ubiquitination assays . Tax promoted the K63-linked autoubiquitination of TRAF6 , however TaxM22 and TaxE345A mutants were both impaired in inducing the TRAF6 autoubiquitination ( Figure 2A ) . TaxE345A only weakly activated TRAF6 , likely due to the residual binding of this mutant with TRAF6 ( Figure 1I ) . These results indicate that Tax requires both IKK/NEMO and TRAF6 binding to activate TRAF6 . Furthermore , both TaxM22 and TaxE345A were deficient in promoting MCL-1 K63-linked polyubiquitination ( Figure 2B ) . Therefore , Tax also requires interactions with IKK/NEMO and TRAF6 to induce the K63-linked polyubiquitination of MCL-1 . TRAF6 likely directly ubiquitinates MCL-1 since purified TRAF6 promoted the ubiquitination of recombinant GST-MCL-1 and co-expression of Tax with TRAF6 further enhanced MCL-1 ubiquitination ( Figure S4 ) . To further understand how Tax interaction with TRAF6 regulated MCL-1 ubiquitination , we examined the subcellular localization of TRAF6 since it has been shown to traffic to the mitochondria as part of its regulation of host innate immune signaling or mitochondria quality control [33] , [34] . Indeed , Tax specifically induced the mitochondrial translocation of TRAF6 but not other TRAFs ( Figure S5 ) . We next conducted biochemical fractionation and immunofluorescence assays to examine the effect of Tax mutants on TRAF6 translocation to the mitochondria . TaxM22 and TaxE345A were less effective than wild-type Tax in facilitating the translocation of TRAF6 to mitochondria ( Figure 2C and D ) . TaxM22 and TaxE345A were also deficient in the stabilization of MCL-1 ( Figure 2C ) . Interestingly , a significant fraction of Tax was found in the mitochondria where it partially co-localized with TRAF6 ( Figure 2C and D ) . However , TRAF6ΔTRAF-C , which does not interact with Tax ( Figure 1J ) , did not translocate to the mitochondria in response to Tax expression ( Figure S6 ) . TRAF6 was mostly localized in the mitochondria in HTLV-1 transformed MT-2 cells and was heavily modified , most likely by polyUb chains ( Figure 2E ) . The Tax- ATL cell line TL-OM1 exhibited less pronounced TRAF6 modification compared to MT-2 cells ( Figure 2E ) . Based on these collective findings , we postulated that Tax could mediate the interaction between TRAF6 and MCL-1 . Indeed , co-IP analysis revealed that Tax specifically enhanced the interaction between MCL-1 and TRAF6 ( Figure 2F ) . Although MCL-1 interacted with TRAF3 under basal conditions , Tax had no effect on this interaction ( Figure 2F ) . Three conserved TRAF6 binding motifs within the PEST domain of MCL-1 were identified and mutations were rendered within each motif as indicated , both singly and in combination ( Figure 2G ) . Notably , the other BCL-2 family members BCL-2 , BCL-x ( L ) and BFL-1/A1 all lack TRAF6 binding sites . In vitro GST pull-down assays revealed that the third TRAF6 binding motif was most critical for MCL-1 binding to TRAF6 since all mutants harboring mutations in the third TRAF6 binding motif failed to interact with TRAF6 ( Figure 2H ) . Taken together , Tax forms a complex with IKK and TRAF6 and triggers TRAF6 redistribution to the mitochondria where it conjugates MCL-1 with K63-linked polyUb chains ( Figure 2I ) . The collective data also raise the possibility that TRAF6 may play a role in Tax-mediated NF-κB activation . MCL-1 is a highly labile protein due to phosphodegron motifs in the PEST domain targeted by the FBW7 ( F-box and WD repeat domain-containing 7 ) E3 ligase complex [28] , [29] . Cycloheximide ( CHX ) chase assays were next conducted to examine MCL-1 stability , which revealed that Tax , but not TaxM22 , significantly prolonged the half-life of MCL-1 in Jurkat cells ( Figure 3A ) . Consistent with these results , Tax failed to enhance the half-life of MCL-1 protein in NEMO-deficient Jurkat cells ( Figure S7 ) , indicating that Tax requires IKK to stabilize MCL-1 . MCL-1 was more stable in Tax+ MT-2 cells compared to the Tax- ATL cell line TL-OM-1 and Jurkat cells ( Figure 3B ) . MCL-1 stability was also increased in Tax-expressing MT-4 cells compared to TL-OM1 cells ( Figure 3C ) . Tax depletion by two distinct shRNAs triggered the loss of MCL-1 in MT-2 cells together with cleavage of poly ADP ribose polymerase ( PARP ) , indicative of apoptosis ( Figure 3D ) . Consistent with these results , shRNA-mediated knockdown of Tax in MT-4 cells with shRNAs #2 and 4 correlated with loss of MCL-1 and PARP ( Figure 3E ) . Therefore , MCL-1 protein levels are under the strict control of Tax in HTLV-1 transformed cell lines , and furthermore Tax is dependent on IKK to stabilize MCL-1 . We next examined the stability of MCL-1 by CHX chase assays in murine embryonic fibroblasts ( MEFs ) with a genetic deletion of TRAF6 [35] . MCL-1 stability was sharply diminished in Traf6−/− MEFs compared to control wild-type MEFs ( Figure 3F ) . Although Tax stabilized endogenous MCL-1 in wild-type MEFs , Tax had no effect on MCL-1 stability in Traf6−/− MEFs ( Figure 3G ) . We next examined if Tax and TRAF6 acted synergistically to stabilize MCL-1 . Overexpression of either Tax or TRAF6 stabilized MCL-1 in 293 cells as expected , and Tax and TRAF6 together further increased MCL-1 stability ( Figure 3H ) . Taken together , our data provides strong evidence that TRAF6 plays a central role in regulating the stability of MCL-1 . We next examined if Tax was able to protect MCL-1 from degradation in response to stimuli that trigger genotoxic stress and apoptosis . Indeed , Tax prevented MCL-1 degradation induced by ultraviolet irradiation and DNA-damaging drugs including the topoisomerase II inhibitor etoposide , and the kinase inhibitor sorafenib ( Figures 4A and S8 ) . However , TaxM22 and TaxE345A mutants failed to prevent etoposide-induced degradation of MCL-1 ( Figure 4B ) . Interestingly , etoposide treatment enhanced the interactions between endogenous TRAF6 and Tax proteins in Tax inducible Jurkat Tet/On-Tax and MT-2 cells as shown by co-IP ( Figure 4C ) . The third TRAF6 binding motif in MCL-1 , that we previously demonstrated mediated the interaction with TRAF6 ( Figure 2H ) , was essential for Tax to prevent etoposide-induced MCL-1 degradation ( Figure 4D ) . Furthermore , shRNA-mediated depletion of IKKα and IKKβ or inhibition of IKKβ with SC-514 , a small molecule IKKβ inhibitor , restored the sensitivity of MT-2 cells to etoposide-induced degradation of MCL-1 ( Figure S9A–C ) . These results provide further evidence that IKK serves a critical role in the protection of MCL-1 from degradation triggered by genotoxic stress . Biochemical fractionation studies using Tax inducible Jurkat cells revealed that Tax increased the mitochondrial localization of NEMO , IKKβ and TRAF6 ( Figure S9D ) , thus raising the possibility that IKK may regulate TRAF6 and/or MCL-1 in the mitochondria . We next examined if Tax induced the mRNA expression of MCL-1 by quantitative real-time PCR ( qRT-PCR ) . MCL-1 was not transcriptionally activated by Tax , although cIAP2 and BFL-1 were strongly induced by Tax as previously described [36] , [37] ( Figure S10 ) . Tax+ HTLV-1 transformed cell lines also did not exhibit elevated levels of MCL-1 mRNA compared to Jurkat or the Tax- ATL cell line ED40515 ( - ) ( Figure S11 ) . Together these results strongly suggest that Tax regulates MCL-1 chiefly by post-translational mechanisms . Etoposide is currently a component of the regimen of conventional chemotherapy for ATL patients , nevertheless acute ATL carries a dismal prognosis due to rapid emergence of chemotherapy resistance [38] . Our data revealed that Tax+ HTLV-1 transformed cell lines ( MT-2 , MT-4 , SLB-1 and ATL-2 ( S ) ) , but not Tax- ATL cells ( ATL-43T and TL-OM1 ) , were highly refractory to etoposide-induced MCL-1 degradation ( Figure 4E ) . Thus , Tax expression may contribute to chemotherapy drug resistance in ATL . It was next examined if TRAF6 required intact RING and TRAF domains to protect MCL-1 from stress-induced degradation . Overexpression of wild-type TRAF6 , but not the RING mutant TRAF6C70A or TRAF deletion TRAF6ΔTRAF-C , stabilized and protected MCL-1 from etoposide-induced degradation ( Figure 4F ) . Thus , TRAF6 requires its E3 activity and oligomerization domain to mitigate the effects of genotoxic stress on MCL-1 stability . Given that TRAF6 functions as an essential signaling molecule downstream of Toll-like receptors ( TLRs ) and the costimulatory molecule CD40 , it is plausible that Tax may have hijacked a TRAF6-dependent signaling pathway that controls MCL-1 stability during immune activation . Thus , we examined if TRAF6 activating stimuli including CD40 ligand ( CD40L ) and lipopolysaccharide ( LPS ) influenced MCL-1 stability . Indeed , CD40L and LPS pre-treatment prevented etoposide-induced MCL-1 degradation in primary mouse B cells in the absence of transcriptional induction of MCL-1 ( Figures 4G and S12 ) . However , CD40L and LPS treatment strongly induced ICAM-1 and A20 mRNAs ( Figure S12 ) . The Epstein-Barr virus ( EBV ) -encoded latent membrane protein 1 ( LMP1 ) is a potent activator of TRAF6 and also protected MCL-1 from etoposide-induced degradation ( Figure S13 ) . Moreover , LPS induced the K63-linked polyubiquitination of MCL-1 in a TRAF6-dependent manner in RAW 264 . 7 mouse macrophages ( Figure S14 ) . Thus , TRAF6 stabilization of MCL-1 may represent a common mechanism of cell survival upon activation of innate immune signaling pathways . The E3 ligases MULE , FBW7 , and β-TRCP interact with MCL-1 and have been implicated in MCL-1 proteasomal degradation in response to apoptotic stimuli [26] , [27] , [29] . We next performed co-IP experiments to determine if Tax blocked MCL-1 degradation by inhibiting the interactions between MCL-1 and its degradative E3 ligases . Surprisingly , Tax had no effect on the interaction between MCL-1 and MULE and actually promoted the interactions of MCL-1 with FBW7 and β-TRCP ( Figure S15 ) . To further delineate the molecular mechanism underlying MCL-1 stabilization by Tax and TRAF6 , we generated a series of compound MCL-1 lysine to arginine mutants ( Figures 5A and S16A ) . A previous study suggested that MCL-1 is ubiquitinated at five N-terminal lysines ( amino acids ( aa ) 5 , 40 , 136 , 194 , and 197 ) to trigger its proteasomal degradation [26] . However , MCL-1 N5-KR , a mutant with the first five lysines substituted with arginine , was still degraded in response to etoposide treatment ( Figure S16B ) , indicating that other lysines may be involved in Ub-dependent proteasomal degradation . Indeed , mutation of the first nine lysines to arginine ( N9-KR ) abrogated etoposide-induced MCL-1 degradation ( Fig . S16B ) . However , the four C-terminal lysines ( aa 276 , 279 , 302 , and 308 ) were required for Tax and TRAF6-mediated stabilization of MCL-1 and the protection of MCL-1 from etoposide-induced degradation ( Figures 5B and C , and S16B-D ) . These four C-terminal lysines appeared to function in a redundant manner ( Figures 5C and S16D ) . Surprisingly , MCL-1 All-KR ( where all lysines were mutated to arginine ) was degraded by etoposide in a proteasome-dependent manner since MCL-1 degradation was inhibited by MG-132 treatment ( Figure 5D ) . Degradation of polyubiquitinated proteins is carried out by the 26S proteasome that includes the core 20S proteasome and a 19S regulatory subunit [39] . Ub-independent degradation of MCL-1 mediated by the 20S proteasome has been previously described [40] . We hypothesized that K63-linked polyubiquitination of MCL-1 impaired its interaction with the proteasome . To test this hypothesis , cells were treated with the DSP ( dithiobis[succinimidylpropionate] ) cross-linker and lysates were subjected to co-IP experiments to examine the effects of Tax and TRAF6 on MCL-1 interaction with the proteasome . Wild-type MCL-1 interacted with the 20S proteasome as expected , however Tax and TRAF6 effectively blocked this interaction ( Figure 5E ) . Consistently , immunofluorescence experiments showed that MCL-1 colocalized with the 20S proteasome , whereas Tax prevented this colocalization ( Figure S17 ) . However , Tax and TRAF6 did not impair interactions between the lysine-less MCL-1 mutant ( All-KR ) and the proteasome ( Figure 5E ) , indicating a requirement of MCL-1 lysines for Tax/TRAF6 to block MCL-1/20S proteasome binding . Next , ubiquitination assays were performed using wild-type and MCL-1 lysine mutants in order to map the Tax responsive K63-linked polyubiquitination sites in MCL-1 . These experiments revealed that the four C-terminal lysines served as the major targets for K63-linked polyubiquitination since MCL-1 C4-KR ( and All-KR ) was not conjugated with K63-linked polyUb chains in the presence of Tax ( Figure 5F ) . Thus , our collective data indicate that Tax/TRAF6-mediated K63-linked polyubiquitination of the four C-terminal lysines of MCL-1 blocks interactions with the core 20S proteasome ( Figure 5G ) . Given the pivotal role of TRAF6 in the control of MCL-1 stability , we hypothesized that TRAF6 plays an essential pro-survival role in ATL cells . Indeed , shRNA-mediated knockdown of TRAF6 resulted in a significant loss of viability of both Tax+ ( MT-2 ) and Tax- ( TL-OM1 ) ATL cells ( Figure 6A ) . Similar results were obtained upon MCL-1 depletion by shRNA ( Figure 6A ) . All shRNAs were validated for specific knockdown of TRAF6 and MCL-1 ( Figure S18A and B ) . Annexin V and propidium iodide ( PI ) staining confirmed that the decreased viability was due to apoptosis ( Figure 6B ) . Interestingly , MCL-1 depletion in MT-2 cells elicited less cell death compared to TRAF6 depletion ( Figure 6B ) , indicating that the Tax/TRAF6 axis may also activate alternative survival mechanisms , possibly the AKT pathway [35] . Nevertheless , MCL-1 protein expression was significantly elevated in primary human T cells immortalized by HTLV-1 ( 12 W ) using a well-established co-culture assay with irradiated MT-2 cells [41] , compared to the parental uninfected primary T cells ( 0 W ) ( Figure 6C ) . These results are congruent with previous studies that demonstrated high levels of MCL-1 protein in HTLV-1 transformed cell lines and immune stimulated Tax expressing cells [42] , [43] . To investigate the role of MCL-1 and TRAF6 in HTLV-1-mediated cell transformation , we conducted a co-culture assay using human peripheral blood mononuclear cells ( PBMCs ) transduced with lentiviruses expressing control scrambled , MCL-1 or TRAF6 shRNAs and lethally irradiated MT-2 cells as a source of infectious HTLV-1 viral particles . Puromycin was added after 4 weeks of co-culture to select for cells containing shRNAs . Cells expressing control shRNAs were immortalized by HTLV-1 and continued to expand indefinitely ( Figure 6D ) . However , cells expressing either TRAF6 or MCL-1 shRNAs ceased to proliferate after 6 weeks in culture and were resistant to immortalization ( Figures 6D and S18C ) , suggesting that TRAF6 and MCL-1 play essential roles in HTLV-1-induced immortalization of primary human T cells . To determine if TRAF6 played a more ubiquitous role in MCL-1 stabilization and survival of cancer cell lines , TRAF6 was knocked down with shRNA in HeLa ( cervical carcinoma ) , MCF-7 ( breast carcinoma ) , DLD-1 ( colorectal adenocarcinoma ) and 293 cells . TRAF6 depletion induced the apoptotic death of HeLa , MCF-7 and 293 , but not DLD-1 cells , while concomitantly downregulating MCL-1 protein ( Figure S19 ) . Thus , TRAF6 regulates the basal stability of MCL-1 in some , but not all cancer cell lines . Knockdown of TRAF6 also sensitized 293 cells to sorafenib-induced cell death ( Figure S20 ) . Taken together , TRAF6 regulates MCL-1 stability in diverse cancer cell lines .
Our findings have uncovered a novel mode of regulation of MCL-1 stability that has been hijacked by the HTLV-1 Tax oncoprotein to promote cell transformation . A ubiquitin proteomics screen revealed that Tax modulated the ubiquitination of 136 cellular proteins , of which 22 of these candidates required IKK for Tax-induced ubiquitination . Tax promoted the K63-linked polyubiquitination of MCL-1 in a TRAF6-dependent manner , which imparted enhanced stability to MCL-1 and protection from degradation in response to genotoxic stress stimuli . Our collective results provide strong evidence that Tax has usurped TRAF6 and the host ubiquitin machinery to evade apoptosis and maintain viral persistence . Although MCL-1 is a highly labile protein , it is commonly overexpressed in cancers and contributes to cell survival and drug resistance although the precise mechanisms have yet to be completely elucidated . Previous studies have shown that the E3 ligases MULE , β-TRCP and FBW7 conjugate MCL-1 with K48-linked polyUb chains to promote its degradation [26]–[29] . The deubiquitinase USP9-X is overexpressed in lymphomas and multiple myeloma and stabilizes MCL-1 by removing K48-linked polyUb chains [44] . Our findings implicate TRAF6 as a key regulator of MCL-1 stability . MCL-1 contains three TRAF6 binding sites in the PEST domain , of which the third one appears to be most critical for binding with TRAF6 and stabilization by Tax . We found that TRAF6 expression was sufficient to protect MCL-1 from etoposide-induced degradation , and TRAF6 required its E3 ligase activity and C-terminal TRAF domain for this function . Intriguingly , treatment of primary mouse B cells with CD40L or LPS , both potent activators of endogenous TRAF6 , protected MCL-1 from etoposide-induced degradation . We also found that TRAF6 stabilized MCL-1 in mouse fibroblasts and in several cancer cell lines including HeLa and MCF-7 . Therefore , TRAF6-mediated MCL-1 stabilization appears to be a common mechanism of cell survival usurped by both viral and non-viral cancers . Tax contains a TRAF6 binding motif between amino acids 343–348 , just upstream of the PBM known to be critical for transformation by Tax [32] . Mutation of this TRAF6 binding site greatly diminished Tax interaction with TRAF6 , Tax-induced autoubiquitination of TRAF6 and MCL-1 K63-linked polyubiquitination and stabilization . Tax interaction with TRAF6 was also critical for the redistribution of TRAF6 to the mitochondria since TaxE345A was defective in promoting TRAF6 mitochondrial localization . Thus , Tax directly engages TRAF6 to trigger its activation , and a subset of Tax translocates to the mitochondria together with TRAF6 which directly ubiquitinates MCL-1 and possibly other substrates . It is intriguing that Tax can localize to the mitochondria , however future studies will be necessary to determine the precise mechanisms by which Tax traffics to the mitochondria . Notably , the Tax2 protein encoded by HTLV-2 has divergent C-terminal sequences from Tax1 , lacks the TRAF6 binding site and PBM and is unable to interact with TRAF6 [45] . These differences in Tax2 may potentially account for the reduced pathogenicity of HTLV-2 , which has not been definitively linked to any lymphoproliferative disorders [46] . Previous studies have shown that Tax enhances the autoubiquitination of TRAF6 [30] , consistent with the findings in this report . Given that TRAF6 is a potent activator of NF-κB , does TRAF6 play a role in Tax-induced NF-κB activation ? A previous study demonstrated that Tax activation of IKK proceeds normally in the absence of TRAF6 in vitro using a cell free assay system [47] . However , our findings in this study raise new questions regarding TRAF6 and Tax-induced NF-κB activation that clearly warrant additional studies performed in T cells , the natural cell host of HTLV-1 . Indeed , we have recently found that IL-17RB and TRAF6 both play essential roles in NF-κB signaling in HTLV-1 transformed cell lines [48] . However , it is unclear if Tax requires TRAF6 and the TRAF6 binding site in Tax for NF-κB activation . It also remains unclear precisely how IKK participates in Tax-induced TRAF6 activation and MCL-1 stabilization . Since Tax forms protein complexes with IKK/NEMO and TRAF6 , Tax may utilize NEMO as part of its mechanism to trigger TRAF6 autoubiquitination . Given that Tax also induces the mitochondrial localization of IKK , another possibility is that IKK may directly phosphorylate and regulate either TRAF6 and/or MCL-1 in the mitochondria . Our results indicated that Tax did not impede the interactions of MCL-1 with its degradative E3 ligases , but rather Tax enhanced binding of MCL-1 with FBW7 and β-TRCP . Consistently , Tax also modestly increased the K48-linked polyubiquitination of MCL-1 . These findings can potentially be explained by a recent study that demonstrated that Tax induces reactive oxygen species , which in turn stimulate DNA damage and genotoxic stress [49] . Tax may counteract the destabilizing effects of genotoxic stress on MCL-1 by triggering the activation and mitochondrial localization of TRAF6 , which interacts with MCL-1 and conjugates the four C-terminal lysine residues with K63-linked polyUb chains . MCL-1 K63-linked polyubiquitination stabilizes MCL-1 by blocking its interactions with the core 20S proteasome , thereby preventing both Ub-dependent and independent degradation of MCL-1 . K63-linked polyUb chains conjugated to the C-terminus of MCL-1 may prevent 20S proteasome binding via a conformational change in MCL-1 or through steric hindrance . Previously it was reported that pro-apoptotic BH3-only proteins ( BOPs ) , PUMA and BIM , stabilize yet inactivate MCL-1 [50]–[52] . This paradoxical finding indicates that functional inactivation of MCL-1 does not always require its degradation . Although the molecular mechanism remains to be elucidated , PUMA and BIM may stabilize MCL-1 by binding to the hydrophobic groove of MCL-1 to inhibit MULE interactions [51] . However , our results indicate that Tax did not prevent MCL-1 and MULE interactions , and also TRAF6 did not bind to the hydrophobic groove of MCL-1 . Therefore , it is likely that Tax and TRAF6 stabilize and enhance the anti-apoptotic activity of MCL-1 , in contrast to PUMA and BIM . This notion is further supported by our results that demonstrate that shRNA-mediated knockdown of MCL-1 in HTLV-1-transformed and ATL cells induced apoptotic cell death and also blocked HTLV-1-induced immortalization of primary T cells . TRAF6 has recently emerged as an oncogene and is overexpressed in diverse human cancers [53] , [54] . TRAF6 conjugates AKT with K63-linked polyUb chains that regulate its membrane localization and phosphorylation [35] . TRAF6 also upregulates the expression of hypoxia-inducible factor 1α ( HIF-1α ) to promote tumor angiogenesis [55] . Together with our findings that TRAF6 governs MCL-1 stability and cell survival , accumulating evidence strongly support the notion that TRAF6 is a bona fide oncogene and when overexpressed can endow cells with at least three of the known hallmarks of cancer ( sustaining proliferative signaling , resisting cell death and inducing angiogenesis ) [56] . In conclusion , our studies have identified a novel TRAF6/MCL-1 anti-apoptotic axis that has been subverted by the HTLV-1 Tax oncoprotein to evade apoptosis . TRAF6 and MCL-1 may therefore represent viable therapeutic targets for ATL and other cancers .
Blood from healthy donors was purchased from Biological Specialty Corporation ( Colmar , PA ) . Animal work was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . The animal protocol ( protocol number MO12M112 ) was approved by the Institutional Animal Care and Use Committee ( IACUC ) of Johns Hopkins University . pCMV-Flag-MCL-1 ( plasmid 25392 ) , pRK5-HA-Ub ( plasmid 17608 ) , pRK5-HA-Ub-K63 ( plasmid 17606 ) and pRK5-HA-Ub-K48 ( plasmid 17605 ) were purchased from Addgene . His-MCL-1 was generated by cloning human MCL-1 cDNA into pcDNA3 . 1/His using BamHI and EcoRI enzyme sites . pCMV4-Tax and M22 were previously described [20] . Flag-Tax ( wild-type and mutants ) was generated by cloning Tax cDNAs into pcDNA3 . 1/Flag vector using BamHI and EcoRI enzyme sites . The plasmids expressing HA-TRAF3 , Flag-TRAF2 and Flag-TRAF6 were previously described [57]–[59] . Flag-TRAF5 [60] was a gift from Drs . Soo Young Lee and Yongwon Choi ( University of Pennsylvania ) . Flag-TRAF3 was generated by cloning TRAF3 cDNA into pDUET . Hyg vector using BamHI and XhoI enzyme sites . Myc/His-TRAF6 was generated by cloning TRAF6 cDNA into pcDNA3 . 1-Myc/His vector using EcoRI and XhoI enzyme sites . The GFP reporter vector pEGFP N1 was purchased from Clontech . GFP-Tax [61] was a gift from Dr . Brian Wigdahl ( Drexel University ) . Flag-EBV LMP1 plasmid [62] was a gift from Dr . Shunbin Ning ( University of Miami ) . Flag-MULE [26] was a gift from Dr . Qing Zhong ( University of California , Berkeley ) . HA-FBW7 plasmid [28] was a gift from Dr . Wenyi Wei ( Harvard University ) . HA-βTRCP [63] was a gift from Dr . Shao-Cong Sun ( M . D . Anderson Cancer Center ) . Tax , IKKα , and IKKβ shRNAs were cloned into lentiviral pYNC352/puro or GFP-puro vector using BamHI and MluI enzyme sites as described previously [64] . The oligonucleotide sequences for shRNAs are listed in Table S2 . The lentiviral pLKO . puro vector expressing control , TRAF6 and MCL-1 shRNAs were purchased from Sigma-Aldrich . For lentiviral transduction of Tax , Tax cDNA was cloned into pDUET . hyg using BamHI and XhoI enzyme sites . GST-MCL-1 was generated by cloning MCL-1 cDNA lacking the C-terminal transmembrane domain ( amino acids 328 to 350 ) in frame with GST into pGEX4T-1 using BamHI and XhoI enzyme sites . Site-directed mutagenesis was performed using Platinum Pfx DNA polymerase ( Invitrogen ) . Anti-MCL-1 ( sc-819 ) , TRAF6 ( sc-8409 ) , TOM20 ( sc-17764 ) , GFP ( sc-8334 ) , GST ( sc-138 ) , His ( sc-804 ) , VDAC1 ( sc-390996 ) , IKKα ( sc-7182 ) , IKKβ ( sc-8014 ) , NEMO ( sc-8330 ) and LDH ( sc-33781 ) antibodies were purchased from Santa Cruz Biotechnology ( Santa Cruz , CA ) . Anti-mouse MCL-1 antibody ( 600-401-394 ) was purchased from Rockland Immunochemicals ( Gilbertsville , PA ) . Anti-TRAF6 ( 8028 ) , Flag ( 2368 ) , PARP ( 9532 ) , and Lys63 specific ubiquitin ( 5621 ) antibodies were purchased from Cell Signaling Technology ( Danvers , MA ) . Anti-ubiquitin ( linkage-specific K63 ) antibody ( ab179434 ) was from Abcam ( Cambridge , MA ) . Anti-Flag ( F1804 ) and β-actin ( A1978 ) antibodies were purchased from Sigma-Aldrich ( St . Louis , MO ) . Anti-Myc ( OP-10 ) and Lys63 specific ubiquitin ( 05-1313 ) monoclonal antibodies were purchased from EMD Millipore ( Billerica , MA ) . Anti-ubiquitin ( ADI-SPA-200 ) and 20S proteasome α4 antibodies ( MCP34 ) were purchased from ENZO Life Sciences ( Farmingdale , NY ) . Anti-HA ( clone 12CA5 ) antibody was purchased from Roche Applied Science ( Indianapolis , IN ) . Monoclonal anti-Tax antibody was prepared from a hybridoma ( 168B17-46-34 ) received from the AIDS Research and Reference Program , NIAID , National Institutes of Health . Jurkat Tet/On-Tax WT and M22 cells [65] were maintained in RPMI 1640 supplemented with 10% tetracycline-free FBS and treated with 1 µg/ml doxycycline for 2 days for inducible Tax expression . Jurkat , Jurkat ( NEMO-deficient ) [66] , DLD-1 , and HTLV-1 transformed and ATL cell lines including MT-2 , MT-4 , SLB-1 , ATL-2 ( S ) , C8166 , TL-OM1 , ED40515 ( - ) and ATL-43T were cultured in RPMI-1640 supplemented with 10% heat-inactivated FBS and antibiotics . HeLa , 293T , 293 , MCF-7 and Raw 264 . 7 cell lines were cultured in DMEM supplemented with 10% FBS and antibiotics . Traf6+/+ and Traf6−/− MEFs , a gift from Dr . Hui-Kuan Lin ( M . D . Anderson Cancer Center ) , were cultured in DMEM supplemented with 10% FBS and antibiotics and transfected with GenJet II ( SignaGen Laboratories , Rockville , MD ) . Primary splenic B cells were isolated using the EasySep mouse B cell enrichment kit ( Stemcell Technologies , Vancouver , Canada ) from mice on a mixed 129×B6 genetic background and cultured in RPMI-1640 supplemented with 10% heat-inactivated FBS . Plasmid DNA transfection of HeLa , 293 and 293T cells was performed using JetPrime ( Polyplus-Transfection , New York , NY ) . Lentiviral transduction of DNA or short hairpin RNAs ( shRNAs ) into suspension cells was performed by spinoculation at 800×g for 30 min with MOIs of 5 to 10 . The lentiviral transduction of adherent cells was conducted in the presence of 5 µg/ml polybrene . For immunoblotting , whole cell lysates were prepared by lysing cells in cell extraction buffer ( 100 mM Tris [pH 7 . 4] , 100 mM NaCl , 1% Triton X-100 , 1 mM EDTA , 1 mM EGTA , 10% glycerol , 0 . 1% SDS , 2 mM Na2VO4 , 1 mM NaF , 0 . 5% deoxycholate , and 20 mM Na4P2O7 ) supplemented with protease inhibitor cocktail ( Roche Applied Bioscience , Indianapolis , IN ) on ice , followed by centrifugation at 15 , 000×g for 10 min . Cell lysates were separated on SDS-PAGE , transferred to nitrocellulose membranes and immunoblotted with appropriate antibodies diluted in SuperBlock blocking PBS buffer ( Thermo Scientific , Rockford , IL ) . For co-IP , cells were lysed in RIPA buffer ( 50 mM Tris [pH 7 . 4] , 150 mM NaCl , 1% Igepal CA-630 , and 0 . 25% deoxycholate ) freshly supplemented with protease inhibitor cocktail on ice . Lysates ( 500 µg protein ) cleared by centrifugation were incubated at 4°C overnight with the indicated antibodies ( 2 µg ) and then incubated with protein A-agarose beads for an additional 3 h . Immunoprecipitates were washed three times with RIPA buffer followed by elution of bound proteins with 1 . 5× SDS sample buffer or 3× Flag peptide ( Sigma , St . Louis , MO ) . For immunostaining , HeLa cells grown on glass coverslips for 24 h were transfected , and fixed and permeabilized in chilled methanol for 5 min . For staining of mitochondria , cells were incubated with 100 nM Red MitoTracker ( Invitrogen , Grand Island , NY ) for 30 min before fixation . Following incubation with SuperBlock blocking PBS buffer overnight at 4°C , coverslips were incubated with primary antibodies , washed with PBS , and then incubated with appropriate fluorescence dye-conjugated secondary antibodies . The coverslips were incubated with DAPI for 3 min and mounted in mounting medium ( Richard-Allan Scientific , Campus Drive Kalamazoo , MI ) , and cells were imaged on Nikon E-800 with a 60× oil-corrected objective . The UbiScan proteomics platform ( performed by Cell Signaling Technology ) was used to identify and quantify differences in ubiquitination between untreated Jurkat Tet/On-Tax WT and M22 cells and treated with Dox . Briefly , cell lysates were digested with trypsin and peptides were separated from non-peptide material by solid phase extraction with Sep-Pak C18 cartridges . Lyophilized peptides were re-dissolved , and ubiquitinated peptides enriched with the ubiquitin branch ( “K-ε-GG” ) antibody ( Cell Signaling Technology ) . Peptides were eluted from antibody resin into a total volume of 100 µl in 0 . 15% TFA . Eluted peptides were concentrated with Eppendorf PerfectPure C18 tips prior to LC-MS/MS analysis with an LTQ-Orbitrap hybrid mass spectrometer . MS/MS spectra were evaluated using SEQUEST and SORCERER 2 . Peptide assignments were obtained using a 5% false positive discovery rate . Searches were performed against the NCBI human database . For ubiquitination assays , an extra wash was performed using RIPA buffer supplemented with 1 M urea after immunoprecipitation . For MCL-1 ubiquitination assays performed with His-MCL-1 , cells were lysed in buffer B ( 100 mM NaH2PO4 , 10 mM Tris , and 8 M urea [pH 8 . 0] ) and His-tagged MCL-1 proteins were precipitated with Ni-nitrilotriacetic acid ( NTA ) agarose ( Qiagen , Valencia , CA ) . After washing in buffer C ( 100 mM NaH2PO4 , 10 mM Tris , and 8 M urea [pH 6 . 3] ) , His-tagged proteins were eluted in buffer E ( 100 mM NaH2PO4 , 10 mM Tris , and 8 M urea [pH 4 . 5] ) and subjected to SDS-PAGE and immunoblotting . For TRAF6 in vitro ubiquitination assays , Flag-tagged TRAF6 or Tax proteins expressed in 293T cells were purified using EZview Red anti-Flag affinity gel ( Sigma-Aldrich , St . Louis , MO ) and incubated in ubiquitin conjugation reaction buffer supplemented with UBE1 ( E-305 ) , UbcH13 ( E2-664 ) and ubiquitin ( U-100H ) purchased from Boston Biochem ( Cambridge , MA ) for 2 h at 30°C . For MCL-1 in vitro ubiquitination assays , Flag-TRAF6 ( alone or with Tax ) was purified and incubated in ubiquitin conjugation reaction buffer supplemented with UBE1 , UbcH5c ( E2-627 ) , GST or GST-MCL-1 , ubiquitin and energy regeneration buffer ( Boston Biochem ) for 2 h at 30°C . The reaction mixtures were boiled in 1× SDS sample buffer and subjected to SDS-PAGE and immunoblotting . Recombinant GST-fusion proteins were purified using standard methods . To enhance the solubility of GST-MCL-1 proteins from bacteria , the C-terminal transmembrane domain of MCL-1 was deleted . GST-MCL-1 protein ( 2 µg ) immobilized on glutathione beads was incubated with Flag-TRAF6 protein ( 500 ng ) immunoprecipitated from 293T cells in binding buffer ( 50 mM Tris [pH 7 . 4] , 150 mM NaCl , 1 mM EDTA , and 1% Triton X-100 ) for 3 h at 4°C and washed with binding buffer four times , and precipitated protein complexes were separated on SDS-PAGE and subjected to immunoblotting . Cells were treated with CHX ( 10 µg/ml ) for the indicated times . Whole cell lysates were prepared in cell extraction buffer , separated on SDS-PAGE , and immunoblotted with anti-MCL-1 and actin antibodies . The intensity of MCL-1 bands was measured using Alphaview software ( Alpha Innotech , San Leandro , CA ) and normalized to actin . The half-life of MCL-1 protein was calculated under the assumption of first-order decay kinetics as described previously [67] . Mitochondria were isolated using Axis-Shield OptiPrep iodixanol ( Sigma-Aldrich , St . Louis , MO ) according to the manufacturer's protocol . Briefly , cells transfected with the indicated plasmid DNAs were homogenized in buffer B ( 0 . 25 M sucrose , 1 mM EDTA , 20 mM HEPES-NaOH [pH 7 . 4] ) with 40 strokes of a Dounce homogenizer and centrifuged at 1 , 000 g for 10 min . An aliquot of homogenate was used as total cell extract . The supernatant was further centrifuged at 13 , 000 g at 4°C for 10 min . The supernatant was used as cytosol fraction . The pellet resuspended in 36% iodixanol was bottom-loaded under 10% and 30% gradients and centrifuged at 50 , 000 g for 4 h . The mitochondrial fraction was collected at the 10%/30% iodixanol interface . Total RNA was isolated using the RNeasy mini kit ( Qiagen , Valencia , CA ) . First-strand cDNA was synthesized from 1 µg of total RNA using SuperScript II RT ( Invitrogen ) with random hexamers . qRT-PCR was performed in a 96-well microplate using an ABI Prism 7500 detection system ( Applied Biosystems , Foster City , CA ) with the RT2Real-Time SYBR green/ROX master mix ( Qiagen , Valencia , CA ) . Reactions were performed in a total volume of 25 µl and contained 50 ng of reverse-transcribed RNA ( based on the initial RNA concentration ) and gene-specific primers . PCR conditions included an initial incubation step of 2 min at 50°C and an enzyme heat activation step of 10 min at 95°C , followed by 40 cycles of 15 seconds at 95°C for denaturing and 1 min at 60°C for annealing and extension . Primer sequences are listed in Table S3 . Peripheral blood mononuclear cells ( PBMCs ) were freshly isolated from blood using Ficoll-Paque Plus ( GE healthcare , Piscataway , NJ ) and stimulated with 5 µg/ml of phytohemagglutinin ( PHA ) in R10 media ( RPMI-1640 supplemented with 25 mM HEPES , 2 mM L-glutamine , and antibiotics ) for 3 days . Following transduction ( MOI of 10 ) with lentiviruses expressing shRNAs by spinoculation at 800×g for 3 . 5 h , the PBMCs ( 1×104 cells ) were co-cultured with MT-2 cells ( 5×104 cells ) that were γ-irradiated at 60 Gy . Puromycin ( 5 µg/ml ) was added to the culture at 4 week after co-culture to select for shRNA-transduced PBMCs . Viable cells were counted using the trypan blue exclusion assay at 4 and 6 weeks . To measure apoptosis , a modified Annexin V/PI staining procedure was performed as described previously [68] . Briefly , cells were washed in phosphate buffered saline ( PBS ) and 1× Annexin V binding ( AV ) buffer ( 10 mM HEPES [pH 7 . 4] , 140 mM NaCl , and 2 . 5 mM CaCl2 ) twice , and resuspended in 100 µl of AV buffer . The cells were incubated with Annexin V Alexa Fluor 488 ( Invitrogen ) in the dark for 15 min . at room temperature and then PI ( Sigma-Aldrich ) was added for an additional 15 min , fixed with 1% formaldehyde and treated with RNase A ( 50 µg/ml ) for 15 min at 37°C . Samples were analyzed using a FACSCalibur flow cytometer ( BD Biosciences ) . CellTiter-Glo Luminescent Cell Viability Assay ( Promega , Madison , WI ) which quantitates ATP as a measure of metabolically active cells was used to measure cell viability .
|
HTLV-1 infection is etiologically linked to the development of the neuroinflammatory disorder HTLV-1 associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) and adult T-cell leukemia ( ATL ) , an aggressive CD4+CD25+ malignancy . The HTLV-1 regulatory protein Tax constitutively activates the IκB kinases ( IKKs ) and NF-κB to promote cell survival , proliferation and transformation . However , the precise mechanisms by which Tax and IKK regulate cell survival are largely unknown . Here , we found that Tax interacts with and activates the host ubiquitin ligase TRAF6 , and promotes a redistribution of TRAF6 to the mitochondria . TRAF6 conjugates the anti-apoptotic BCL-2 family member MCL-1 with lysine 63 ( K63 ) -linked polyubiquitin chains that antagonize MCL-1 interaction with the 20S proteasome , thereby protecting MCL-1 from degradation elicited by chemotherapeutic drugs . TRAF6 and MCL-1 both played pivotal roles in the survival of ATL cells and the immortalization of primary T cells by HTLV-1 . Overall , our study has identified a novel TRAF6/MCL-1 axis that has been subverted by the HTLV-1 Tax protein to maintain the survival of HTLV-1 infected T cells .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"biology",
"and",
"life",
"sciences",
"medicine",
"and",
"health",
"sciences"
] |
2014
|
HTLV-1 Tax Stabilizes MCL-1 via TRAF6-Dependent K63-Linked Polyubiquitination to Promote Cell Survival and Transformation
|
The innate immune response constitutes the first line of defense against infections . Pattern recognition receptors recognize pathogen structures and trigger intracellular signaling pathways leading to cytokine and chemokine expression . Reactive oxygen species ( ROS ) are emerging as an important regulator of some of these pathways . ROS directly interact with signaling components or induce other post-translational modifications such as S-glutathionylation , thereby altering target function . Applying live microscopy , we have demonstrated that herpes simplex virus ( HSV ) infection induces early production of ROS that are required for the activation of NF-κB and IRF-3 pathways and the production of type I IFNs and ISGs . All the known receptors involved in the recognition of HSV were shown to be dependent on the cellular redox levels for successful signaling . In addition , we provide biochemical evidence suggesting S-glutathionylation of TRAF family proteins to be important . In particular , by performing mutational studies we show that S-glutathionylation of a conserved cysteine residue of TRAF3 and TRAF6 is important for ROS-dependent activation of innate immune pathways . In conclusion , these findings demonstrate that ROS are essential for effective activation of signaling pathways leading to a successful innate immune response against HSV infection .
The innate immune response constitutes the first line of defense against invading pathogens , and relies on pattern recognition receptors ( PRR ) s for detection of infections through recognition of either molecular structures specific for non-self , or aberrant localization of molecules used by both host and microbe [1] , [2] . Toll-like receptors ( TLR ) s are membrane-bound PRRs localized in the plasma membrane and endosomes , which recognize microbes at these sites . Other families of PRRs are localized in the cytoplasm , such as Retinoic acid-inducible gene ( RIG ) -I-like receptors ( RLR ) s which detect cytosolic RNA [1] , [2] . The RLRs , including RIG-I and Melanoma differentiation-associated gene-5 ( Mda-5 ) , signal from the outer mitochondrial membrane via anchoring of the mitochondrial RLR adaptor molecule , MAVS ( mitochondrial antiviral signaling protein ) [3] . Cytosolic DNA potently stimulates innate immunity through a series of DNA receptors ( DNAR ) s including the AIM2-like receptors ( ALR ) s [3]–[7] . Herpes simplex virus ( HSV ) is a significant human pathogen , and whilst HSV-1 is an important cause of viral encephalitis , HSV-2 predominantly causes genital infections [8] . HSV-1 and 2 are closely related at the genetic level and accordingly share many biological and pathological properties [8] . HSV is recognized by TLR2 and TLR9 [9] , [10] , which act synergistically to control HSV infection in the brain in mice [11] . In humans , a dominant-negative TLR3 allele has been reported in otherwise healthy children with HSV-1 encephalitis [12] . In addition to TLRs , both RLRs and DNARs have been implicated in recognition of herpes viruses [3] , [5]–[7] , [13] . Most recently , we have identified γ-interferon-inducible protein 16 ( IFI16 ) as a novel intracellular sensor of HSV DNA and mediator of expression of type I interferon ( IFN ) and inflammatory cytokines [7] . Thus , innate recognition of HSV involves a large spectrum of PRRs , which together orchestrate the innate immune response to infections by this virus [reviewed in ref . 14] . The innate immune system interacts closely with basic cellular processes such as autophagy and reactive oxygen species ( ROS ) [15] , [16] , thereby coordinating these cellular processes with the innate immune response . Oxygen is consumed during many cellular processes in a manner leading to production of superoxide anions ( O2- ) [17]–[19] , which are readily converted to hydrogen peroxide ( H2O2 ) by a reaction catalysed by superoxide dismutases . Both superoxide anions and hydrogen peroxide can be converted to other reactive species ( e . g . , hydroxyl radical , peroxynitrite , etc . ) through reactions with a range of reactive molecules in the cell . ROS is a collective term for superoxide and the downstream-derived ions and small molecule metabolites . Due to the highly reactive nature of ROS that can modify any cellular macromolecule , the cell contains several scavengers , called antioxidants , as well as enzymes that maintain cellular redox homeostasis . Glutathione ( GSH ) is the most abundant cellular antioxidant . GSH can form molecular disulfides with the thiol group of the cysteine residues in oxidative conditions , thereby regulating protein function [20] . This modification has been shown to be essential for the function of several proteins such as c-jun , p50/NF-κB and IκB kinase subunit ( IKK ) β [21]–[23] . It has long been known that ROS play an important part in the innate immune system as a microbicidal compound produced by NADPH oxidases ( NOXs ) in phagosomes of phagocytic cell types such as macrophages and neutrophils [18] . More recently , it has been demonstrated that ROS are also involved in activation and regulation of a wider range of processes in the innate immune system , including autophagy , signal transduction , gene expression , activation of the inflammasome , and programmed necrosis [15] , [16] , [24]–[28] . Thus , cellular sensing of the intracellular redox status impacts on the immune response to infectious agents . The recognition of viruses by the immune cells is a very complex process in which several signal transduction pathways regulating numerous cellular processes are involved . Moreover , the recent observation of the involvement of ROS as a regulator of these processes gives an additional layer of complexity that needs to be described . In this work , we demonstrate that HSV infection induces an early production of ROS , which regulates the activation of innate immune responses initiated by HSV infection , via TLRs , RLRs , and IFI16 . Mechanistically , we show that ROS induces S-glutathionylation of tumor necrosis factor receptor-associated factor ( TRAF ) 3 and 6 , which is essential for optimal stimulation of the cellular response to innate HSV recognition .
ROS has gained increasing importance in the role as a second messenger in the activation of signaling pathways . In order to examine how these radicals are involved in the recognition of HSV , we first examined if HSV infection induced formation of cellular ROS . RAW264 . 7 cells were infected with HSV-2 and ROS formation at different time points was monitored by live microscopy using the oxidant-sensitive fluorescent probe CM-DCFDA . ROS production was induced 1 h post-infection ( Figure 1A ) , and this was abrogated by pre-treating the cells with the antioxidant N-acetyl-L-cysteine ( LNAC ) , which can scavenge endogenous ROS ( Figure 1A ) . LNAC did not affect viral entry as determined by staining for viral capsids in the cytoplasm following infection ( data not shown ) . Next , the role of ROS in HSV-stimulated cytokine expression was examined . We first investigated the effect of exogenous ROS on expression of IFNs and IFN-stimulated genes ( ISG ) s in response to HSV infection . Murine primary macrophages responded to HSV-2 infection with production of type I IFN ( Figure 1B ) and the IFN-inducible chemokines CXCL10 ( Figure 1C ) and CCL5 ( Figure 1D ) , and in all cases exogenous hydrogen peroxide ( H2O2 ) lead to a modest but significant elevation of this response . On the contrary , pre-treatment with the general antioxidants pyrrolidine dithiocarbamate ( PDTC ) and LNAC strongly inhibited the cytokine response following infection with HSV-2 ( Figure 1E , F ) . All small molecule inhibitors were used in concentrations not affecting cell viability as determined by annexin V and propidium iodide staining in macrophages ( data not shown ) . To examine how modulation of ROS levels affected the ability of HSV to activate the innate immune response in different immune cell types , we treated pDCs and macrophages with LNAC prior to infection with HSV-1 or -2 . The subsequent CCL5 expression was evaluated by ELISA . Although these cell types stimulate the innate immune response against HSV infection through different combinations of PRRs , they all displayed the same inhibitory effect of the antioxidant ( Figure 2A , B ) . Similar findings were obtained with murine embryonic fibroblasts ( MEF ) s treated with LNAC and infected with HSV-1 or -2 ( Figure 2C ) . Once the role of ROS was confirmed in vitro , we evaluated their potential function in innate antiviral defense in vivo . With this purpose , we treated mice with LNAC prior to infection with HSV-2 . The virus-induced production of type I IFN was inhibited in mice receiving LNAC ( Figure 2D ) , and these mice exhibited elevated viral load in the liver after day 1 post infection ( Figure 2E ) . Collectively , these data demonstrate that HSV infection induces formation of cellular ROS , which are essential for the activation of an antiviral innate immune response in vitro and in vivo . The signaling pathways upstream of the transcription factors NF-κB and IRF-3 are important for establishment of innate antiviral defense against HSV [1] , [2] . To clarify whether activation of these pathways was modulated by ROS , we examined how treatment with LNAC prior to HSV infection affected phosphorylation of the NF-κB inhibitor IκBα and activation of IRF-3 by nuclear translocation . We observed that the infection led to strong phosphorylation of IκBα after 5 h , and that this was potently inhibited by LNAC ( Figure 2F ) . Likewise , HSV-induced translocation of IRF-3 to the nucleus was largely abrogated by pre-treatment with LNAC ( Figure 2G ) . Thus , the signaling pathways activating NF-κB and IRF-3 are sensitive to the general depletion of cellular ROS . The MAPK kinase kinase apoptosis signal-regulating kinase ( ASK ) 1 has been reported to be involved in ROS-dependent innate immune signaling pathways upstream of MAPKs and IRF-3 following lipopolysaccharide ( LPS ) treatment [16] , [26] . To test the role of ASK1 in ROS-dependent innate antiviral immune responses , peritoneal macrophages from C57BL/6 and ASK1-/- mice were infected with HSV-2 in vitro . As expected , the culture supernatants from infected cells contained elevated levels of CCL5 and IFN-α/β as compared to the supernatants from untreated cells , but no significant difference was observed between the C57BL/6 and ASK1-/- cells ( Figures 2H , I ) . Mice were infected with HSV-2 via the intraperitoneal route , and livers harvested after 3 days for analysis of viral load by plaque assay . High levels of virus were observed in the livers from C57BL/6 mice , and no impairment in the antiviral defense was observed in the ASK1 -/- mice ( Figure 2J ) . Thus , HSV infection stimulates ROS-dependent activation of NF-κB and IRF-3 in a manner independent of ASK1 . HSV stimulates the innate immune system through multiple PRRs and in cell-type specific manners [13] . In macrophages , HSV-1 induced expression of type I IFN , independently of MAVS ( RLRs ) ( Figure 3A ) and TLR2/9 ( Figure 3B ) . We have recently shown that HSV-1 is sensed by IFI16 [7]; consistent with this , expression of CCL5 in macrophages in response to HSV-1 was dependent on the IFI16 murine ortholog , p204 ( Figure 3D ) . In contrast , pDCs responded to the infection in a manner entirely dependent on TLR9 ( Figure 3C ) . To further investigate the effect of ROS modulation on the activation of the PRRs that were reported to be involved in HSV recognition , we treated RAW264 . 7 cells with LNAC prior to stimulation with specific ligands for: cytoplasmic DNA sensors ( HSV-1 60-mer ) , TLR2 ( Pam3Csk4 ) , TLR3 ( poly ( I∶C ) ) , TLR9 ( ODN1826 ) , and RLRs ( poly ( I∶C ) :LyoVec ) ) ( Figure 3E–G ) . Pre-incubation with LNAC strongly diminishes CCL5 production in response to HSV-1 60mer ( Figure 3E ) . Similarly , stimulation via TLRs and RLRs induced a strong expression of CCL5 , which was abrogated by pre-treatment with LNAC ( Figure 3F , G ) . Collectively , these data demonstrate that all the PRRs reported to be involved in recognition of HSV are inhibited by general depletion of cellular ROS , indicating a positive role for ROS in the regulation of the innate immune response to HSV . ROS can influence signal transduction through many mechanisms [16] , [29] , [30] . However , one of the most important mechanisms involves reversible S-glutathionylation of cysteine residues that alters protein function either by changing the protein active site or modifying protein-protein interactions [20] . To address whether S-glutathionylation could be involved in the positive stimulatory roles of ROS on innate immune signaling , we treated cells with the GSH-depleting agents buthionine sulfoximine and diethylmaleate prior to infection with HSV-2 [31] . Diminution of cellular GSH levels decreased the ability of HSV-2 to induce IFN-β and CCL5 mRNA expression ( Figure 4A , B ) . To further address the potential role of S-glutathionylation in ROS-dependent positive stimulation of innate immune signaling , we transduced RAW264 . 7 cells with recombinant adenovirus ( AdV ) expressing glutaredoxin ( Adv-Grx ) 1 , in order to decrease S-glutathionylation , or with an empty vector adenovirus ( Adv-vector ) . Glutaredoxin overexpression decreases S-glutathionylation of proteins by catalyzing deglutathionylation freeing the protein thiol groups . Infection of the cells with HSV-2 or stimulation with the TLR9 ligand ODN1826 led to accumulation of CCL5 and IFN-α/β in the culture supernatants of the cells expressing AdV-vector and this response was significantly decreased in cells expressing Adv-Grx ( Figure 4C , D ) . TRAF family proteins have been reported to be sensitive to redox-dependent regulation of signal transduction [16] , [32] , and we therefore examined whether ROS affected the glutathionylation of TRAF3 and TRAF6 , which are critically involved in activation of pathways upstream of IRF-3 and NF-κB , respectively . Interestingly , H2O2 treatment , which elevated HSV-induced cytokine expression ( Figure 1B–D ) , led to a clear increase in GSH moieties associated with both TRAF3 and TRAF6 ( Figure 4E , F ) . Importantly , glutathionylation of TRAF 3 and 6 was also observed following HSV-2 infection ( Figure 4G , H ) . Structure-based sequence alignment of TRAF family proteins revealed a cysteine residue in the β3 sheet in the TRAF C-terminal domain , which is conserved between TRAF2 , 3 , and 6 , in both rodents and humans ( Figure 5A ) . Examination of the structure of the TRAF6 C-terminal domain complexed with a peptide from RANK ( PDB ID 1LB5 ) showed that this residue ( Cys390 ) is surface-exposed and localized in close proximity to the peptide-binding pocket [33] . We docked GSH onto the structure of the TRAF6 C-terminal domain in the proximity of Cys390 . The docking revealed that the structure was compatible with glutathionylation of Cys390 , as the only clash was with Arg466 ( data not shown ) . However , this clash could be relieved by changing the side chain conformation of Arg466 , directing it further towards the RANK binding pocket . Thus , glutathionylation of Cys390 and contact between the GSH group and the binding partners of TRAF6 is in accordance with the current structural knowledge . To determine the role of this potential glutathionylation site in TRAF6 mediated signaling , wild type TRAF6 or TRAF6 C390S mutant were transfected into Traf6-/- fibroblasts and P-IκBα monitored in response to treatment with HSV-2 or poly ( I∶C ) . Traf6-/- fibroblasts displayed reduced P-IκBα compared to wild type fibroblasts in response to both HSV-2 infection and treatment with poly ( I∶C ) ( Figure 5B ) . Reconstitution of Traf6-/- fibroblasts with wild type TRAF6 returned HSV-2 and poly ( I∶C ) -stimulated P-IκBα to levels consistent with wild type fibroblasts ( Figure 5B , C ) . However , reconstitution of Traf6-/- fibroblasts with TRAF6 C390S was unable to completely rescue P-IκBα and IFN-β mRNA induction in response to HSV-2 and poly ( I∶C ) ( Figure 5B , C ) . Furthermore , cells transfected with the C390S mutant were less sensitive to LNAC treatment than cells transfected with WT TRAF6 after poly ( I∶C ) treatment and HSV-2 infection , as measured by IκBα phosphorylation and IFN-β mRNA ( Figure 5D , E ) . By contrast , the ability of IL-1β to activate the NF-κB pathway was not compromised in TRAF6 C390S-expressing cells , despite strong sensitivity towards LNAC treatment ( Figure 5F ) , hence suggesting differences in the mode of action of ROS in different signaling pathways . Finally , introduction of a Cys-to-Ser mutation at position 455 in TRAF3 , which aligned with Cys390 of TRAF6 , led to reduced induction of IFN-β by HSV-2 after transfection into traf3-/- MEFs , and importantly abolishment of the sensitivity towards LNAC treatment , which was found in traf3-/- cells transfected with WT TRAF3 ( Figure 5G ) . Collectively , these data indicate that HSV infection leads to production of ROS , which is essential for activation of innate antiviral immune responses , and this proceeds via a mechanism involving S-glutathionylation of TRAF family proteins .
The innate immune response constitutes the first component of the host defense machinery against pathogens . Exposure to an invading pathogen triggers recruitment and activation of phagocytic cells that initiate a respiratory burst , consisting in a robust production of ROS , in order to eliminate the invading pathogen . However , it is increasingly believed that moderate intracellular concentrations of ROS can act as a second messenger involved in the activation of innate immune signaling pathways against pathogens . In this study , we demonstrate that ROS are produced early during HSV infection and are essential for triggering an effective antiviral response against HSV in vivo and in vitro , involving type I IFNs , CXCL10 and CCL5 secretion . There are several reports describing ROS production induced by viral infection , such as hepatitis C virus , influenza A , respiratory syncytial virus , and Sendai virus [34]–[37] . Lipopolysaccharide also stimulates ROS production via NOX and xanthine oxidase [30] , [38] . Due to the high reactivity attributed to ROS , these radicals can quickly interact with the surrounding macromolecules . This has been seen during pulmonary H5N1 influenza A virus infection where ROS production generates oxidized phospholipids , which are TLR4 agonists playing a key role in lung injury during infection [36] . The pro-inflammatory response triggered by H5N1 infection can be reduced with LNAC treatment and as a result reduce lung injury [39] . Likewise , it has been previously reported that HSV infection-induced ROS are responsible for the formation of lipid peroxidation byproducts and neurotoxicity , characteristic of encephalitis [40] , [41] . Thus , ROS are not only produced by macrophages and neutrophils to play a part in the killing of invading pathogens , but are a key component for the activation of PRR signaling . In the case of HSV , several PRRs are involved in the recognition of the virus [14] , including membrane-bound TLRs , and cytoplasmic RLRs and DNARs [3] , [6] , [7] , [9] , [12] , [13] , [42] . Here , we have examined the role of ROS in induction of innate immune responses by HSV as well as all the reported receptors recognising HSV . The data demonstrate that ROS are necessary for optimal signaling and gene expression during HSV infection or stimulation of PRRs . Most notably , the recently identified DNA receptor IFI16 senses HSV DNA in the cytoplasm and initiates a signaling cascade that is also redox-sensitive [7] . However , the sources of ROS might be different for the various receptors since they are located in different cellular compartments . It has previously been reported from separate groups that increased mitochondria-derived ROS production inhibits TLR4-mediated NF-κB activation [43] , [44] , whereas cytoplasmic NOX-derived ROS positively affects this activity [30] . Woo et al . have recently shed some light to this problem . They have shown that cells stimulated via immune receptors , such as T and B cell receptors , inactivate membrane-bound peroxiredoxin I ( PrxI ) and not cytoplasmic PrxI [45] . This allows transient localized accumulation of H2O2 around the membranes where signaling molecules such as protein tyrosine phosphatases are being recruited to the activated receptor [45] . Consequently , this intracellular oxidant gradient allows a regulated management of the signaling to the otherwise uncontrolled radical interactions , due to the nature of ROS . The recognition of viruses is coordinated at the subcellular level by triggering signaling pathways leading to activation of transcription factors , and consequent expression of antiviral cytokines and chemokines . All these processes are regulated at the post-transcriptional level , with phosphorylation being the best described . However , other post-translational modifications have gained increasing importance , such as oxidation , ubiquitination , SUMOylation or glutathionylation . In this work we show that HSV oxidative-dependent activation of NF-κB and IRF-3 pathways plays a key role downstream of the PRRs recognizing HSV . NF-κB was the first transcription factor recognized to be redox-regulated [46] . ROS not only activates NF-κB through promotion of the degradation of IκBα , but also induces several post-translational modifications of the NF-κB subunit p65 that are necessary for the transcription of NF-κB-dependent pro-inflammatory genes [47] . TNF-α-induced IL-8 gene expression in monocytes is mainly controlled at the transcriptional level through NF-κB although redox-sensitive phosphorylation at Serine 276 of p65 is required [48] . For the IRF-3 pathway , different outcomes have been described in redox-dependent activation of IRF-3 . It has been reported that NOX-derived ROS stimulate activation of the non-canonical IKK-like kinase , IKKε , and so IRF-3 activation , during infection with respiratory syncytial virus [34] . Recently , it has been shown that NOX2-derived ROS are critical for efficient RIG-I-mediated activation of IRF-3 and induction of IFN expression [37] , whereas activation of this pathway through TLR4 is dependent of NOX4-derived ROS and ASK1 [26] . Prinarakis et al . , on the other hand reported that IRF-3 is constitutively S-glutathionylated in HEK293 cells and is deglutathionylated by Grx upon Sendai virus infection , which is essential for the ability of IRF-3 to promote transcription [49] . In the primary murine peritoneal macrophages used in this study , constitutive glutathionylation of endogenous IRF-3 was not detected ( data not shown ) , but our data do not exclude that ROS enhance signaling at the level of IKKε and MAVS . Collectively , the literature on ROS and activation of PRRs suggest that ROS interacts with signaling at multiple steps , which may implicate that the subcellular sources of ROS as well as the microenvironment in which a PPR operates ( e . g . endosomes and mitochondrial surfaces ) impacts on the role and mechanism of action of ROS in innate immunity . This adds a layer of complexity to the already intricate signaling of the innate immune response . S-glutathionylation has also been suggested to be an important player in redox signaling [20] , [31] . We found that depletion of cellular GSH levels or reversing of S-glutathionylation by overexpression of Grx strongly inhibited HSV-induced expression of IFNs and ISGs . Moreover , HSV-2 infection led to detectable glutathionylation of both TRAF3 and TRAF6 , which was also seen in response to the potent ROS stimuli H2O2 . Finally , a conserved surface-exposed cysteine residue is presented in the TRAF domain of TRAF2 , 3 , and 6 , which is involved in redox-sensitive signaling pathways from PRRs and cytokine receptors [2] , but not in TRAF4 , which is generally believed not to be involved in these pathways [50] . We found that mutation of this cysteine in TRAF3 and 6 decreased the signaling and cytokine response to HSV-2 infection as well as the redox-sensitive nature of the response . Interestingly , a recent study demonstrated that human glutathione S-transferase P1-1 interacts with the TRAF domain of TRAF2 and specifically down-regulates signaling to MAPKs [32] , which together with our present findings suggests that S-glutathionylation of TRAFs or associated protein represents a mechanism by which ROS re-direct and amplify signaling in the innate immune system . In contrast to this model , a recent report based on Grx-1 knock-down in 293HEK and Hela cells suggested that TRAF6 gets deglutathionylated in the RING finger motif upon IL-1/TLR stimulation , which was proposed to be essential for subsequent activation [51] . Therefore , with the available data , it is clear that we are still at an early stage in our understanding of the role of S-glutathionylation in immunological signal transduction . In conclusion , we provide evidence that HSV infection quickly induces intracellular ROS which are necessary for proper activation of innate antiviral immune responses . The stimulatory functions of ROS appear to be mediated through S-glutathionylation , and we suggest TRAF family proteins to be important targets in positive ROS signaling . Recently , novel important roles for ROS in the innate immune system have been described , including inflammasome activation and programmed necrosis [28] , [52] . Thus , it is becoming increasingly evident that ROS are key players in the host response to infection and inflammation , and that further understanding of the molecular details underlying the production and action of ROS may provide important knowledge on antiviral response mechanisms and pathogenesis of many human diseases .
This study was carried out in accordance with the recommendations in the Guide for Care and Use of Laboratory animals , Institute of Laboratory Animal Resources , National Academy press ( 1996 ) . All animal experiments were done in accordance with a protocol ( permit number 2009/561-1613 ) , which was approved by The Danish Committee for Animal Research ( Ministry of Justice ) . C57BL/6 , TLR9-/- , TLR2/9-/- and ASK1-/- mice [11] , [53] , [54] were bred at M&B Taconic ( Laven , Denmark ) and kept in the animal facility at The Faculty of Health Sciences , AU between the time of delivery ( at 4 to 6 weeks of age ) and the time of the experiments , and used for experiments between the age of 7 and 9 weeks old . Peritoneal cells were harvested by lavage of the peritoneal cavity with PBS containing 5% foetal calf serum ( FCS ) and 20 IU/ml heparin . Cells were washed , counted , and re-suspended in RPMI 1640-5% FCS for sub-culturing . For in vitro experiments , the cells were used at a concentration of 3 . 0×106 cells per well in 96-well plates in 100 µl of RPMI 1640 with 5% FCS . BMMs were obtained as follows: femurs and tibia were surgically removed from C57BL/6 and MAVS-/- mice , freed of muscles and tendons , and briefly suspended in 70% ethanol . Ends were cut , the marrow was flushed with 10% RPMI 1640 , and cell suspension was filtered over a 70-µm cell strainer ( BD Falcon ) and centrifuged for 5 min at 1330 rpm . After 2 washes , cells were resuspended at 2×105/ml in RPMI 1640 with 10% FCS and GM-CSF ( 10 ng/ml ) and seeded in bacteriological petri dishes and incubated at 37°C with 5% CO2 and media changed after 3 and 5 days . On day 7 , adherent cells were harvested from the dishes with medium containing 10 ng/ml GM-CSF . The cells were centrifuged , washed , and resuspended in RPMI 1640 , 10% FCS , and GM-CSF ( 20 ng/ml ) , and examined by flow cytometry for expression of showing CD11b and CD11c ( data not shown ) . For in vitro experiments , the cells were used at a concentration of 1 . 0×106 cells per well in 96-well plates in 100 µl medium . To isolate primary pDCs cells from spleens from C57BL/6 and TLR9-/- mice , organs were surgically removed and transferred to RPMI 1640 with 5% FCS . The spleens were then transferred to a 1 mg/ml suspension of collagenase D ( Roche ) . The enzyme was injected into the organ , which was subsequently cut into small pieces , followed by incubation in collagenase D suspension for 30 min at 37°C . The suspension was filtered over a 70-µm pore size cell strainer , spun down , and suspended in RPMI 1640–5% FCS , and the cells were counted . After centrifugation , the cells were resuspended in PBS with 2 mM EDTA-0 . 5% BSA ( MACS running buffer ) in accordance with the manufacturer's instruction ( Miltenyi Biotec ) . Anti-mPDCA-1 microbeads were added , and after incubation for 15 min at 4°C the suspension was spun down and suspended in running buffer . pDCs were then isolated in an autoMACS separator by positive selection . For in vitro experiments , the cells were used at a concentration of 1 . 0×106 cells per well in 96-well plates in 100 µl RPMI 1640 with 5% FCS . RAW264 . 7 cells and MEFs ( C57BL/6 , traf3-/- ( G . Cheng , LA , CA , USA ) and traf6-/- ( T . Mak , Toronto , Canada ) were grown in DMEM supplemented with 5-10% FCS . The oxidant-sensitive dye 5- ( and-6 ) -chloromethyl-2′ , 7′-dichlorodihydrofluorescein diacetate ( CM-H2DCFDA ) was purchased from Invitrogen . Recombinant cytokines used for ELISAs were purchased from R&D Systems . The PRR agonists Poly ( IC ) :LyoVec , Poly ( I:C ) , Pam3Csk4 , and ODN1826 were obtained from InvivoGen , IL-1β was from R&D Systems , and the HSV-1 dsDNA 60mer , described earlier [7] , was from DNA Technology . Activation and inhibition of ROS production and function was achieved using H2O2 ( Sigma-Aldrich ) , LNAC ( Sigma Aldrich ) , PDTC ( Sigma-Aldrich ) , and for GSH depletion was used buthionine sulfoximine ( Fluka ) , diethylmaleate ( Sigma-Aldrich ) . All small molecule inhibitors were used in concentrations that did not affect cell viability as determined by annexin V and propidium iodide staining . The pCMV hTRAF6 and pRK5 hTRAF3 expression plasmids were kindly provided by Andrew G . Bowie ( Trinity College , Dublin ) . The TRAF6 C390S mutant was generated using the quick change kit ( Stratagene ) as described by the manufacture . The primers used were as follows . Forward: 5′-CCCGGGTACAAACTGTCCATGCGCTTGCACCTTCAGTTACCG-3′ , Reverse: 5′-CGGTAACTGAAGGTGCAAGCGCATGGACAGTTTGTACCCGGG-3′ . The TRAF3 C455S mutant was generated using a similar approach and the following primers; Forward: 5′-GGCTATAAGATGTCTGCCAGGGTCTACC-3′ , Reverse: 5′-GGTAGACCCTGGCAGACATCTTATAGCC-3′ . Oligonucleotides for HSV-1 60mer were synthesised by MWG Biotech; sequence is as follows: 5′-TAAGACACGATGCGATAAAATCTGTTTGTAAAATTTATTAAGGGTACAAATTGCCCTAGC-3′ . Forward and reverse strands were annealed by heating for 5 min to 95°C and cooling to room temperature . The viruses used were HSV-1 ( KOS strain ) , HSV-2 ( MS strain ) , Sendai virus ( Cantrell strain ) , AdV empty vector , and AdV-Grx [55] . Viruses were propagated and quantified as previously described [13] , [55] . For infections in vivo , mice were injected intraperitoneally with 5×106 pfu of HSV-2 as described previously [13] . At later time-points , serum , peritoneal cells , and livers were harvested for further analyses as described below . For modulation of ROS levels in vivo , mice were treated i . p . with LNAC ( 1 . 1 mmol/kg body weight ) in saline 4 h prior to virus infection . Samples of snap-frozen livers were weighed , thawed , and homogenized three times for 5 s in MEM supplemented with 5% FCS . After homogenization the organ suspensions were pelleted by centrifugation at 1 , 620 x g for 30 min , and the supernatants used for analysis . Plaque assays were performed on Vero cells as described previously [11] . The siRNAs were chemically synthesised by Qiagen ( p204-specific siRNA , 5′-CGGAGAGGAAUAAAUUCAUTT-3′; control siRNA , 5′-UUCUCCGAACGUGUCA CGUTT-3′ ) . RAW264 . 7 cells were seeded in 12-well plates at a density of 1×105 cells per well and were transfected with siRNA at a concentration of 12 . 5 pmol/ml with Lipofectamine 2000 ( 1 µl/ml ) . The cells were treated twice with siRNA on consecutive days and were grown for a further 48 h before stimulation . The degree of p204 knock-down was between 50 and 80% . RAW 264 . 7 macrophages were plated at 1×105 cells per well onto 10mm round coverslips . Cells were treated with inhibitors and infected with HSV-1 for 3 hr at 37°C , fixed with 4% formaldehyde ( 10 min , room temperature ) , permeabilised with methanol ( 90 sec at ÷20°C ) and blocked with 5% normal goat serum ( 15 min , room temperature ) . Cells were incubated with polyclonal rabbit anti-IRF-3 ( Santa Cruz Biotechnology , CA , USA ) at room temperature for 1 hr , and counterstained with Alexa Fluor 568 conjugated anti-rabbit antibodies ( Molecular Probes ) for an additional 1 hr at room temperature . Finally , cells were stained with DAPI and coverslips mounted in Pro-Long Gold ( Molecular Probes ) . Images were collected using Zeiss LSM710 confocal microscope 63 x oil objective . For quantification of cells positive for nuclear IRF-3 , regions of interest were identified based on DAPI staining , and levels of nuclear IRF-3 was determined . An arbitrary threshold was set , and percentage of positive cells in each group was calculated . Between 95 and 150 cells were counted for each condition and scored for IRF-3 localization . RAW 264 . 7 macrophages were seeded at 5×104 cells per well on a 8-wells chambered coverglass ( NUNC ) . Cells were either pre-treated with 6 , 4 mM LNAC for 30 min or with vehicle and subsequently infected with HSV-2 for 1 , 2 , or 3 hours . Media was then changed to a Phenol red-free DMEM and transferred to 37°C warmed heated-chamber of the confocal microscope . Cells were then incubated with 5 µM CM-H2DCFDA and imaged immediately after as described by manufacturer using Zeiss LSM710 confocal microscope 40x or 20x objectives . Z-stack images were taken ( 4–5 slices of 1 µm size ) at every 30 seconds for up to 1 , 5 min . For quantification of ROS production , background fluorescence was eliminated and total fluorescence was measured in each frame using Image J software . The number cells per frame was calculated and a minimum of 100 cells were counted per condition . Amount of ROS was represented as fluorescence per cell . Cytokine measurements were carried out using ELISAs based on matched antibody pairs from R&D Systems as described [13] . Levels of phosphorylated IκBα was determined using Luminex technology and kits from Bio-Rad , and following the instructions from the manufacturer . The bioactivity of type I IFN in culture supernatants and serum was determined using a L929-cell based bioassay as described previously [11] . High levels of IFN-γ or IFN-λ did not interfere with the assay . Total RNA from human and murine macrophages was purified using the High Pure RNA Isolation kit from Roche . For cDNA generation , RNA was subjected to reverse transcription with oligo ( dT ) as primer and Expand reverse transcriptase ( Roche ) . For data from Figure 3D , RNA was subjected to one-step qPCR ( Stratagene ) . For measurement of cytokines , the cDNA was PCR-amplified using the following primers: IFN-β , forward: 5′-CACTGGGTGGAATGAGACTAT-3′ , reverse: 5′-GACATCTCCCACGTCAATC-3′ , CCL5 , forward: 5′-ACTCCCTGCTGCTTTGCCTAC-3′ , reverse: 5′-GCGGTTCCTTCGAGTGACA-3′ , β-actin , forward: 5′-TAGCACCATGAAGATCAAGAT-3′ , reverse: 5′-CCGATCCACACAGAGTACTT-3′ . Products were measured using SYBR Green I ( Qiagen ) and normalized against β-actin . For processing of data , we selected the cycle threshold value of samples and normalized to β-actin following standard procedures . Data were presented as relative units compared to the average of untreated controls Cells were washed twice in phosphate-buffered saline , and lysed in 850 µl of lysis buffer ( 50 mM HEPES , pH 7 . 5 , 150 mM NaCl , 2 mM EDTA , 10% ( v/v ) glycerol , 1% ( v/v ) Nonidet P-40 , 0 . 5 mM DTT , Complete protease inhibitor ( Roche ) , 0 . 5 mM Na3VO4 , 20 mM N-ethylmaleimide , and 50 mM α-Iodoacetoamide ) . For immunoprecipitation , rabbit polyclonal anti-TRAF3 and rabbit polyclonal anti-TRAF6 ( Santa Cruz Biotechnology , CA , USA ) were precoupled to protein A/G-Agarose beads ( Santa Cruz Biotechnology , CA , USA ) overnight at 4°C . The beads were washed twice in lysis buffer and incubated with 1 . 5 mg of cell lysate per sample overnight at 4°C . The immune complexes were washed 3 times in lysis buffer , boiled , and analyzed by standard SDS-PAGE and Western blotting techniques using rabbit polyclonal anti-TRAF3/6 ( as above ) , mouse monoclonal anti-glutathione ( ViroGen , Watertown , VA , USA ) , and rabbit polyclonal anti-glutaredoxin-1 antibodies ( Santa Cruz Biotechnology , CA , USA ) for blotting . One molecule of glutathione was docked onto C390 in the published structure of TRAF6 ( 1LB5 and 1LB6 ) using the program Coot . This was possible with only minor changes of side chain conformation . The docking checked for steric clashes using the built-in routines in Coot . Coordinates for glutathione was taken from the Hicup server . The data are presented as means ± SD . The statistical significance was estimated with the Student t-test or Wilcoxon rank sum test ( p values of <0 . 05 were considered to be statistically significant ) . The results shown in this work are derived from data that are representative for the results obtained . For each series of experiments , two to six independent repetitions were performed .
|
Herpes simplex virus ( HSV ) type 1 and 2 are important human pathogens , which can give rise to severe diseases during both primary and recurrent infections . In addition to activating “classical” innate and adaptive immune responses , many infections stimulate other cellular activities such as and production of reactive oxygen species ( ROS ) . However , there is little knowledge on the cross-talk between ROS and the innate antiviral response . In this article we show that HSV infection leads to production of ROS , and that ROS play a critical role in activation of innate immune responses to these viruses . At the mechanistic level , we show that ROS stimulate glutathionylation ( a protein modification ) of the signaling molecules TRAF3 and 6 , which promotes redox-sensitive signaling . Our data support the idea that the innate immune system not only detects specific HSV molecules but also senses the cellular oxidative stress level , and integrates this into the innate immune response to infections .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"infectious",
"diseases",
"immunology",
"biology",
"microbiology"
] |
2011
|
HSV Infection Induces Production of ROS, which Potentiate Signaling from Pattern Recognition Receptors: Role for S-glutathionylation of TRAF3 and 6
|
Entamoeba nuttalli is an intestinal protozoan with pathogenic potential that can cause amebic liver abscess . It is highly prevalent in wild and captive macaques . Recently , cysts were detected in a caretaker of nonhuman primates in a zoo , indicating that E . nuttalli may be a zoonotic pathogen . Therefore , it is important to evaluate the pathogenicity of E . nuttalli in detail and in comparison with that of E . histolytica . Trophozoites of E . nuttalli GY4 and E . histolytica SAW755 strains were inoculated into liver of hamsters . Expression levels of proinflammatory factors of hamsters and virulence factors from E . histolytica and E . nuttalli were compared between the two parasites . Inoculations with trophozoites of E . nuttalli resulted in an average necrotic area of 24% in liver tissue in 7 days , whereas this area produced by E . histolytica was nearly 50% . Along with the mild liver tissue damage induced by E . nuttalli , expression levels of proinflammatory factors ( TNF-α , IL-6 and IL-1β ) and amebic virulence protein genes ( lectins , cysteine proteases and amoeba pores ) in local tissues were lower with E . nuttalli in comparison with E . histolytica . In addition , M2 type macrophages were increased in E . nuttalli-induced amebic liver abscesses in the late stage of disease progression and lysate of E . nuttalli trophozoites induced higher arginase expression than E . histolytica in vitro . The results show that differential secretion of amebic virulence proteins during E . nuttalli infection triggered lower levels of secretion of various cytokines and had an impact on polarization of macrophages towards a M1/M2 balance . However , regardless of the degree of macrophage polarization , there is unambiguous evidence of an intense acute inflammatory reaction in liver of hamsters after infection by both Entamoeba species .
The enteric protozoan Entamoeba histolytica causes an estimated 50 million cases of amebic colitis and liver abscess in humans , resulting in 40 , 000 to 100 , 000 deaths annually [1–5] . Entamoeba dispar is morphologically indistinguishable from E . histolytica , but is nonpathogenic . E . histolytica and E . dispar are also found in feces of nonhuman primates [6] . Recently , Entamoeba nuttalli , which is phylogenetically closer to E . histolytica than E . dispar , has also been identified in nonhuman primates [7] , and there is a high prevalence of E . nuttalli infections in wild and captive macaques , including Macaca mulatta , M . fasciculalis , M . fuscata , M . thibetana and M . sinica , and other nonhuman primates in zoos [8–15] . Most macaques with E . nuttalli infections are asymptomatic , suggesting that the host-parasite relationship in macaques may be commensal in natural infection [12] . More recently , cysts of E . nuttalli were detected in a caretaker of nonhuman primates in a zoo [16] . The infected person was asymptomatic , but this finding raises the possibility that E . nuttalli is a zoonotic pathogen . Fatal cases with liver abscess due to E . nuttalli have been reported in Abyssinian colobus and Geoffroy’s spider monkey in a zoo [17 , 18] , and inoculation of E . nuttalli trophozoites in liver of hamsters causes formation of abscesses and is lethal in some cases [7 , 10 , 12] . Hamsters inoculated with E . nuttalli are weakened and have decreased body weight . The liver lesions produced by E . nuttalli trophozoites are characterized by extensive necrosis associated with inflammatory reactions [7 , 10] . These histological changes are similar to those caused by E . histolytica trophozoites , suggesting similar pathological mechanisms of tissue damage [7 , 10 , 19 , 20] . However , E . histolytica infection in liver generally results in large single abscesses [1 , 21] , whereas E . nuttalli infection in hamsters induces small multiple abscesses [7] . Thus , the detailed mechanisms of how hosts with E . nuttalli develop a different pathogenic manifestation from that in E . histolytica infection are poorly defined . E . nuttalli is as virulent as E . histolytica in animal models , but it remains unclear whether E . nuttalli is virulent in humans . These findings , coupled with in vivo observations that E . nuttalli causes histological lesions in similar conditions and has few sequence differences in some important genes [7 , 14] in comparison with E . histolytica , have reinforced the idea that E . nuttalli is incapable of generating human lesions because of the host specificity of E . nuttalli and E . histolytica parasites . Therefore , it is important to evaluate the pathogenicity of E . nuttalli in comparison with that of E . histolytica to examine the molecular basis of the pathophysiology of amebic liver abscess ( ALA ) formation . In this study , expression levels of proinflamatory factors in hamsters and virulence factors from E . histolytica and E . nuttalli were compared between these parasites . The histopathological and immunopathological analyses of ALA provide valuable information on the pathogenicity of E . nuttalli .
All animal experiments were performed in strict accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals ( 1988 . 11 . 1 ) and were approved by the Institutional Animal Care and Use Committee ( IACUC ) of our institutions ( Permit Numbers 20110307–051 and 20160225–097 ) . All efforts were made to minimize suffering . Trophozoites of E . histolytica SAW755CR and E . nuttalli GY4 strains were grown under axenic conditions at 36 . 5°C in YIMDHA-S medium [22] containing 15% ( v⁄v ) heat-inactivated adult bovine serum . Trophozoites were harvested during the logarithmic growth phase ( 48 to 72 h ) by chilling on ice for 5 min . RAW264 . 7 cells were cultured in DMEM ( Thermo ) supplemented with 10% fetal bovine serum ( FBS ) ( Thermo ) , 100 U/ml penicillin , and 100 μg/ml streptomycin . CHO-K1 cells were cultured in Ham’s F12 nutrient medium ( Thermo ) supplemented with 10% FBS , 100 U/ml penicillin , and 100 μg/ml streptomycin . The mammalian cells were grown in a 37°C incubator with 5% CO2 . Six-week-old male hamsters were obtained from Shanghai Songlian Experimental Animal Factory . ALA was induced by direct inoculation of 1×106 axenic trophozoites of E . histolytica SAW755CR strain or E . nuttalli GY4 strain into liver , as previously described [23] . After intrahepatic inoculation of trophozoites , hamsters were euthanized at 3 h , 6 h , 12 h , 24 h , 48 h , 72 h and 168 h post-inoculation . At each time point , 6 to 7 hamsters were used . Liver tissues were harvested and fixed in 4% paraformaldehyde followed by paraffin embedding . Sections were stained with hematoxylin and eosin ( HE ) or periodic acid-Schiff ( PAS ) for histopathology [7] . Tissue damage and inflammatory cell infiltration were quantified in high quality images ( 2560×1920 pixels ) captured using a Nikon light microscope . Areas of leukocyte infiltration and liver necrosis were measured using Image-Pro Plus 4 . 5 . 1 software ( Media Cybernetics ) . Areas of interest are expressed as a percentage of the total tissue area . Immunohistochemical staining was performed as described elsewhere [24] . Briefly , paraformaldehyde-fixed liver sections were deparaffinized , rehydrated by standard protocols and incubated overnight at 4°C with rabbit anti mouse IFN-γ , TNF-α , IL-1β and IL-6 polyclonal antibodies ( Abcam ) . The slides were subsequently incubated with horseradish peroxidase-labeled goat anti-rabbit immunoglobulin and then with chromogen substrate ( 3 , 3′-diaminobenzidine ) for 2 min before counterstaining with hematoxylin . The cytokine score ( intensity area/total image area ) was determined in areas of leukocyte infiltration using Image-Pro Plus 4 . 5 . 1 software ( Media Cybernetics ) . Each cytokine score was determined by counting more than 50 high-power fields ( ×20 ) . Gene expression of IFN-γ , TNF-α , IL-1β , IL-4 , IL-6 , IL-10 , nitric oxide synthase ( iNOS ) , arginine enzyme I ( Arg-1 ) and mannose receptor I ( MRC-I ) in liver tissues was examined by quantitative real-time PCR ( qRT-PCR ) using the primers listed in Table 1 [22 , 25–26] . Briefly , total RNA ( 1 μg ) of tissue from the edge of a liver abscess was purified with an RNeasy Plus Mini kit ( Qiagen ) . cDNA was synthesized with a Primescript first-strand cDNA synthesis kit ( Takara ) using oligo ( dT ) primers . qRT-PCR was carried out in a final reaction volume of 20 μl on an ABI 7500 Real-time PCR system ( Applied Biosystems ) . Reactions were performed in a 96-well plate with SYBR Premix Ex Taq ( Takara , Japan ) containing primers listed in Table 1 . The amplification cycling conditions were as follows: 30 s at 95°C and 40 cycles of 5 s at 95°C and 35 s at 60°C . Analysis by qRT-PCR of gene expression was conducted during the log phase of product accumulation , during which Ct values correlated linearly with relative DNA copy numbers . Each experiment was performed at least three times . Gene expression of heavy subunit of galactose/N-acetylgalactosamine lectin ( Hgl ) , intermediate subunit of galactose/N-acetylgalactosamine lectin ( Igl ) , cysteine proteinase 2 ( CP2 ) , cysteine proteinase 5 ( CP5 ) , amoebapore A ( AP-A ) , and amoebapore B ( AP-B ) of Entamoeba trophozoites in ALA tissues were examined using the same RNA samples extracted from tissue at the edge of a liver abscess . Genes of virulence proteins of E . nuttalli GY4 strain were amplified and sequenced ( S1 Text ) . Primers for these genes were designed using the identical sequence regions of E . histolytica SAW755CR strain and E . nuttalli GY4 strain , and are listed in Table 1 . The primers for Hgl and Igl used in this study can amplify all known subtypes of Hgl and Igl genes . Reactions were performed as described above . Each experiment was performed at least three times . To assess whether secretory proteins from E . histolytica and E . nuttalli cause polarization of macrophages , trophozoites were first incubated with CHO cells and then lysed . RAW264 . 7 cells were stimulated with the lysates of ameba trophozoites . Briefly , CHO-K1 cells ( 106 ) were cultured in a 35-mm dish ( Costar ) , and then 5×105 trophozoites were added and coincubated for 30 min . Trophozoites were then harvested , and lysed by repeated freezing and thawing of 106 trophozoites per ml in PBS . After centrifugation at 20 , 000 g for 10 min , trophozoite lysates were used to stimulate RAW264 . 7 cells ( 5×105 ) in 24-well culture plates ( Costar ) overnight . The cells were stimulated with trophozoite lysates , LPS ( 1 μg/ml final conc . ) ( Sigma ) or PBS . After incubation for 6 h , 12 h , 24 h and 48 h , culture supernatants of RAW264 . 7 cells were assayed for cytokines and NO production . Cells were collected and frozen for measurement of arginase activity and expression of iNOS and Arg-1 genes by qRT-PCR . Each experiment was performed at least three times . Total RNA ( 1 μg ) of treated RAW264 . 7 cells was purified with an RNeasy Plus Mini kit ( Qiagen ) . cDNA was synthesized with a Primescript first-strand cDNA synthesis kit ( Takara ) using oligo ( dT ) primers . Gene expression of iNOS and Arg-1 was examined by qRT-PCR using the primers listed in Table 1 . Reactions were performed as described above . Each experiment was performed at least three times . A Griess assay was performed using 20 μl of culture supernatant of RAW264 . 7 cells mixed with 30 μl of distilled water and 50 μl of Griess reagent ( Sigma ) . Absorbance was measured at 548 nm in a microplate reader [27 , 28] . Experiments were performed at least three times . Arginase activity in cell lysates was measured in RAW264 . 7 cells that were harvested and lysed with mammalian tissue lysis/extraction reagent ( Sigma ) for 15 min on a shaker and centrifuged at 13 , 000 g for 10 min to remove insoluble material . Sample supernatant ( 20 μl ) was added to a well of a 96-well plate , 10 μl of substrate buffer was added , and the mixture was incubated at 37°C for 120 min for arginine hydrolysis . The reaction was stopped with 200 μl of urea in each well at room temperature for 30 min . Absorbance was measured at 430 nm in a microplate reader . One unit of arginase is the amount of enzyme that converts 1 . 0 mM of L-arginine to ornithine and urea per minute at pH 9 . 5 and 37°C . Each experiment was performed at least three times . To assess whether secretory proteins from E . histolytica and E . nuttalli cause proliferation of macrophages , a cell proliferation assay was performed . RAW264 . 7 cells ( 5×104 ) were cultured in 96-well culture plates ( Costar ) overnight . Cells were stimulated with 5 μl of PBS , trophozoite lysates or LPS ( 1 μg/ml final ) . CCK-8 reagent ( Dojindo ) was added to each well at 4 h , 10 h , 22 h or 46 h , and optical density ( OD ) was measured at 450 nm using a microplate reader ( Bio-rad ) at 6 h , 12 h , 24 h or 48 h . Each experiment was performed at least three times . A Luminex multiplex immunoassay was performed to determine the concentrations of inflammatory cytokines using a customized Milliplex Mouse Cytokine/Chemokine Magnetic Bead Panel ( Merck Millipore ) for IL-1β , IL-6 and TNF-α . Briefly , 25μl of cell supernatant , control or standard was added to a 96-well plate containing 25μl of capture antibody-coated , fluorescent-coded beads . Biotinylated detection antibodies and streptavidin-PE were added to the plate after the appropriate incubation periods . After the last washing step , 150μl of sheath fluid was added to the wells , and the plate was incubated and read on a Luminex100 instrument . Five-PL regression curves were used to plot standard curves for all analytes with xPonent 3 . 1 software by analyzing the bead median fluorescence intensity . Results are expressed in pg/ml . Samples with quantification below the detection limit were registered as “zero” and samples above the quantification limit of the standard curve were given the value equal to the highest value of the curve . Each experiment was performed at least three times . Statistical analyses were performed using IBM SPSS ( ver . 20 , SPSS Statistics/IBM Corp . , Chicago , IL , USA ) . qRT-PCR data were analyzed by two-tailed Mann-Whitney U test . Other data were analyzed with a two-tailed Student t-test . P < 0 . 05 was considered significant in all analyses .
After inoculation of trophozoites into hamster liver , both E . histolytica SAW755 and E . nuttalli GY4 caused ALA , with a clear boundary between the abscess and normal tissue . The main area of inflammatory cell infiltration and living trophozoites was located at the edge of the abscess . At 3 h post-inoculation , inflammatory cell infiltration ( mainly neutrophils ) was observed ( Figs 1A , S1–S3 ) , and then infiltration of inflammatory cells increased in liver tissues . These cells were mainly monocytes and macrophages . A clear liver abscess was seen from 24 to 168 h with E . histolytica and 48 to 168 h with E . nuttalli ( Fig 1B ) . ALA areas increased to nearly 50% with E . histolytica at 168 h , whereas the average ALA area was only 24% at 168 h with E . nuttalli . These results indicate that mild liver tissue damage was induced by E . nuttalli GY4 strain . Similarly , the area of inflammatory cell infiltration with E . nuttalli was smaller than that with E . histolytica at each time point . At 168 h , the inflammatory cell infiltration area was significantly lower with E . nuttalli ( 7% ) than with E . histolytica ( 12% ) ( Fig 1C ) . To evaluate expression of cytokines during ALA formation by E . nuttalli trophozoites , hamster livers were used for immunohistochemistry at different time points , and expression levels of IFN-γ , TNF-α , IL-6 and IL-1β in liver abscesses were analyzed . The control group had low levels of these cytokines ( Fig 2A ) at each time point , whereas areas positive for TNF-α , IL-6 and IL-1β in liver were increased after inoculation with E . histolytica SAW755CR ( Fig 2B–2E ) . These areas were also increased in liver inoculated with E . nuttalli GY4 , but to a lesser extent compared with E . histolytica ( Fig 2C–2E ) . These results indicate a smaller increase in expression of proinflammatory factors ( TNF-α , IL-6 and IL-1β ) in local tissues inoculated with E . nuttalli GY4 . Little IFN-γ was detected in liver tissue after inoculation of either strain . To quantify the changes in cytokines during ALA formation , qRT-PCR was used to amplify IFN-γ , TNF-α , IL-6 , IL-1β , IL-4 and IL-10 genes , with β-actin amplified as a reference . The results for expression levels of TNF-α , IL-6 and IL-1β in tissue at the edge of liver abscesses were similar to the immunohistochemistry data . At most time points , IL-6 and IL-1β increased significantly with E . histolytica and E . nuttalli ( Fig 3 ) . qRT-PCR showed that expression of TNF-α , IL-6 and IL-1β with E . nuttalli GY4 was lower than with E . histolytica SAW755CR at 48 h , 72 h and 168 h , and expression of IL-6 was particularly significantly lower . Macrophages are immune effector cells that play an important role in ALA development . Expression of iNOS and Arg-1 genes was analyzed by qRT-PCR to examine differences in macrophage polarization in liver abscesses induced by the two species of Entamoeba trophozoites . Expression of iNOS rose rapidly after inoculation , whereas that of Arg-1 decreased ( Fig 4A and 4B ) . MRC-I is a highly expressed surface receptor on M2 macrophages , and expression levels of MRC-I and Arg-1 changed similarly in qRT-PCR . In the early stage of ALA formation , MRC-I expression increased in all damaged liver tissue ( including in the control group ) . The MRC-I level with E . nuttalli GY4 continued to increase at 72 h and 168 h , but did not rise further in the control group or with E . histolytica SAW755CR ( Fig 4C ) . The higher iNOS/Arg-1 ratio ( macrophage polarization M1/M2 ) at 72 h and 168 h suggests macrophage polarization toward M1 with E . histolytica SAW755CR , but toward a M1/M2 balance with E . nuttalli GY4 at 168 h ( Fig 4D ) . These results suggest that M2 macrophages increased at 168 h after inoculation of the GY4 strain , and the milder liver tissue damage caused by E . nuttalli GY4 strain might be attributable to the increase in these macrophages . To compare changes of virulence proteins with the two Entamoeba species during ALA progression , qRT-PCR was performed to examine expression levels of Hgl , Igl , CP-2 , CP-5 , AP-A and AP-B genes . There were significant increases in Hgl ( 2- to 5-fold ) , CP2 ( 5- to 16-fold ) , CP5 ( 2- to 11-fold ) , AP-A ( 2- to 3-fold ) and AP-B ( 2- to 7-fold ) after inoculation in hamster liver ( Fig 5 ) . There were few differences between the expression levels of virulence protein genes of E . histolytica and E . nuttalli in vitro; only the CP5 level was half in E . nuttalli compared to E . histolytica ( Fig 5D ) . However , significantly lower expression of CP2 , CP5 , AP-A and AP-B of E . nuttalli was found at 168 h , and this lower level of virulence proteins in vivo may contribute to the milder liver tissue damage caused by E . nuttalli GY4 . To study whether secretory proteins of Entamoeba play an important role in polarization of macrophages , in vitro stimulation of mice macrophage RAW264 . 7 cells was performed , and expression of iNOS and Arg-1 was analyzed by qRT-PCR . The levels of both of these genes rose rapidly after stimulation with trophozoite lysates of E . histolytica or E . nuttalli . Lysate of E . histolytica induced significantly higher iNOS expression than that of E . nuttalli at 48 h ( 1334-fold vs . 627-fold compared to PBS ) . In contrast , lysate of E . nuttalli induced higher Arg-1 expression than E . histolytica ( 122-fold vs . 61-fold compared to PBS ) ( Fig 6A and 6B ) . To examine the effects of secretory proteins of Entamoeba on NO production , RAW264 . 7 cells were stimulated with PBS , trophozoite lysate of E . histolytica or E . nuttalli , and LPS . A stable oxidized product of NO in the cell culture supernatants was determined by the Griess assay . Nitrate was increased by stimulation with lysate of E . histolytica at 24 h ( 24 . 8 μM ) and 48 h ( 36 . 9 μM ) ( Fig 6C ) . The effect of trophozoite lysates on arginase activity was also examined . Arginase activity induced by lysates of E . histolytica and E . nuttalli , and LPS was increased at 24 h and 48 h , with lysate of E . nuttalli inducing higher arginase activity ( 5 . 4 unit/L ) than that of E . histolytica at 48 h ( Fig 6D ) . These results indicate that secretory proteins of Entamoeba play important roles in the polarization balance of macrophages . A cell proliferation assay indicated that RAW264 . 7 cells were capable of proliferation after stimulation with trophozoite lysates of E . histolytica and E . nuttalli . An additional 33% to 42% proliferation of RAW264 . 7 cells occurred in comparison with PBS-stimulated cells at 48 h . There was no significant difference between lysates of E . histolytica and E . nuttalli ( Fig 6E ) . Cytokine expression of RAW264 . 7 cells was determined using a Luminex multiplex immunoassay after stimulation with PBS , trophozoite lysate of E . histolytica or E . nuttalli , and LPS . Lysate of E . histolytica induced a significant increase in TNF-α ( 507 . 9 pg/ml ) , IL-6 ( 2573 . 0 pg/ml ) and IL-1β ( 17 . 3 pg/ml ) at 48 h . Lysate of E . nuttalli also caused a significant increase in TNF-α ( 435 . 5 pg/ml ) at 48 h , but had no significant effect on IL-6 and IL-1β ( Fig 7 ) . These results are consistent with the lower expression levels of IL-6 and IL-1β with E . nuttalli GY4 in the hamster ALA model .
The aim of this study was to determine the histopathological features of ALAs that regulate host inflammatory immune responses following their interaction with parasites and to examine whether these features differ between E . nuttalli and E . histolytica . Our data show that both E . nuttalli and E . histolytica cause liver abscesses with a clear boundary between the abscess and normal tissue . Interestingly , lesions in hamster differed between E . nuttalli and E . histolytica , including the size of the abscess and inflammatory cell infiltration region . The ability of trophozoites to produce a liver abscess in hamsters also differs among strains of E . histolytica and E . dispar , and the SAW755 strain of E . histolytica used in this study is a highly virulent strain . Even in hamsters inoculated with E . nuttalli trophozoites , lethal cases occurred within 7 days using the NASA06 strain [10] , and in hamsters inoculated with the E . nuttalli SSS212 , the mean abscess size was >50% of the liver [12] . In liver tissue sections , the necrotic area with inflammatory reactions was highly extended . Therefore , the virulency of the E . nuttalli GY4 strain used in the present study may have been relatively low . Tissue destruction during ALA formation is generally attributable to both the cytotoxicity of trophozoites and the resultant host inflammatory immune response [29] . A typical amebic lesion is characterized by a necrotic zone with edges consisting of cellular debris and inflammatory cell infiltration [30] . Such necrosis is produced by virulence factors of trophozoites , such as galactose/N-acetylgalactosamine lectin ( Gal/GalNac lectin ) , APs and CPs [29–33] . In the present study , the expression levels of virulence factors of trophozoites in tissue was higher than that of axenically cultured trophozoites , but the levels of major CPs and APs differed between the two strains , with lower levels in E . nuttalli . Immunopathological effects also contribute to tissue destruction during liver abscess formation in the hamster model . The host inflammatory response suppresses invasive trophozoites , but also leads to severe tissue damage . During this infectious process , multiple types of inflammatory cells are recruited to the infected liver of hamsters . Infiltrating neutrophils are predominant in inflammatory regions in the initial phase of invasive liver amebic infection , followed by macrophages that accumulate rapidly during abscess formation . Evidence from in vivo and in vitro studies suggests that macrophage-mediated anti-ameba activity is a major mode of host defense against E . histolytica infections , and has essential functions throughout ALA formation [34] . When pathogens attack , naïve macrophages can be polarized in a direction to classical activated ( M1 ) macrophages that strongly express iNOS , which produces NO through catabolism of arginine , subsequently causing proinflammatory effects and tissue damage [35 , 36] . During abscess formation , M1 macrophages release NO into infected tissue , and NO combined with toxic products from the oxidative burst then kill trophozoites . The macrophage-mediated anti-ameba activity is inhibited by arginase in a dose-dependent manner through competition with iNOS that depletes the common substrate , L-arginine [37 , 38] . Macrophages can also be polarized into alternatively activated ( M2 ) macrophages and induce Arg-1 , which competes with iNOS by degrading arginine into ornithine and polyamines , giving rise to macrophages with anti-inflammatory effects and tissue repair functions [35 , 36 , 39] . The present study showed that E . nuttalli GY4 induced small liver abscesses at 168 h after inoculation compared with large abscesses driven by E . histolytica SAW755CR . Moreover , infiltration of inflammatory cells remained lower in abscess lesions of E . nuttalli compared to those caused by E . histolytica . In amebic liver lesions , secretion of Gal/GalNac lectin , APs and CPs by trophozoites also results in destruction of neutrophils and liberation of their toxic products , which may play an important role in enlargement of abscess lesions [29 , 30] . Immunosuppressive and tissue repair functions play critical roles in control of inflammation by producing anti-inflammatory mediators [36 , 39] . In this study , both Entamoeba species caused increased levels of iNOS in liver lesions of hamsters and decreased arginine at the early stage of ALA formation , indicating elevation of M1 macrophages , which are involved in host defense and tissue damage . Significantly , the level of arginine increased with E . nuttalli at 168 h after trophozoite inoculation , which suggests greater elevation of M2 macrophages compared with E . histolytica infection . The increased proportion of M2 macrophages in liver abscess lesions might attenuate tissue damage through accelerated tissue repair , and this might explain the smaller abscesses and milder liver tissue damage in the animal model infected with E . nuttalli . There are several key factors in macrophage polarization during infection , with pathogens and their virulence proteins being the fundamental regulators [39–41] . The current study indicated that proteins secreted by both Entamoeba species were able to induce macrophage polarization and skew differentiation towards M1 or M2 phenotypes . With E . histolytica , the macrophage polarization skewed towards the M1 phenotype , as shown by the significant increase in iNOS expression and multiple proinflammatory cytokines , such as TNF-α , IL-1 and IL-6 , exerting immunoregulatory roles during infection . With E . nuttalli , the polarization trend of macrophages was not as clear , based on the lower levels of iNOS and cytokines and higher production of arginine , compared to E . histolytica infection . These results suggest an equilibrium in macrophage polarization . Several studies have shown amebicidal activity of macrophages mediated by iNOS mRNA expression and NO production [30 , 42 , 43] . There is also evidence of direct macrophage activity by amebic virulence factors , and E . nuttalli secreted fewer virulence factors than E . histolytica based on protein profiles . Taken together , these data indicate that virulence factors inducing macrophage polarization in hamster liver lesions switch to a protective M2 phenotype from a destructive M1 phenotype , leading to decreased NO production , which reduces immunopathological tissue damage . T-cell cytokine responses can be divided into different classes based on the combination of cytokines produced . Th1 cells secrete cytokines including IL-2 , IFN-γ and TNF-β that promote differentiation and activity of macrophages and cytotoxic T cells , and lead primarily to a cytotoxic immune response . In contrast , the Th2 cytokine response is characterized by IL-4 , IL-5 , IL-6 , IL-9 and IL-10 production [44] . These cytokines , the levels of which correlate with the degree of tissue damage , are released by attacked host cells or effector cells . IFN-γ is a suppressive cytokine that can clear the parasite [45] . In the present study , there was no significant increase in IFN-γ during ALA development in hamsters , perhaps suggesting that persistent progression of lesions facilitates invasive amebiasis . In contrast , TNF-α and IL-6 , which are inflammatory factors , were strongly sustained and expressed during progression of tissue damage . Macrophages were clearly the major effector cells secreting these cytokine mediators . M1 macrophages secreted proinflammatory cytokines , including TNF-α , IL-1β and IL-6 , which activate phagocytes to kill pathogens , but also cause tissue damage [46 , 47] . The results of immunohistochemistry and qRT-PCR indicated that the levels of multiple cytokines increased during ALA formation . These results suggest that macrophage polarization might profoundly affect the degree of tissue damage in ALA formation . At the early stage of infection , TNF-α , IL-1β and IL-6 showed an increasing trend with both Entamoeba species , but with higher levels with E . histolytica than with E . nuttalli . The increased TNF-α and IL-6 during amebic tissue damage results in activation of macrophages to release NO and thereby exert an anti-inflammatory effect . Our data show that macrophage polarization by E . histolytica SAW755CR induced greater upregulation of iNOS expression at the transcriptional level , resulting in a higher proportion of M1 polarized macrophages , which then secreted higher levels of proinflammatory cytokines and aggravated amebic tissue damage . The released TNF-α and IL-1β feedback to further skew macrophage differentiation towards the M1 phenotype . Consequently , small multiple abscesses merge with each other and coalesce to form large single abscesses after infection with E . histolytica SAW755CR . M2 macrophages generally have high levels of mannose receptors and scavenger receptors , and play important roles in polarized Th2 reactions . For example , M2 macrophages promote the encapsulation and killing of parasites and have immunoregulatory and anti-inflammatory functions [48] . This macrophage population is also thought to play a critical role in negative regulation of host protective immunity against microbial infections . Thus , M2 macrophages modulate expression of anti-inflammatory cytokines such as MRC or transforming growth factor , and thereby modulate suppression of tissue inflammation and enhance tissue repair [49] . In E . nuttalli infection , expression levels of TNF-α and IL-1β decreased at 168 h after inoculation , whereas expression of MRC was upregulated with the same trend as that of Arg-1 . The downregulation of TNF-α and IL-1β suggests that tissue damage might be slowed in E . nuttalli infection . Macrophage polarization tends to reach an equilibrium with the increase in the M2 phenotype . Repair-associated factors begin to take effect and inhibit ALA formation and development . Finally , this leads to formation of multiple small abscesses that are incapable of coalescing into larger lesions , consistent with the finding that E . nuttalli GY4 forms smaller liver abscesses than E . histolytica SAW755 . Thus , our results show that differential secretion of amebic virulence factors in E . nuttalli infection may trigger lower cytokine secretion and promote polarization of macrophages towards a M1/M2 balance . In consideration of intestinal immunity that is the first line of defense against amoeba trophozoites , a critical aspect in Entamoeba pathogenesis is to overcome the colonic epithelial barrier [50–53] . The intestinal bacterial microbiota is another important factor that influence the pathogenesis of E . histolytica . This could be interrelated to direct ingestion of intestinal bacteria that increase the expression of virulence proteins of E . histolytica [54–58] . The bacteria could also alter the immune status of host intestine to prevent or promote amoebiasis . For instance , the increasing of IL-23 , IL-17 , dendritic cells and neutrophil induced by segmented filamentous bacterium in the cecum mediated protection from E . histolytica [59 , 60] . E . nuttalli trophozoites may elicit ALA formation with intense inflammatory reaction in human if the parasites translocate to liver . However , E . nuttalli is probably not adapted to intestinal microenvironment of human and unable to invade beyond the colonic epithelial barrier of human under natural conditions . There is unambiguous evidence of an intense acute inflammatory reaction in hamster liver in infection by both Entamoeba species , but no evidence showing that these events in hamster liver also occur in human liver . Moreover , it is unknown whether E . nuttalli trophozoites produce intestinal ulcer in human and non-human primates . Additionally , as well known in genetic restriction , a T-cell receptor recognizes a particular antigenic peptide presented by a specific histocompatibility complex ( MHC ) molecule , and this interaction is associated with susceptibility or resistance to pathogen infection [61–64] . For instance , highly polymorphic HLA genes have an enormous capacity to bind to viral peptides associated with HBV infection [65] and a single MHC supertype confers qualitative resistance to Plasmodium relictum infections in avian malaria [66] . Consequently , host MHC molecule may also play a key role in determination of host susceptibility to E . nuttalli . In conclusion , histopathological features and expression levels of proinflamatory factors in ALAs formed by E . nuttalli were identified in this study . The results also suggest that the difference of tissue damage in infection by E . histolytica and E . nuttalli is due to the levels of secretion of various cytokines , regardless of the extent of macrophage polarization . In any event , both Entamoeba species induced intense acute inflammatory reactions in liver of hamsters after infection .
|
Entamoeba nuttalli is the phylogenetically closest protozoan to Entamoeba histolytica and is highly prevalent in macaques . Previous studies have indicated that E . nuttalli is virulent in a hamster model . In this study , we compared the immunopathological basis of formation of liver abscess in hamsters between E . nuttalli and E . histolytica . Mild liver tissue damage developed after intrahepatic injection of trophozoites of E . nuttalli , and lower expression levels of genes for host proinflammatory factors and amebic virulence proteins were detected at the edges of liver abscesses induced by E . nuttalli . In addition , alternatively activated macrophages were increased in E . nuttalli-induced liver abscesses in the late stage of disease progression . The lysate of E . nuttalli trophozoites also induced higher arginase expression than E . histolytica in vitro . Polarization of macrophages is likely to affect the degree of acute inflammatory reactions in liver in an animal model during E . nuttalli infection . Our data reveal new characteristics of abscess formation by E . nuttalli .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"blood",
"cells",
"trophozoites",
"parasite",
"groups",
"medicine",
"and",
"health",
"sciences",
"innate",
"immune",
"system",
"immune",
"cells",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"cytokines",
"pathogens",
"immunology",
"vertebrates",
"parasitic",
"protozoans",
"animals",
"parasitology",
"mammals",
"developmental",
"biology",
"apicomplexa",
"protozoans",
"signs",
"and",
"symptoms",
"molecular",
"development",
"abscesses",
"hamsters",
"white",
"blood",
"cells",
"inflammation",
"animal",
"cells",
"entamoeba",
"histolytica",
"immune",
"response",
"immune",
"system",
"rodents",
"eukaryota",
"diagnostic",
"medicine",
"cell",
"biology",
"virulence",
"factors",
"physiology",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"macrophages",
"amniotes",
"organisms"
] |
2018
|
Characteristics of inflammatory reactions during development of liver abscess in hamsters inoculated with Entamoeba nuttalli
|
Infection of the developing fetus with human cytomegalovirus ( HCMV ) is a major cause of central nervous system disease in infants and children; however , mechanism ( s ) of disease associated with this intrauterine infection remain poorly understood . Utilizing a mouse model of HCMV infection of the developing CNS , we have shown that peripheral inoculation of newborn mice with murine CMV ( MCMV ) results in CNS infection and developmental abnormalities that recapitulate key features of the human infection . In this model , animals exhibit decreased granule neuron precursor cell ( GNPC ) proliferation and altered morphogenesis of the cerebellar cortex . Deficits in cerebellar cortical development are symmetric and global even though infection of the CNS results in a non-necrotizing encephalitis characterized by widely scattered foci of virus-infected cells with mononuclear cell infiltrates . These findings suggested that inflammation induced by MCMV infection could underlie deficits in CNS development . We investigated the contribution of host inflammatory responses to abnormal cerebellar development by modulating inflammatory responses in infected mice with glucocorticoids . Treatment of infected animals with glucocorticoids decreased activation of CNS mononuclear cells and expression of inflammatory cytokines ( TNF-α , IFN-β and IFNγ ) in the CNS while minimally impacting CNS virus replication . Glucocorticoid treatment also limited morphogenic abnormalities and normalized the expression of developmentally regulated genes within the cerebellum . Importantly , GNPC proliferation deficits were normalized in MCMV infected mice following glucocorticoid treatment . Our findings argue that host inflammatory responses to MCMV infection contribute to deficits in CNS development in MCMV infected mice and suggest that similar mechanisms of disease could be responsible for the abnormal CNS development in human infants infected in-utero with HCMV .
Viral infections in the fetus and young infant are well described causes of abnormal brain development that often result in permanent neurological sequelae , including disorders of motor and cognitive functions . Altered CNS development and neurologic disease have been documented in the developing fetus and young infant following infection with a number of viruses , such as herpes simplex virus ( HSV ) , rubella , lymphocytic choriomeningitis ( LCMV ) and human cytomegalovirus ( HCMV ) [1]–[7] . A variety of mechanisms can lead to interruption of the developmental program of the CNS including: damage to the brain parenchyma secondary to apoptotic or necrotic loss of resident cells within the CNS , damage to the supporting vasculature and microvascular supply of the CNS resulting in decreased blood flow and/or damage to the blood brain barrier , altered cellular positioning and disruption of synapse formation leading to a failure in neuronal connectivity and circuitry formation [8] , [9] . In the case of infection with viruses that exhibit specific cellular tropism , the loss or dysfunction of specific populations of resident cells within the CNS often underlies disease . In other cases , cellular tropism is broad and disease is thought to result from direct viral damage to supporting structures , such as the vasculature or the glial architecture . Additionally , indirect mechanisms of disease following CNS infection include viral induced host inflammatory responses [10] , [11] . Host responses following virus infections often lead to more global CNS damage secondary to the production of soluble effector molecules that can amplify proinflammatory responses of resident cells as well as promote cytotoxic activity by effector cells of the adaptive immune system [12]–[23] . Although these mechanisms of disease , as well as other proposed mechanisms , are consistent with clinical findings in patients with viral encephalitis , a precise description of the pathogenesis of CNS disease in virus infected human fetuses and infants is often limited by the lack of informative tissue specimens . Because of limitations inherent in studies of the human CNS , small animal models have been developed to elucidate mechanisms of disease associated with viral infections of the developing CNS . These models have utilized a number of different viruses including HSV , murine cytomegalovirus ( MCMV ) , LCMV , alphaviruses and more recently West Nile Virus ( WNV ) [4] , [24]–[30] . Studies of CNS disease following both peripheral and intracerebral HSV inoculation have described a necrotizing encephalitis , which is more severe in animals with deficits in innate and adaptive immunity [31]–[33] . However , more recent studies have argued that in addition to the direct cytopathic effects associated with HSV replication , host derived innate immune responses contribute to CNS damage in infected mice [34] , [35] . Similarly , experimental models employing LCMV infection have provided direct evidence that host-derived inflammation is a major component of CNS disease [4] , [36] . In these models , limiting CD8+ virus specific T lymphocyte responses , or more global immunosuppression , dramatically reduced the severity of CNS disease [4] , [37] . The contribution of immunopathological responses are particularly relevant to disease in young animals because expression of inflammatory genes during the dynamic developmental program of the CNS appears to result in a disease phenotype that differs from that seen in adult animals . Thus , substantial CNS damage in young infants could result from infection with viruses that are infrequently pathogenic in adults . In contrast , an effective immune response does appear to be necessary to limit the severity of CNS infection with alphaviruses and WNV [24]–[26] , [38]–[42] . Responses derived from the adaptive immune system , in particular the production of antiviral antibodies , determine the susceptibility of newborn animals to alphavirus infection of the CNS [25] , [38] , [43] , [44] . Thus disease outcome in young animals with viral infections of the CNS reflects a balance between unregulated inflammation and the control of virus replication [18] , [32] , [45]–[51] . Intrauterine infection with HCMV is the most common cause of congenital ( present at birth ) infection in humans and occurs in approximately 1/200 live births in the United States [52] . A small but significant number of newborn infants infected in-utero exhibit a variety of neurodevelopmental abnormalities secondary to HCMV infection of the CNS [5] , [6] . Because little is known about the mechanisms of disease associated with this intrauterine infection , we developed a murine model of CNS infection that utilizes peripheral inoculation of newborn animals with limiting amounts of MCMV . In contrast to other murine models that have utilized intracranial inoculations of MCMV almost exclusively , the model we have developed uses intraperitoneal inoculation of limiting amounts of MCMV and requires virus replication in the periphery , viremia and neuroinvasion . These latter features of this murine model , particularly the hematogenous spread to the CNS , appear to more closely recapitulate the presumed pathogenesis of fetal CNS infection with human cytomegalovirus . MCMV infection of the brain in these newborn mice results in a focal , non-necrotizing encephalitis with little evidence of specific cellular tropism but with global and symmetric deficits in brain development [53] . Altered development occurred in areas of the brain that exhibited no evidence of viral proteins or nucleic acids , suggesting that inflammatory responses to infection , and not direct effects of virus infection , were responsible for the altered development in the brain of neonatal animals [53] . To determine the potential role of host derived inflammation as a mechanism of disease in this model , we first needed to separate the linkage between virus replication and host inflammatory responses . This was accomplished by treating young animals with corticosteroids to limit host responses , and therefore inflammation , during virus infection . Although inflammation in MCMV infected animals was reduced at several levels , viral replication was unaffected . More importantly , the anti-inflammatory activity of corticosteroids attenuated the previously described developmental abnormalities in the cerebella of infected animals . This finding strongly argued that virus replication was not a direct cause of the developmental abnormalities within the CNS following MCMV infection and suggested that inflammatory responses played a major role in the disease phenotype [53] .
In an earlier report we described altered cerebellar development , including delayed cortical lamination , associated with MCMV infection of the CNS in newborn mice [53] . Disruption of lamination within the cerebellar cortex was frequently observed; however , altered lamination in areas immediately adjacent to virus infected cells was atypical in an overwhelming number of examined sections . Thus , histologic evidence of direct virus cytopathology as a cause of abnormal lamination of the cerebellar cortex was rare ( Figure 1A ) . The predominant histopathologic findings of this CNS infection were widely distributed foci of virus infected cells and surrounding mononuclear cells throughout the cerebrum and cerebellum [53] . In contrast to the focal nature of virus infection and mononuclear cell infiltration , defects in cerebellar morphogenesis were global and , most importantly , symmetric as illustrated by the delayed foliation and reduced cerebellar area in virus infected animals ( MCMV ) compared to uninfected ( control ) animals at post-natal day ( PND ) 8 ( Figure 1B ) . Notably , studies of infants infected in-utero by HCMV have also described global and symmetric deficits in brain morphogenesis without a significant component of focal or asymmetric loss of brain parenchyma , in the majority of documented cases [6] , [54]–[61] . From these findings , we have proposed that global alterations in cerebellar development are likely associated with soluble factors produced by the host inflammatory response and not related to direct effects of viral cytopathology . To characterize the nature of the inflammatory response in the cerebellum of infected animals , we analyzed several immunologic parameters in the brains of control and infected animals at PND8 . This time point was selected because virus replication in the CNS was established and deficits in cerebellar development were clearly observable [53] . Initially , we assayed the phenotype of CNS mononuclear cells in control and virus infected animals . Although CD8+ and CD4+ T-lymphocyte infiltrates , peripheral blood macrophages and activated microglia could be readily detected in the cerebellar parenchyma at PND14 , mononuclear cells were present in the CNS of MCMV infected mice by PND8 , prior to the detection of infiltrating T-lymphocytes [62] . Mononuclear cells isolated from control and infected brains were stained with two markers for tissue macrophages , F4/80 , a marker for cells of myeloid lineage and CD45 , a pan-leukocyte marker . The differential expression of CD45 by F4/80+ cells has been employed to distinguish between quiescent microglia ( low ) , activated microglia ( intermediate ) and infiltrating macrophages ( hi ) [63] . In control animals , F4/80+ cells expressing CD45hi/int were present in low abundance ( 3 . 0% ) ( Figure 2A ) . We observed an increase in the proportion of CD45hi/intF4/80+ cells in the CNS of infected mice ( 9% ) ( Figure 2A ) [62] . Furthermore , MHC class II expression was increased in this population of cells in MCMV infected mice , a finding consistent with the activation of these cells following infection ( Figure 2B ) . These results demonstrated an increase in the inflammatory response within the CNS , including increased activation of resident macrophages and recruitment of peripheral blood macrophages early in infection , prior to the appearance of virus specific CD8+ T-lymphocytes . To further define the activation state of brain macrophages in the CNS of MCMV infected mice , cerebellar sections from PND8 control and infected animals were stained with anti-Iba-1 , a marker for activated microglia/macrophages [64] , [65] . In sections from the cerebella of control mice , few Iba-1+ cells were observed ( Figure 2C ) . However , the number of Iba-1+ cells in the cerebellum was significantly increased following infection with MCMV ( Figure 2C ) . In addition , Iba-1 staining was observed in the meningial layer within the cerebellum of MCMV infected animals , suggesting an infiltration of cells from the periphery ( Figure 2C ) . Importantly , cellular infiltrates and activated mononuclear cells in the cerebellum were readily detected in the parenchyma of the cerebellum and not limited to foci of virus infected cells ( data not shown ) , suggesting that the generalized inflammation observed in the brains of MCMV infected mice was induced by soluble mediators produced in response to virus infection . Finally , we attempted to determine the frequency of Iba-1+ cells with an ameboid morphology suggestive of activated microglia and/or macrophages as compared to Iba-1+ cells with a ramified morphology consistent with quiescent or resting microglia/macrophages . We found cells consistent with both morphologies in infected and control animals but were unable to definitively assign differences in populations between the two experimental groups ( data not shown ) . Given the increase in the number of Iba-1+ cells and the increased activation of CD45hi/intF4/80+ cells , we next quantified the expression of inflammatory cytokines in virus-infected cerebella by quantitative real time PCR . We selected several proinflammatory cytokines , as well components of interferon induced responses ( IFIT1 and STAT1 ) , as markers for inflammation in the cerebella of infected animals . The expression of TNFα ( 10-fold ) , IFNβ ( 7-fold ) , STAT1 ( 10-fold ) and IFIT1 ( 175-fold ) were significantly increased in infected animals as compared to control animals ( Figure 2D ) . Together , these results demonstrated that by PND8 activated cells of the innate immune response and proinflammatory cytokines were present in the developing cerebellum of mice infected with MCMV as newborns . Thus far our findings suggested that soluble factors produced by the inflammatory response to virus infection in the CNS were responsible for the global alterations in cerebellar development . Endogenous glucocorticoids have been demonstrated to protect against immune-mediated pathology in MCMV infected adult mice , suggesting that treatment with glucocorticoids could alter the pathological changes in the CNS of MCMV infected newborn mice [66] , [67] . To examine the effects of glucocorticoid treatment on postnatal cerebellar development , control and MCMV infected mice were treated with dexamethasone ( dexa ) , a glucocorticoid with potent anti-inflammatory activity , which has been routinely used in the treatment of CNS inflammation in both clinical medicine and experimental animal models of human disease [17] , [68]–[72] . Control and MCMV infected newborn mice were treated daily with dexa or vehicle on PND4-6 and liver , spleen , brain and cerebellum were isolated from all groups on PND8 . There was no significant difference in the number of plaque forming units ( PFU ) of virus in the spleen , liver or brain of dexa treated/infected animals when compared to vehicle treated/infected animals , signifying that treatment with dexa had minimal effects on viral replication ( Figure 3A ) . We next assessed whether dexa treatment exhibited an anti-inflammatory effect following MCMV infection . Dexamethasone significantly reduced the frequency of CD45hi/intF4/80+ macrophages in the brains of infected mice compared to vehicle treated/infected mice ( Figure 3B ) . Interestingly , the frequency of CD45lo F4/80+ cells was reduced in the brains of MCMV infected mice as compared to control and dexa treated mice suggesting that the number of quiescent , or resting , microglia was decreased in infected animals , perhaps secondary to an increase in activated microglia in this experimental group ( Figure 3B ) . A reduction of MHC class II expression in this population was also observed in dexa treated/infected mice ( data not shown ) . Similarly , the number of Iba-1+ cells was significantly decreased in the cerebellum of dexa treated/infected mice compared to vehicle treated/infected mice ( Figure 3C ) . Consistent with the findings described above , the expression of IFIT1 was significantly decreased in the cerebellum of infected animals following treatment with dexa ( Figure 3D ) . We also determined that dexa treatment normalized the expression of IFIT2 and STAT1 in the cerebellum of MCMV infected mice ( Figure 3D ) . Together , these results demonstrated that dexa treatment decreased inflammation in the CNS of MCMV infected animals without significantly altering levels of virus replication . The finding that dexa treatment of MCMV infected mice significantly reduced the inflammatory response in the CNS raised the possibility that dexa treatment could also prevent the aberrant cerebellar development observed following infection . Dexamethasone treatment of infected mice normalized the expression of the developmentally regulated genes gli1 and N-myc ( both effectors of the sonic hedgehog ( SHH ) pathway responsible for granule neuron proliferation ) , as well as GABRA6 ( a marker for granule neuron differentiation ) and CDK5 ( primarily expressed in differentiated neurons ) ( Figure 4A ) [73]–[75] . Notably , some of these genes have previously been shown to be altered following MCMV infection [53] . However , dexa treatment of control animals also resulted in a significant reduction in the expression of both GABRA6 and CDK5 in the cerebellum when compared to control animals receiving only vehicle ( Figure 4A ) . These differences in expression were not due to an effect of dexa on transcription because the expression of Zic2 , a transcription factor expressed predominantly in granule neuron progenitors , was unaltered following treatment ( Figure 4A ) [76] . Morphometric measurements from the cerebella of infected mice demonstrated that the increased thickness of the EGL , previously associated with delayed migration of granule neuron progenitors , appeared to have been normalized following treatment with dexa ( data not shown ) . However , the EGL in dexa treated/control animals was decreased in thickness compared to vehicle treated/control animals ( data not shown ) . Dexamethasone treatment of control mice also resulted in a significant decrease in cerebellar area when compared to vehicle treated/control mice ( Figure 4B ) . In addition , the cerebellar area of dexa treated/infected animals was further decreased compared to vehicle treated/infected animals ( Figure 4B ) . Importantly , we did not observe a significant increase in activated caspase 3 staining in sections from these mice , indicating that increased apoptosis of granule neuron progenitor cells ( GNPCs ) did not contribute to the reduced size of the cerebellum in dexa treated/infected animals ( data not shown ) . These findings suggested that dexa treatment of MCMV infected mice resulted in significant off-target effects in cerebellar development , a result that would limit the interpretation of findings from our studies of cerebellar development in dexa treated animals . Similar off-targets effects of dexa on cerebellar development have been previously described and thought to be secondary to the anti-proliferative effects of this specific glucocorticoid on GNPCs [77] , [78] . Finally , our findings raised the possibility of an additive effect of dexa and MCMV infection on cerebellar development . The off-target effects of dexa on cerebellar development have been attributed to the resistance of this glucocorticoid to inactivation by 11β-hydroxysteroid dehydrogenase type 2 ( 11β-HSD2 ) , an enzyme which is highly expressed in the postnatal cerebellum in rodents as well as humans [78]–[80] . This enzyme is induced by SHH during development of GNPCs in the cerebellar cortex and appears to be protective in terms of limiting both the apoptotic and anti-proliferative effects of corticosteroids [78] , [79] , [81] , [82] . In contrast to dexa , other glucocorticoids such as hydrocortisone and prednisolone can be inactivated by 11β-HSD2 and have not been associated with the level of off-target effects observed following treatment with dexa [78] . Thus , we repeated the previous experiments using prednisolone ( pred ) , a corticosteroid with predominant glucocorticoid activity , which has also been used to attenuate inflammation associated with infections of the CNS , both in animal models and clinical medicine [69] , [83]–[86] . Control and MCMV infected newborn mice were treated once a day on PND4-7 with vehicle or pred . This time course of treatment was necessary secondary to the shorter in-vivo half-life of pred compared to dexa ( Figure 5A ) [87] , [88] . Initially , we determined the effects of pred treatment on virus replication in MCMV infected mice . We found no significant difference between the level of virus replication in the liver or brain of pred treated animals compared to vehicle treated/infected animals . However , minimal increases in viral genome copy number were observed in both the spleen and cerebellum of pred treated/infected animals ( Figure 5B ) . We next determined the effect of pred treatment on the frequency of Iba-1+ cells in the cerebellum of both uninfected and MCMV infected mice . As described previously , the number of Iba-1+ cells was increased in the cerebellum of MCMV infected mice compared to control mice ( Figure 5C ) . Following pred treatment , the frequency of Iba-1+ cells was reduced in MCMV infected animals ( 59% reduction ) compared to vehicle treated/infected animals ( Figure 5C ) . The number of Iba-1+ cells in pred treated/control animals was not significantly different from the number of positive cells in the cerebellum of vehicle treated/control animals ( Figure 5C ) . The observed reduction of Iba-1+ cells in the cerebellum of pred treated/infected mice indicated that pred decreased macrophage/microglia activation in the CNS of newborn mice following MCMV infection . In agreement with our previous findings , treatment of infected mice with pred also decreased the frequency of CD45hi/intF4/80+ macrophages in the CNS of infected mice ( Figure 5D ) . Treatment of control animals with pred had no significant effect on either the CD45hi/intF4/80+ macrophage population or the CD45lo F4/80+ resting microglial population ( Figure 5D ) . The observed reduction of Iba-1+ cells in the cerebellum and the decreased percentage of CD45hi/intF4/80+ macrophages in the CNS of pred treated/infected mice indicated that pred decreased the number of activated macrophage/microglia in the CNS of newborn mice following MCMV infection . We next determined the effects of pred treatment on the expression of proinflammatory cytokines previously shown to be elevated in the cerebellum following MCMV infection ( Figure 2D ) . Consistent with the findings described above , we observed a reduction in the transcription of TNFα ( 25% ) , IFNβ ( 70% ) and IFIT1 ( 65% ) within the cerebellum of MCMV infected mice treated with pred ( Figure 6A ) . Pred treatment also decreased cytokine levels of IFNβ ( 25% ) and IFNγ ( 43% ) within the cerebellum ( Figure 6B ) . Interestingly , cytokine levels of TNFα were not affected following pred treatment . These results illustrated that treatment with pred could attenuate MCMV induced inflammation in the CNS independent of changes in virus replication , thereby uncoupling the level of virus replication and the host inflammatory response within the cerebellum . Since treatment with pred significantly reduced the inflammatory response in the CNS and has been reported to lack the off-target effects observed with dexa , we next determined if pred treatment could also limit the abnormal development of the cerebellum that was observed in MCMV infected animals . Because of the large number of mice used in these experiments , the variation in animal size and the size dependent variation in brain area , we normalized measurements of cerebellar area between experimental groups by expressing cerebellar area as a percentage of brain area . The ratio of cerebellar area/brain area was found to be similar for pred treated/control and pred treated/infected animals when compared to vehicle treated/control animals; however , vehicle treated MCMV infected mice showed a significant reduction in this ratio ( Figure 7A ) . These results confirmed the decrease in cerebellar area previously observed following infection with MCMV and , more importantly , demonstrated normalization of altered cerebellar size in infected mice by treatment with pred . These findings were consistent with our hypothesis that inflammatory mediators , released in response to MCMV infection , were a primary cause of altered cerebellar development . In addition to the decrease in cerebellar area , we have previously documented an increase in the thickness of the EGL in MCMV infected animals [53] . Since treatment of infected mice with pred lead to normalization of cerebellar area , we next determined whether this treatment would also normalize the increased thickness of the EGL following infection . As expected , the EGL was thicker in MCMV infected mice compared to control mice . This abnormality in cerebellar development was corrected in infected mice following treatment with pred ( Figure 7B , D ) . There was no measureable difference in the thickness of the EGL in control animals treated with pred compared to vehicle treated/control animals ( Figure 7B , D ) . To determine if the increase in the thickness of the EGL following infection was secondary to an increase in cellularity , the number of GNPCs in the EGL was quantified . Consistent with an increase in thickness , we found an increase in the number of GNPCs within the EGL following infection ( Figure 7C ) . Concomitant with normalizing the increased thickness of the EGL , treatment of infected mice with pred also normalized the number of GNPCs within the EGL . We did not find any significant difference in the number of GNPCs in the EGL between vehicle treated/control animals or pred treated/control animals ( Figure 7C ) . The normalization of MCMV induced abnormalities in the morphogenesis of the cerebellar cortex following treatment with pred demonstrated that we could limit morphogenic abnormalities within the cerebellum of infected mice by modulating inflammatory responses . Previously , we documented that following infection , morphological deficits within the cerebellum coincided with a significant reduction in the transcription of developmentally regulated genes expressed within GNPCs [53] . Since pred treatment reduced inflammation and corrected morphological deficits within the cerebellum of infected mice , we hypothesized that pred treatment could also correct abnormalities in the transcription of these genes . Similar to our studies using dexa , we assayed gli1 , N-myc , GABRA6 and CDK5 expression in the cerebella of uninfected and MCMV infected mice treated with vehicle or pred . Consistent with our previous findings , expression of both GABRA6 and CDK5 was decreased following infection with MCMV when compared to control mice ( Figure 7E ) . Following treatment with pred the expression of both genes was normalized within the cerebella of MCMV infected mice . Similarly , the transcription of gli1 and N-myc was elevated in the cerebellum following infection and treatment of infected mice with pred decreased the expression of both genes ( Figure 7E ) . Importantly , pred treatment had no effect on the transcription of gli1 , N-myc , GABRA6 or CDK5 in control animals ( Figure 7E ) . As a control , the expression of Zic2 was analyzed and was found to be similar in the cerebella of all groups ( Figure 7E ) [76] . These results indicated that decreasing inflammation in MCMV infected animals by treatment with pred normalized the expression of developmentally regulated genes in the absence of measurable off-target effects . In MCMV infected mice , the upregulation of gli1 and N-myc was inconsistent with the deficit in GNPC proliferation observed in our previous studies [53] . This suggests that an alternative mechanism could be responsible for the deficit in GNPC proliferation within the cerebellum of infected mice [53] . Given our previous findings ( increased thickness of the EGL , decreased GNPC differentiation , decreased GNPC migration to the IGL and decreased thickness of the IGL ) , we postulated that a block or delay within the GNPC cell cycle , downstream from the actions of gli1 and N-myc , would be most consistent with our observations . The failure of GNPCs to complete a program of proliferation in the EGL would prevent their differentiation and subsequent migration into the IGL . This mechanism would also account for the increased cellularity of the EGL and the decreased cellularity of the IGL in infected animals [53] , [89]–[91] . To investigate this possibility , PND8 animals were injected with BrdU , a marker of cells in S phase . Serial sections from the cerebellum were stained with antibodies reactive with BrdU and Ki67 , a marker of cycling cells , and the number of positive cells was quantified for each marker ( Figure 8A ) . No difference was observed in the percent of total GNPCs that were positive for Ki67 in the EGL of MCMV infected animals compared to control animals ( Figure 8B ) . However , a decrease in the percent of cycling cells ( Ki67+ ) positive for BrdU was detected in infected animals when compared to control animals ( Figure 8C ) . The decrease in BrdU reactivity within GNPCs of infected mice was therefore not secondary to a decrease in the overall number of cells in the cell cycle . Moreover , the previously described minimal level of apoptosis of GNPCs in either group of animals indicated that there is likely a block or delay in the cell cycle of GNPCs following infection [53] . If the inflammatory response in the CNS of infected mice contributed to the block/delay in the proliferation of GNPCs , our results described above would argue that the anti-inflammatory effects of pred could alleviate this block and restore the proliferative capacity of GNPCs in the EGL . Analysis of Ki67 expression in pred treated groups revealed that the percent of GNPCs in the cell cycle was similar to that of infected or control animals that were treated with vehicle ( Figure 8B ) . When compared to vehicle treated/control animals there was no significant difference in the percent of BrdU+ cells in EGL of pred treated/infected animals indicating that pred treatment of infected animals normalized the deficit in GNPC proliferation associated with MCMV infection ( Figure 8C ) . Importantly , the percent of BrdU+ GNPCs in the EGL of pred treated/control animals was not significantly different from vehicle treated/control animals . Together , these findings argue that pred treatment alleviated alterations in the cell cycle of GNPCs that were associated with MCMV infection . Furthermore , these results support our hypothesis that modulating the inflammatory response following MCMV infection could limit deficits in cerebellar morphogenesis , likely through reversing the delay in GNPC proliferation . To further define the disruption in the cell-cycle of GNPCs following infection we assayed the levels of two cyclins , cyclin D1 and cyclin B1 , in control and MCMV infected mice . Levels of cyclin D1 were not significantly different between control or infected animals , suggesting that infection with MCMV did not alter the signals associated with entry of GNPCs into G1 ( data not shown ) [92] . Similarly , pred treatment did not alter cyclin D1 levels in infected or control animals ( data not shown ) . Although there was no observable difference in the levels of total cyclin B1 expression between infected and control mice ( Figure 8D ) , the level of phosphorylated-cyclin ( p-cyclin ) B1 , a marker for G2/M , was decreased within the cerebella of infected animals compared to control animals ( Figure 8D ) [93] , [94] . Together with the decreased number of BrdU+ GNPCs , this data further argued for a block/delay in the cell cycle following infection . Cerebella from both control and MCMV infected mice treated with pred displayed levels of p-cyclin B1 that were similar to vehicle treated/control mice ( Figure 8D ) . Although this data did not reveal the precise point where cell cycle progression was delayed , it further confirmed that altered development of the cerebellum in infected animals was associated with delayed proliferation of GNPCs within the EGL . Treatment with pred corrected this deficit and normalized the morphological abnormalities within the cerebellum following infection . These results were consistent with a mechanism in which the developmental abnormalities associated with focal encephalitis in MCMV infected newborn mice resulted from the host inflammatory response as opposed to a direct virus-mediated mechanism .
Previously we have shown that intraperitoneal inoculation of newborn mice with MCMV resulted in a focal CNS infection that involved all regions of the brain that but did not exhibit specific cellular tropism [53] . Histologically , the foci consisted of a small number of virus-infected cells , mononuclear cells and reactive astroglial cells [53] . Although there was no observable difference in the size of the cerebrum between infected and uninfected animals , cerebellar hypoplasia was readily apparent in infected animals and was associated with delayed foliation and decreased area of the cerebellar cortex , findings attributable to the decreased proliferation of GNPCs within the EGL [53] . Morphogenic abnormalities of the cerebellar cortex included increased thickness of the EGL , decreased thickness of the IGL , abnormal arborization of Purkinje neuron dendrites and thinning of the molecular layer [53] . Interestingly , the altered morphogenesis of the cerebellum was symmetric even though foci containing virus infected resident cells and infiltrating mononuclear cells were scattered widely throughout the parenchyma of the cerebellum . These later findings strongly argued that the developmental abnormalities were secondary to a soluble mediator generated during virus-induced inflammatory responses in the CNS and not from direct cytopathic effects of virus infection . In this report , we have described findings consistent with this mechanism; specifically , evidence that attenuation of inflammatory responses in infected mice , by treatment with anti-inflammatory glucocorticoids , normalized developmental abnormalities in the cerebellum without affecting the level of virus replication . Our results demonstrated that several measures of GNPC proliferation were altered in MCMV infected mice , including a decrease in the frequency of cells in S phase and a decrease in the levels of phospho-cyclin B1 within the EGL of MCMV infected mice . Several explanations could account for these findings , including a decrease in the number of GNPCs entering the cell cycle , premature exit of GNPCs from the cell cycle and a block or delay in the cell cycle of GNPCs following infection . Premature exit of GNPCs from the cell cycle represented an obvious explanation for the decreased cerebellar size but other measures of GNPC proliferation were inconsistent with this explanation . The increased cellularity of the EGL following MCMV infection and the similar percentages of Ki67+ GNPCs in infected and control mice argued that there was no difference in the number of GNPCs entering the cell cycle nor was there an increased number of GNPCs exiting the cell cycle . Because we found a decrease in certain markers of proliferation but no change in the number of cycling GNPCs following MCMV infection in this study as well as in a previous study , a more consistent interpretation of our data is that the cell cycle of GNPCs in the EGL is prolonged in MCMV infected animals [53] . Prolongation of the cell cycle could delay the completion of the programmed proliferation and subsequent differentiation of GNPCs that is required for normal morphogenesis of the cerebellar cortex . Variation in the rate of cell division of GNPCs in the EGL has been described , suggesting that the duration of the cell cycle in these cells is not autonomous and can be influenced by extracellular cues [90] , [95] , [96] . Though we have not fully characterized the nature of this alteration in the cell cycle of GNPCs , it was reversible , in that the delay was corrected when MCMV infected animals were treated with glucocorticoids . Although a unifying mechanism for the normalization of cerebellar development in pred treated MCMV infected mice remains incompletely described , our results were most consistent with a decrease in the inflammatory response in the CNS leading to normalization of the proliferative capacity of GNPCs in the cerebellar cortex . This mechanism is based on previous studies that have demonstrated that GNPCs undergo what is thought to be a programmed number of cell divisions prior to exiting the cell cycle , entering a differentiation program and then migrating from the EGL into deeper layers of the cerebellar cortex [90] , [95] , [96] . This well choreographed developmental pathway has been extensively studied and many of the molecular signals associated with this pathway have been described [74] , [91] , [95]–[100] . We are proposing that if the cell cycle of GNPCs is prolonged , subsequent to inflammation in the cerebellum , then normal morphogenesis of the cerebellar cortex fails to take place and the expression of developmentally regulated genes that depend on differentiation and correct cellular positioning will be delayed . Findings from this study are consistent with a reversible , generalized slowing of the GNPC cell cycle in infected mice . Reversal of this slowing could be expected to result in a rebound in GNPC proliferation , permitting the completion of the developmentally programmed cell divisions , differentiation into migrating granule neurons , migration into the IGL and expression of the associated differentiation genes . The reversibility of this mechanism is consistent with the partial resolution of defects in cerebellar development observed in vehicle treated MCMV infected mice following virus clearance and regulation of the inflammatory response later in infection [53] . Additional support for the reversibility of a slowing of the cell cycle has been reported in a study of 11β-HSD2 −/− transgenic mice treated with corticosterone [79] . Findings from this study demonstrated a rebound in the cerebellar area and the size of the IGL in these transgenic mice following withdrawal of steroid treatment [79] . Even though the effector molecules and pathways that lead to altered proliferation of GNPCs and cerebellar development in this model of a human CNS infection remain undefined , such a mechanism could argue for a common pathway leading to the developmental abnormalities associated with inflammation following infection of the developing brain of the fetus and newborn infant by a number of microbial agents . Alteration in the rate of proliferation of progenitor cells in the developing CNS could lead to deficits in developmental , stage dependent cell positioning and potentially result in a number of long term neurological abnormalities . A recent study that carefully detailed the effects of glucocorticoids on the developing cerebellum described several phenotypes following treatment with different glucocorticoids [78] . These investigators demonstrated that the phenotypic response of GNPCs to glucocorticoids was dependent on the presence of 11β-HSD2 , an enzyme that is expressed at higher levels in the cerebella of both newborn rodents and humans as compared to other regions of the CNS [78]–[80] , [101] . Previous studies have indicated that the inactivation of glucocorticoids by 11β-HSD2 limits the anti-proliferative and apoptotic inducing activities of endogenous and exogenous glucocorticoids [78] , [79] . Because dexamethasone ( dexa ) is not efficiently inactivated by 11β-HSD2 , treatment of neonatal mice with dexa resulted in increased GNPC apoptosis ( short term treatment ) or decreased GNPC proliferation ( chronic treatment ) , secondary to exit from the cell cycle presumably from accelerated GNPC differentiation [77] , [78] . Interestingly , in this study chronic prednisolone ( pred ) treatment resulted in an intermediate phenotype due to the inactivation of this specific glucocorticoid by 11β-HSD2 [78] . Our findings were consistent with the results presented in this report in that treatment with dexa , but not pred , resulted in a significant decrease in the size of the cerebellar cortex in both uninfected and infected mice . We also noted that in two independent experiments the cerebellar area in dexa treated/infected mice was smaller than that of both dexa treated/control mice or vehicle treated MCMV infected mice . These findings suggested that the effects of dexa and MCMV infection were additive and raised the possibility that the effect of dexa on GNPC proliferation in this setting differed from those that followed MCMV infection . Interestingly , dexa treatment did result in normalization of the expression of genes associated with GNPC differentiation ( GABRA6 and CDK5 ) in the absence of normalization of GNPC proliferation , a finding consistent with accelerated GNPC differentiation in animals following treatment with dexa [77] , [78] . The premature exit of GNPCs from the cell cycle likely accounted for the cerebellar hypoplasia and decreased cerebellar area that was observed in dexa treated animals . In contrast , when infected mice were treated chronically with pred , we observed a correction of the abnormal cell cycle of GNPCs that was also associated with normalization of the morphogenic abnormalities in the cerebellar cortex . Following normalization of the cell cycle in pred treated animals , GNPCs completed their programmed proliferation in the EGL , migrated into the deeper layers of the cerebellum and expressed development specific genes . We have not identified a specific mechanism ( s ) to explain the correction of proliferation deficit ( s ) in GNPCs following pred treatment , but it is unlikely that in pred treated mice , GNPCs exited the cell cycle and differentiated as was observed in dexa treated mice . This argument is based on three findings; ( i ) a similar frequency of GNPCs were cycling in both pred treated and vehicle treated mice , ( ii ) the frequency of BrdU+ GNPCs in the EGL was increased following pred treatment and ( iii ) measures of cerebellar morphogenesis ( EGL thickness , cerebella area and EGL cellularity ) were normalized in infected mice following treatment with pred . Several experimental models of CNS infection in newborn animals have also noted beneficial outcomes following treatment with anti-inflammatory agents , but in some cases and in contrast to our findings , increased disease severity secondary to increased replication of the microorganism was also observed [17] , [67] , [68] . Experimental rodent models of herpes simplex encephalitis have demonstrated a beneficial effect of steroid treatment when combined with an antiviral agent suggesting that host-derived inflammation contributes to the outcome of CNS infection with this virus [102] , [103] . In findings that paralleled our results , treatment of Borna disease virus ( BDV ) infected adult rats with dexa limited inflammation and also appeared to improve neurologic function in infected animals [17] . In clinical medicine , the use of glucocorticoids to limit CNS inflammation in patients with mycobacterial infections of the brain is well established [69] , [70] . These agents have also been utilized to limit neurological sequelae that follow bacterial meninigitis associated with pyogenic bacteria [71] . Several studies have demonstrated that glucocorticoids efficiently limit the innate immune response to microorganisms in the CNS , including the expression of proinflammatory cytokines , chemokines and interferon stimulated genes [17] , [104] . However , the use of glucocorticoids , particularly dexa , in young infants remains controversial because of the well documented adverse effects this agent has on brain development [105] , [106] . The importance of SHH in the proliferation of GNPCs in the cerebellar cortex has been studied extensively [107]–[113] . The proliferation of these neuron progenitors in response to SHH has been reported to involve the transcription factors gli1 and N-myc [109] , [114]–[117] . It was therefore somewhat unexpected to find that expression of both gli1 and N-myc was increased in the cerebella of MCMV infected mice as compared to control mice . Interestingly , we noted that transcription of patched ( Ptch ) was also increased in the cerebella of MCMV infected mice , a finding that paralleled the increased expression of gli1 and could represent a regulatory response to SHH induced responses [118] , [119] . We do not have a definitive explanation for the increase in gli1 and N-myc expression but noted that when MCMV infected mice were treated with glucocorticoids the expression of these SHH effectors was normalized . Consistent with our observations , previous reports have suggested that proinflammatory cytokines can modulate the SHH pathway [120] , [121] . As an example , increases in GNPC proliferation have been documented in transgenic mice with constitutive IFNγ expression in the CNS [122] . In these engineered mice , SHH and gli1 expression was induced by IFNγ via a STAT1 dependent pathway . More recent studies have reported that IFNγ treatment of cultured granule neurons leads to increased proliferation and that STAT1 binds directly to the SHH promoter [123] , [124] . Interestingly , both IFNγ and STAT1 were upregulated in the cerebella of MCMV infected mice coincident with an increase in the expression of N-myc and gli1 ( Figure 3D; Figure 6B ) . Moreover , treatment with pred reduced the cytokine levels of IFNγ and normalized the expression of both N-myc and gli1 following MCMV infection . Studies of cytokines during CNS development have detailed both neuroprotective and deleterious roles , suggesting a delicate balance between the homeostatic and immune functions of cytokines in the developing CNS [125]–[128] . Our findings suggest that cytokines released following neonatal infection with MCMV could have deleterious effects on developing GNPCs within the cerebellum and that modulating the inflammatory response associated with this infection could limit damage to the developing CNS . An important aspect of this study is that the pathological and histopathological findings in this murine model appear very similar to those reported in human infants with congenital CMV infection . The focal encephalitis , characteristic of MCMV infection in mice , has also been noted in autopsy findings from infants with congenital HCMV infections . Furthermore , in this model histopathological findings of mononuclear cell infiltrates and reactive gliosis , termed micronodular gliosis , are remarkably similar to those found in infected human infants [55] , [57] , [60] , [129] , [130] . Cerebellar hypoplasia is an invariant finding in this murine model and also frequently reported in infants with congenital HCMV infections that have been studied by imaging or , in a smaller number , following autopsy [57] , [131] , [132] . Reports describing MRI findings in infants with congenital HCMV infection have suggested that cerebellar hypoplasia is characteristic of this intrauterine infection . However , it should also be noted that the murine model we have developed has a significant limitation , dictated by the route of virus inoculation and the age of the developing brain at the time of infection . CNS development in newborn mice is believed to be at a stage similar to that of a mid to late 2nd trimester human fetus . Thus , in the murine model we have developed , cortical damage associated with an earlier gestational age of fetal infection will not be adequately modeled . Yet it is also important to note that the vast majority of infants with congenital HCMV infections also do not exhibit structural damage to the cerebral cortex , raising the possibility that only a minority of infants are infected early in gestation . In agreement with this possibility , recent studies have provided evidence suggesting that transmission of virus to the developing fetus occurs more frequently in the later stages of pregnancy [133] . Thus , with the awareness of limitations inherent in studies carried out in rodents , we would argue that the findings we have generated from our studies suggest that inflammation in the developing brain should be considered a potential contributor to at least some of the developmental abnormalities that have been associated with intrauterine HCMV infections . Furthermore , if inflammation and the soluble mediators present in the CNS account for the altered proliferative capacity of neural progenitor cells , our results could be extrapolated as a potential explanation for maldevelopment of the brain associated with other intrauterine infections resulting in CNS inflammation . Even though our findings in this murine model of congenital CMV infection have demonstrated a beneficial effect of glucocorticoid therapy in maintaining the developmental program during MCMV infection , we cannot directly extrapolate our findings in this model system to human disease or other infections of the CNS . However , the potential intersections between neurodevelopmental pathways and those that contribute to CNS inflammation in neonatal animals would suggest that more selective approaches to limiting CNS inflammation could open new therapeutic avenues and lead to improved outcomes . These approaches combined with antiviral therapy , to limit virus replication until host responses can efficiently clear virus from the CNS , could offer a more optimal approach for management of this important perinatal infection . Further exploitation of this model could provide insight into the feasibility of such an approach and perhaps aide in defining markers of CNS inflammation , allowing for a more selective introduction of anti-inflammatory therapy .
All animal breeding and experiments were performed in accordance to the guidelines of the University of Alabama – Birmingham Institutional Animal Care and Use Committee ( IACUC ) in strict compliance with guidelines set forth by the NIH ( OLAW Assurance Number - A3255-01 ) . Research was conducted under a protocol approved by IACUC . All experiments done at the University of Rijeka were in accordance with the University of Rijeka – Croatia animal use and care policies in accordance to the guidelines of the animal experimentation law ( SR 455 . 163; TVV ) of the Swiss Federal Government . Infection of mouse pups was performed as previously described [53] . Briefly , newborn Balb/c mice ( 6–18 hrs post-partum ) were infected with 500 PFU of MCMV-Smith ( ATCC VR-1399 ) by i . p . ( intraperitoneal ) inoculation . Control and MCMV infected pups were treated on PND4-6 by i . p . injection with dexamethasone sodium phosphate ( dexa; APP Pharmaceuticals ) ; 1 mg/kg in 50 µl of sterile PBS . Dexa was administered once a day and mice were sacrificed on PND8 between 36 and 42 hours after the last treatment was administered . For Prednisolone experiments , animals were treated with prednisolone sodium phosphate ( pred; commercial pharmacy ) ; 7 mg/kg ( equivalent to 1 mg/kg dexa ) in 50 µl of sterile PBS on PND4-7 . Treatments were administered once a day and mice were sacrificed on PND8 between 16 and 18 hrs post injection . As a control , uninfected and MCMV infected animals were given i . p . injections with 50 µl sterile PBS alone ( vehicle ) . Animals were sacrificed on PND8 , perfused with ice cold PBS and organs were harvested and processed for the appropriate downstream application . All mice were purchased from The Jackson Laboratory ( Bar Harbor , ME ) . Stocks of MCMV-Smith strain were propagated by infection of mouse embryonic fibroblasts ( MEFs ) . Infected media was harvested at 5–7 days post-infection and frozen at −80°C . For dexa experiments , organs were collected , weighed and homogenized . A 10% homogenate in media was utilized for standard plaque assays [134] . For pred experiments , organs were collected and DNA was isolated using Trizol according to the manufacturer's instructions ( Roche Applied Science ) . 1 µg of DNA was then used for quantitative real-time PCR with the following primers for MCMV IE-1 Exon 4: Forward: 5′-GGC TTC ATG ATC CAC CCT GTT A – 3′; Reverse: 5′-GCC TTC ATC TGC TGC CAT ACT – 3′ . Primers were used at a concentration of 250 nM/reaction . The following FAM-TAMRA ( BHQ-2 ) probe was used at a concentration of 300 nM/reaction for real-time detection: 5′-/56-FAM/AGC CTT TCC TGG ATG CCA GGT CTC A – 3′ . Real time PCR was performed by Taqman based assay using the StepOne Plus system from Applied Biosystems ( Carlsbad , CA ) . For immunofluorescence studies , mice were injected on PND8 with 50 µg/g of BrdU ( Sigma Aldrich ) in 1× PBS , 6 hrs . prior to harvest . Mice were then perfused with PBS and brains were fixed in 4% paraformaldehyde ( PFA ) overnight , cryoprotected in 30% sucrose-PBS and embedded in Tissue Tek O . C . T . compound ( Andwin Scientific ) . 8-µm sagittal sections were cut using a Leica cryostat . Cut sections were dried for 4 hours at room temperature ( RT ) , rehydrated in 1× PBS then used for immunofluorescence assays . For Iba-1 staining , sections were blocked in 1× PBS , . 05% Triton X-100 , 20% normal goat serum , 5% BSA for 2 hr . at RT . Sections were then stained with anti-Iba-1 overnight at 4°C . Subsequently , sections were washed with PBS , . 05% Triton X-100 and then incubated for 2 hrs . at RT in the dark with secondary antibody , followed by a 15 min . incubation with TOPRO-3 iodide ( 1∶1000 , Molecular Probes ) at RT . Following staining for Iba-1 , sections were post-fixed with 2% PFA for 20 min . at RT . Sections were washed and mounted using Vectashield Fluorescent mounting medium ( Vector Laboratories ) . For BrdU/Ki67 , sections were blocked in 1× PBS , 1% Triton X-100 , 20% normal goat serum , 1 M glycine , 5% BSA for 1 hr . at RT . Blocking was followed by a 2 N HCL acid wash for 10 min . on ice , 10 min . at RT and 20 min at 37°C . Sections were then buffered in . 1 M Borate buffer for 12 min . at RT , washed in PBS , 1% Triton X-100 and labeled as previously described . Primary antibodies utilized in this study were anti-Iba-1 ( 1∶200 , Wako , Japan ) , anti-Ki67 ( 1∶200 , ab66155; Abcam ) , anti-IE1 ( Chroma101 [53] ) and anti-BrdU ( 1∶50 , ab6326; Abcam ) . Secondary antibodies used were: Alexa Fluor 594 - conjugated anti-Rabbit; Alexa Fluor 488 - conjugated anti-mouse ( Molecular Probes ) and Goat anti-Rat – FITC ( Southern Biotech ) , respectively . Images of stained sections were collected by using an Olympus Fluoview confocal microscope ( 20× objective for Iba-1 and 60× objective for BrdU/Ki67 ) . For cell counts , images were saved as TIFF files and opened in Image J [135] . An area box was created and the number of cells in the EGL within this box was counted for each section . Frozen sections were used for all morphometric measurements . EGL measurements were done on serial sections using Image J software . Measurements were obtained from sections stained with BrdU , Ki67 and TOPRO3 . Images were collected with a confocal microscope . 4 measurements were taken from the primary fissure of the EGL in each section and 8 serial sections were measured per animal . For area measurements , the first 5 sections in each series were stained with 1% cresyl violet in ethanol for 10 min . followed by washing with 1× PBS until dye no longer ran off . Sections were mounted with 50% glycerol , 50% PBS and pictures were taken using an Olympus BX41 microscope with a 2× objective . Representative sections showing a close up of the cerebellum used in the paper were obtained with a 4× objective . Cerebellar area and brain area was measured using Image J software [135] . CNS mononuclear cells were isolated by using a percoll density gradient protocol [62] . Isolated cell populations were stained in FACS buffer ( 2% BSA and 0 . 2% sodium azide ) for 30 min at 4°C in the dark and fixed in 2% PFA . All samples were stained with CD45-FITC and F4/80-APC ( eBioscience ) and MHCII-IA/IE ( Biolegend ) . Samples were acquired using a FACSCalibur ( BD Biosciences ) flow cytometer and analyzed using FlowJo7 . 6 . 1 . Due to low cell number and poor cell viability , mononuclear cell isolations from neonatal brain was performed as follows for prednisolone treated groups . Brains were homogenized using a GentleMACs tissue homogenizer ( Milteniy Biotech ) . Homogenates were strained through a 40 µm nylon strainer , followed by centrifugation at 400×g for 4 min at 4°C . Homogenates were washed once with 1×PBS ( without Ca++/Mg++ ) and centrifuged again at 400×g , 4 min at 4°C . Mononuclear cells were isolated by resuspending the pellet in a 37% continuous Percoll gradient followed by centrifugation at 690×g for 20 min , 4°C with gentle braking . Pellets were washed once with FACS buffer ( 1×PBS , 2% BSA , . 2% Sodium Azide ) , then lysed for 5 min with 1 ml RBC lysis buffer ( Sigma Aldrich ) . Lysis was inhibited by adding 10 mls FACS buffer and the pellet was collected by centrifugation ( 400×g , 4 min at 4°C ) . Pellets were again washed with FACS buffer , followed by resuspension in FACS buffer with FC block ( 1∶100 , eBioscience ) . Mononuclear cells were blocked for 30 min on ice , counted using a TC20 cell counter ( Bio-Rad ) and 100 µl of cell suspension was transferred to individual wells of a round bottom , polystyrene 96 well plate . 100 µl of FACS buffer was added to each well and the plate was centrifuged ( 400×g , 4 min at 4°C ) to pellet the cells . Mononuclear pellets were washed 2× with FACS buffer , followed by staining with CD45 – PerCP ( 1∶300 ) , Cd11b – PE ( 1∶200 ) and F480 – FITC ( 1∶300 ) ( eBioscience ) for 1 hr at 4°C in the dark . Following staining , 150 µl of FACS buffer was added to each well and cells were pelleted by centrifugation . Cells were again washed 2× with FACS buffer followed by fixation with 4% PFA for 20 min at 4°C in the dark . Following fixation , cells were washed 2× with FACS buffer , resuspended in 200 µl FACS buffer and transferred to 5 ml polystyrene FACS tubes ( BD Falcon ) . Samples were acquired using a FACSCalibur ( BD Biosciences ) flow cytometer and analyzed using FlowJo7 . 6 . 1 . Dexamethasone experiments were repeated using this protocol and data were compared to the previous protocol . No differences were observed in the frequency of CD45lo or CD45hi/int/F480+ mononuclear cell populations in any group when compared to our previous findings; however , mononuclear cell numbers were greatly improved . Total cerebellar RNA from control and experimental mice was isolated using Trizol reagent ( Roche Applied Science ) ; 500 µl Trizol/cerebellum according to manufacturer's protocol . cDNA from each sample was synthesized using the Superscript III First Strand synthesis kit ( Invitrogen ) . Taqman based real time PCR was employed for determining the mRNA expression of genes of interest in experimental animals relative to uninfected controls . Taqman assay mixes for TNF-α ( Mm99999068 ) , IFN-β ( Mm00439552 ) , STAT1 ( Mm00439518 ) , IFN-γ ( Mm99999071 ) , gli1 ( Mm00494645 ) , N-myc ( Mm00476449 ) , Zic2 ( Mm01226725 ) , CDK5 ( Mm00432437 ) and GABRA6 ( Mm01227754 ) were obtained from Applied Biosystems . Real time PCR was performed using the StepOne Plus system from Applied Biosystems . The housekeeping gene 18S was used as a control for all experiments . The fold change ( target gene expression relative to 18S ) for control animals was set to a value of 1 +/− SEM and the relative fold change for each experimental group was determined by normalizing to control animals . Cerebella were harvest from PND8 animals . Samples were pooled ( 3 cerebella/sample ) and homogenized in ELISA buffer ( 1×PBS , . 25% Triton X-100 ) containing protease/phosphatase inhibitors ( Thermo Scientific ) . Lysates were rotated for 20 min at 4°C then sonicated 3× for 5 sec , followed by centrifugation at 12K× g for 10 min at 4°C . Aliquots were made and stored at −80°C until use . ELISAs were performed according to the manufacturer's instructions: TNFα ( eBioscience ) , high sensitivity IFNγ ( ebioscience , San Diego , CA ) and IFNβ ( PBL Interferon Source ) . Cytokine concentrations ( pg/ml ) were normalized for amount of tissue used ( mg ) . Cerebella harvested from control and experimental groups at PND8 were homogenized in RIPA buffer ( 50 mM Tris-HCl , NaCl 150 mM , 1% NP-40 , 0 . 25% Na-Deoxycholate , 1 mM EDTA ) containing protease/phosphatase inhibitors ( Thermo Scientific ) and cleared of insoluble material by centrifugation at 12K× g . 50 µg of protein solubilized in sample buffer ( 5% SDS , 2% 2-mercaptoethanol , Tris pH 8 ) and separated by SDS-PAGE electrophoresis using a 10% acrylamide gel . Electrophoretically separated proteins were immobilized on nitrocellulose membranes and used for Western blot analysis . Membranes were probed overnight at 4°C for actin ( 1∶1000 , MAB1501; Millipore ) , cyclin D1 , cyclin B1 and phospho-Cyclin B1 ( Ser 147 ) ( 1∶500 , 2978 , 4138 and 4131 respectively; Cell Signaling Technology ) . Immunoblots were incubated for 1 hr with HRP-conjugated anti-mouse or anti-rabbit secondary antibodies ( Southern Biotech ) then developed with ECL reagent ( Perkin Elmer ) . Densitometry was performed using Quantity One software ( Bio-Rad ) and levels of protein were normalized to actin for each lane . Statistical significance of comparisons of mean values was assessed by a two-tailed Student's t test , one-way analysis of variance ( ANOVA ) followed by Bonfferronni's multiple comparison test , two-way ANOVA followed by Bonfferronni's posttest , or a Mann-Whitney test using Prism 4 software ( GraphPad ) .
|
Intrauterine infection with human cytomegalovirus ( HCMV ) is a leading cause of developmental brain damage . In the U . S . , an estimated 2 , 000 infants a year develop brain damage as a result of intrauterine infection with HCMV . In this study , we examined the contribution of host immune responses induced by CMV infection to abnormal development of the CNS by treating neonatal mice infected with MCMV with glucocorticoids . We found that glucocorticoid treatment of infected mice decreased the inflammatory response within the CNS without altering the level of virus replication . In addition , abnormalities in the structure of the cerebellum , as well as abnormalities in granule neuron precursor cell proliferation were normalized in MCMV infected mice following glucocorticoid treatment . These studies suggest that the host immune response to CMV infection is damaging to the developing CNS and that it may be possible to limit CNS disease by modulating inflammation . Moreover , understanding how inflammation and the immune response may alter the developmental program within the CNS could offer important insight into the mechanisms of disease leading to abnormal brain development following intrauterine infection .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"neurogenesis",
"immunologic",
"subspecialties",
"immunology",
"microbiology",
"host-pathogen",
"interaction",
"cytomegalovirus",
"infection",
"neuroscience",
"cell",
"differentiation",
"developmental",
"biology",
"stem",
"cells",
"emerging",
"infectious",
"diseases",
"immune",
"defense",
"immunoregulation",
"infectious",
"disease",
"control",
"morphogenesis",
"immunomodulation",
"immunotherapy",
"animal",
"models",
"of",
"infection",
"infectious",
"diseases",
"developmental",
"neuroscience",
"infectious",
"diseases",
"of",
"the",
"nervous",
"system",
"inflammation",
"biology",
"immune",
"response",
"immunopathology",
"neural",
"stem",
"cells",
"clinical",
"immunology",
"immunity",
"virology",
"viral",
"diseases",
"neuroimmunology"
] |
2013
|
Glucocortiocoid Treatment of MCMV Infected Newborn Mice Attenuates CNS Inflammation and Limits Deficits in Cerebellar Development
|
Natural progression of HIV-1 infection depends on genetic variation in the human major histocompatibility complex ( MHC ) class I locus , and the CD8+ T cell response is thought to be a primary mechanism of this effect . However , polymorphism within the MHC may also alter innate immune activity against human immunodeficiency virus type 1 ( HIV-1 ) by changing interactions of human leukocyte antigen ( HLA ) class I molecules with leukocyte immunoglobulin-like receptors ( LILR ) , a group of immunoregulatory receptors mainly expressed on myelomonocytic cells including dendritic cells ( DCs ) . We used previously characterized HLA allotype-specific binding capacities of LILRB1 and LILRB2 as well as data from a large cohort of HIV-1-infected individuals ( N = 5126 ) to test whether LILR-HLA class I interactions influence viral load in HIV-1 infection . Our analyses in persons of European descent , the largest ethnic group examined , show that the effect of HLA-B alleles on HIV-1 control correlates with the binding strength between corresponding HLA-B allotypes and LILRB2 ( p = 10−2 ) . Moreover , overall binding strength of LILRB2 to classical HLA class I allotypes , defined by the HLA-A/B/C genotypes in each patient , positively associates with viral replication in the absence of therapy in patients of both European ( p = 10−11–10−9 ) and African ( p = 10−5–10−3 ) descent . This effect appears to be driven by variations in LILRB2 binding affinities to HLA-B and is independent of individual class I allelic effects that are not related to the LILRB2 function . Correspondingly , in vitro experiments suggest that strong LILRB2-HLA binding negatively affects antigen-presenting properties of DCs . Thus , we propose an impact of LILRB2 on HIV-1 disease outcomes through altered regulation of DCs by LILRB2-HLA engagement .
HIV-1 disease progression is influenced by host genetic factors and varies greatly among infected individuals . Polymorphism in the HLA class I locus has been consistently shown to associate with HIV-1 infection outcomes by both the candidate gene approach [1] and genome-wide association studies [2] , [3] . The influence of specific HLA class I alleles on HIV-1 disease is particularly obvious for HLA-B alleles , among which HLA-B*57 and -B*27 exhibit consistent protective effects [4] , [5] , [6] , [7] and an allelic group called HLA-B*35-Px associates with accelerated disease progression [8] . HLA class I involvement in HIV-1 disease is primarily thought to be linked to cytotoxic CD8+ T lymphocyte ( CTL ) responses , which are restricted by the host's class I allotypes [9] , [10] . However , alternative mechanisms may exist , given the fact that the HLA class I molecules represent important ligands for receptors regulating activities of innate immune cells . These include the killer cell immunoglobulin-like receptors ( KIRs ) and leukocyte immunoglobulin-like receptors ( LILRs ) . Members of both receptor families have been implicated in anti-HIV immunity . For instance , certain combinations of HLA-B and KIR3DL/S1 alleles encoding receptor-ligand pairs associate with slower disease progression , which may be due to increased natural killer cell responsiveness to infected cells [11] , [12] . In addition , a strong LILRB2-HLA-B*35-Px interaction is suggested to impair dendritic cell ( DC ) function during HIV-1 infection , possibly leading to faster disease progression [13] . Down-modulation of DC function was also observed as a result of a stronger interaction between LILRB2 and HLA-B*27 loaded with the viral escape mutant KK10 L6M compared to the wild type peptide loaded complex [14] . LILRB1 and LILRB2 are the most well-studied members of the LILR family [15] , [16] . These two receptors share 82% sequence homology and bind both classical and non-classical HLA class I molecules [17] , [18] . LILRB2 is exclusively expressed on cells of the myeloid lineage , including conventional DCs , whereas LILRB1 can also be expressed by lymphoid cells . Upon ligand engagement , LILRB1 and LILRB2 induce inhibitory signals via immunoreceptor tyrosine-based inhibitory motifs ( ITIMs ) in their cytoplasmic tails . Thus , these inhibitory receptors , whose ligands are ubiquitously expressed , might play a role in elevating the activation threshold of the myelomonocytic cells and preventing self-damage . LILRB1/B2 interactions with HLA class I involve β2-microglobulin ( β2m ) and the α3 domain of the class I molecule , which are relatively conserved across allotypes [19] , [20] , [21] . A recent study demonstrated variability in binding of LILRB1- and LILRB2-Fc fusion proteins to individual class I allotypes , which included 31 HLA-A , 47 HLA-B and 16 HLA-C allotypes , indicating that additional regions of HLA class I molecules are involved in the interaction [22] . Compared to LILRB1 , LILRB2 showed a greater degree of variability in binding to HLA allotypes . Notably , HLA-B*57:01 and -B*27:05 , which associate with protection in HIV/AIDS , were among the weakest LILRB2 binders . Such a low binding level may reduce inhibitory effects of LILRB2 in DCs and thus contribute to the protective effect of the corresponding alleles . Based on these findings , we hypothesized that the differential LILRB1/2-HLA binding may impact overall immune response to HIV-1 through modification of DC function and thus influence HIV-1 disease outcomes . Specifically , HLA molecules that bind more strongly to LILRB2 were predicted to blunt DC function , which may ultimately contribute to reduced immune control of viral replication and more rapid disease progression . To test this hypothesis , we used epidemiological and HLA genotyping data from several natural history cohorts of HIV-1-infected persons and analyzed clinical outcomes in these patients in relation to in vitro determined levels of interactions between individual HLA class I allotypes and LILRB1/B2 . Our data suggest that the binding strength between LILRB2 and HLA may contribute to HIV-1 control .
To evaluate the influence of LILR-HLA interactions on HIV-1 disease , we tested for a potential correlation between LILR-HLA binding level and the strength of HLA allelic associations with viral control . Previously defined binding scores for HLA class I allotypes ( Table S1 ) were used as a measure of LILR-HLA binding strength ( Material and Methods , [22] ) . We compared the distribution of the HLA alleles in HIV-1 controllers and noncontrollers , all in the absence of therapy . Controllers were defined as individuals whose longitudinal mean viral load ( mVL ) remained below 2 , 000 copies per ml of plasma in the absence of therapy , whereas noncontrollers were patients whose mVL exceeded 10 , 000 copies per ml . Odds ratios ( ORs ) were calculated for each HLA allele using a univariate logistic regression model ( Table S1 ) , and significant ORs ( p<0 . 05 ) were tested for correlations with the LILRB1/B2 binding scores in patients of European and African descent ( referred to as whites and blacks , respectively ) . No relationship was found between the strength of LILRB1-HLA binding scores and the ORs of the corresponding alleles ( Table S2 ) . However , LILRB2 binding strength to HLA-B demonstrated a significant positive correlation with the ORs of the respective alleles in white patients ( r = 0 . 64 , p = 0 . 01; Table 1 and Figure 1 ) . The correlation in our smaller cohort of black patients occurred in the same direction , but did not reach significance ( r = 0 . 24 , p = 0 . 6 ) . Permutation analyses indicated that the significant positive correlation in whites is unlikely to have occurred by chance ( p = 0 . 03 ) . This finding suggests that the interaction between HLA-B and LILRB2 may participate in the overall effect of HLA-B alleles on HIV-1 control , where weaker binding of a given HLA-B allotype to LILRB2 correlates with greater protection of the corresponding allele , possibly as a consequence of enhanced DC function . A more rigorous test for an effect of LILRB2 on HIV-1 outcomes was performed by assigning to each patient four LILRB2-related scores , three locus-specific ( A , B , C for HLA-A , -B and -C , respectively ) scores and one combined ( ABC ) score , based on each patient's class I genotype , and then correlating these scores with measurements of HIV-1 disease outcomes in each patient . Locus-specific scores were generated as a sum of binding scores corresponding to the two alleles at each locus to reflect average LILRB2 binding . The combined ABC binding score was a sum of A , B and C scores , but for HLA-C , only 1/10 of the sum was incorporated into the final score since HLA-C is known to be expressed on the cell surface at roughly 1/10 the level of HLA-A and -B [23] . This combined score , which was used as a measure of average LILRB2 binding to class I on the cell surface , may be more relevant to the physiological consequences of the LILRB2 ligation than the locus-specific scores , since LILRB2 binds to all HLA-A , -B and -C allotypes [17] , [22] . Notably , the variation in the B binding scores appears to be the main contributor to the variation of the ABC scores at the population level ( Figure 2 ) , given the relatively small range of the A scores and low 10% contribution of the C scores . A significant positive correlation of the LILRB2-ABC binding scores with mVL used as a continuous variable was observed in both white and black patients using a univariate model ( r = 0 . 21 , p = 3×10−30 and r = 0 . 14 , p = 5×10−8 , respectively; Table 2 ) . Locus specific analyses indicated that this correlation is driven mainly by the B scores ( r = 0 . 24 , p = 1×10−38 in whites and r = 0 . 16 , p = 6×10−10 in blacks ) , since A scores show no significant correlation and C scores actually trend in the opposite direction , where higher binding of HLA-C to LILRB2 confers slight protection . To confirm that the LILRB2-HLA binding effect on HIV-1 disease outcomes is independent of the effects of individual class I alleles that are not related to LILRB2 binding , we used regression models with stepwise selection with p<0 . 05 as a threshold for inclusion , which included all class I alleles with phenotypic frequency of >2% , and LILRB2-HLA binding scores as continuous variables . The LILRB2-HLA binding effect on viral control was tested first in a categorical analysis comparing controllers to noncontrollers . The A , B and ABC binding scores demonstrated significant independent effects on viral control in white patients ( OR = 1 . 2–1 . 3 for a change of 0 . 1 binding unit , p = 10−3–10−18; Table 3 ) , whereas C score did not remain in the model . The B and ABC binding scores predicted viral control independently of all individual class I alleles in blacks as well ( OR = 1 . 1–1 . 3 for a change of 0 . 1 unit binding , p = 10−5–10−6; Table 4 ) , whereas the A and C binding scores did not stay in the model . Thus , the inverse correlation between the level of HIV-1 control and LILRB2 binding scores to HLA-B allotypes and to combined ABC allotypes were consistent in the two racial groups . Next , we applied the linear regression model with stepwise selection to the analyses of mVL in the absence of therapy where mVL was a continuous variable . Among the four binding scores tested , the B and ABC scores showed significant positive correlations with mVL independently of the effects of individual class I alleles in both whites and blacks ( Tables S3–S4 ) . This analysis indicates that an increase in 0 . 1 unit of the ABC binding score would predict 0 . 08 and 0 . 03 log10 higher mVL in white and black patients , respectively , independently of individual HLA class I alleles . This translates to an increase of 1 . 1 and 0 . 5 log10 mVL in whites and blacks , respectively , when comparing patients with the highest ABC binding score to patients with the lowest score . To test the stability of the regression models , we applied more stringent conditions in stepwise selection ( p<0 . 01 and p<0 . 001 cut-offs ) . The B and ABC scores remained significant in the categorical analysis of viral load control in whites at both cut-offs ( Table 3 ) . While similar stability was observed for the B score in blacks , the ABC score remained significant in categorical analyses only at the intermediate cut-off ( Table 4 ) . In the continuous analysis of mVL , the binding scores demonstrated variable stability ( Tables S3–S4 ) . Thus , we observed consistent associations for LILRB2-B and -ABC binding scores with HIV-1 control tested in both categorical and continuous analysis of mVL across the two racial groups . The effects were always less pronounced in the black population perhaps due to smaller number of individuals in this group . We also tested for a potential effect of LILRB2-HLA binding level on disease progression using a Cox model in a smaller cohort of seroconverts ( 780 whites and 287 blacks ) , but there was no significant effect on time to AIDS outcomes ( see Materials and Methods ) when individual class I alleles were included as covariables ( data not shown ) . This negative result may be due to low statistical power , or the LILRB2 binding effect on HIV-1 control may be outcome-specific and influence viral load only . Functional properties of DCs that result from altered LILRB2-HLA interactions were interrogated using mixed leukocyte reactions , an assay that measures the ability of DCs to stimulate antigen-specific T cell responses . Monocyte-derived dendritic cells ( MDDC ) were exposed to a panel of different recombinant HLA molecules , followed by cytokine-mediated maturation and incubation with CFSE-labeled allogeneic T cells according to a previously described protocol [24] . We observed divergent effects of different HLA allotypes on proliferative activities of allogeneic T cells , where the highest levels of proliferation were observed after MDDC exposure to HLA class I allotypes that have weakest binding to LILRB2 , and the lowest proliferative activities were observed following exposure to HLA class I molecules with strongest binding to LILRB2 ( Figures 3A and S2 ) . These data are consistent with an inverse relationship between MDDC function and corresponding LILRB2-HLA binding strength ( Figure 3B ) . siRNA-mediated knockdown of LILRB2 surface expression on MDDC ( Figure S1 ) reversed inhibitory effects of HLA class I allotypes in a reciprocal hierarchical order ( Figures 3A and S2 ) , leading to a positive association between fold changes in MDDC function after LILRB2 knockdown and corresponding LILRB2-HLA binding scores ( Figure 3C ) . However , inhibitory effects of two specific HLA class I allotypes ( HLA-A*02:01 and -C*01:02 ) on DC function were not significantly affected by LILRB2 knockdown , suggesting that these HLA allotypes may interact with additional , as of yet unidentified immunoregulatory receptors on DCs . In contrast to antigen-presentation properties , secretion of TNFα , IL-6 or IL-12p70 by MDDC was not significantly influenced by LILRB2-HLA-B interactions ( Figure S3 ) . Together , these results suggest that LILRB2-HLA impact immune control of HIV-1 through alterations of the functional antigen-presenting properties of DCs .
Among all human MHC class I alleles , those encoded at the HLA-B locus have the highest degree of genetic variation and the dominant influence on HIV/AIDS [5] . Association of particular HLA-B alleles with HIV-1 infection outcomes is traditionally linked to the ability of the corresponding allotypes to elicit CTL responses . This concept is supported by numerous studies of HLA-restricted CTL responses and viral sequence evolution in carriers of specific HLA class I alleles [25] . The distinct effect of the HLA-B locus on cellular immune responses to HIV-1 is likely due to its greater level of diversity , which results in the presentation of a broader repertoire of viral peptides that can be presented by HLA-B allotypes as compared to HLA-A or HLA-C . In addition , relative resistance of HLA-B to downregulation by HIV-1 viral protein Nef compared to HLA-A [26] as well as low expression level of HLA-C were suggested to contribute to the principal role of the HLA-B locus in HIV-1 disease . However , the structural polymorphism of HLA-B can also influence its binding to receptors other than the T cell receptor . Based on the work presented herein , we propose that variation in binding properties of HLA-B to the inhibitory myelomonocytic receptor LILRB2 can contribute to the overall HLA effects on HIV-1 infection outcomes . The ORs of individual HLA-B alleles determined by comparing HIV-1 controllers to noncontrollers correlate significantly with their LILRB2 binding strength in white patients ( Figure 1A ) . A similar trend was observed in blacks , though not significantly so ( Figure 1B ) , perhaps due to a smaller number of alleles considered in blacks ( n = 8 ) as compared to whites ( n = 14 ) . B*81:01 , an allotype present almost exclusively among blacks , appears to be an outlier in that it binds strongly to LILRB2 , but associates with robust protection against HIV-1 . B*81:01 contains an unusual polymorphism in the α3 domain that dramatically decreases CD8 binding ( the same domain that is centrally involved in LILRB2 binding [27] ) , which may explain in part the protective role of the B*81:01 in HIV/AIDS [28] , [29] . A more powerful and direct analysis of a correlation between LILRB2 binding scores and the level of viremia was conducted by assigning to each patient a LILRB2 binding score based on their class I genotypes and correlating these with the mVL determined from each patient . Our analyses included locus-specific ( A , B , C ) scores as well as a global ABC score , which was used as a measure of average LILRB2 binding to HLA class I overall . Highly significant correlations between mVL and B and ABC binding scores were observed in both white and black patients ( Table 2 ) . Two confounding factors may contribute to this strong correlation , including linkage disequilibrium between the HLA class I loci and the effects of individual HLA alleles on HIV-1 that are not related to LILRB2 function . Therefore , regression models with stepwise selection that included all individual class I alleles and LILRB2 binding scores were employed . The analyses indicated consistent effects for the B and ABC binding scores , both of which associated with viral replication tested in a categorical analysis ( controllers vs . noncontrollers , Tables 3–4 ) and when mVL was used as a continuous variable in white and black patients ( Tables S3–S4 ) . Whereas the ABC score demonstrated effects similar to the B score , the B score accounts for all or nearly all of the combined ABC effect . The OR for viral control was 1 . 1–1 . 2 per 0 . 1 unit increase of the ABC binding score when comparing controllers to noncontrollers ( Tables 3–4 ) . It is not possible to compare directly the strength of the LILRB2 binding effect to the strength of individual HLA allelic effects since the former is based on a continuous variable ( LILRB2 binding score ) and the latter on a dichotomous variable ( presence vs . absence of each allele ) . However , a comparison of the two patient groups at the extreme ends of the ABC binding scores ( i . e . 10% of patients with the lowest scores vs . 10% of patients with the highest scores ) results in an OR of 0 . 3–0 . 4 , which is close to the strength of the protective effect of B*57 in the same model ( OR = 0 . 2–0 . 3 , Tables 3–4 ) . Neither the A nor the C binding scores demonstrated consistent effects on viral load when individual class I alleles were included in the model: the A score remained only in the categorical model when the least stringent p-value cut-off ( <0 . 05 ) was used ( Table 3 ) , and the C score was not significant in any of the analyses . Thus , the negative correlation with mVL that was observed for the C scores in the univariate model ( Table 2 ) is likely due to effects of individual alleles that are not related to LILRB2 binding and/or to linkage disequilibrium between HLA-B and -C . Notably , the effects of B*27:05 , B*57:01 and B*57:03 were substantially diminished in models that included LILRB2 binding scores relative to their effects in the absence of this covariable ( Tables 3–4 and S3–S6 ) . Alternatively , the effect of B*81:01 was not markedly influenced by inclusion of the binding scores in the model . These results indicate that the protective effects of the B*27 and B*57 alleles may be partially due to low LILRB2 binding , but this does not appear to be the case for B*81:01 . The correlation between HIV-1 immune control and the binding strength of LILRB2 to HLA-B allotypes specifically ( and not HLA-A or -C ) is difficult to comprehend , since LILRB2 binds all class I molecules without discrimination [22] . The substantially greater variation in binding scores for HLA-B as compared to HLA-A allotypes ( Figure 2 ) may result in a greater influence of HLA-B on differential immune responses to HIV-1 across infected individuals . While HLA-C allotypes also show fairly broad variation in binding scores similar to HLA-B , their lower expression levels may diminish their effect in regulating myelomonocytic cells in HIV-1 infection . Alternatively , HLA-B expressed on the cell surface may behave in a distinct manner , for example due to the presence of intracellular cysteines as suggested by Gruda et al . [30] . Nevertheless , our model with combined ABC binding scores supports the idea that the average class I binding strength to LILRB2 can influence viral control , and the variation in this binding is mostly due to the allotypic diversity of the HLA-B binding strength to LILRB2 . The effect of LILRB2 binding to HLA class I on immune response to HIV-1 may be mediated by subsets of DCs expressing this receptor . Recent work demonstrated that dermal CD14+ DCs express both LILRB1 and LILRB2 [31] . These cells , along with Langerhans cells ( LCs ) and CD1a+ dermal DCs , are among the first immune cells encountered by HIV-1 in sexual transmission . Interestingly , CD14+ dermal DCs are less efficient at priming CTL than are LCs , and this difference has been attributed to the lack of LILRB1/2 expression by LCs [31] . The reduced ability of dermal CD14+ DCs to prime CTL was suggested to be due to competition between LILRB1/2 and CD8 in binding HLA class I , which has been demonstrated previously [27] . This competition may happen at the DC-T cell interface where LILRB molecules can interact in cis with HLA class I [32] on the DC surface , masking class I molecules from CD8 expressed by the T cells in a manner that does not necessarily involve inhibitory receptor signaling . Variation in the strength of LILRB2 binding to HLA class I may influence the capacity of dermal CD14+ DCs to prime virus-specific CTL and explain the effect of LILRB2 binding on viral load described herein . An alternative mechanism that is supported by our in vitro data implicates inhibition of DCs after LILRB2 ligation and receptor-mediated signal transduction . Our experiments demonstrate that stronger ligation of LILRB2 on the surface of MDDC by HLA in trans at an immature stage result in decreased capacity of these cells to stimulate T cell proliferation when they mature . This is in line with earlier work suggesting a regulatory role of the LILRB2 ligation in DC function [13] , [14] , [31] , [33] , [34] . Taken together , these data suggest that LILRB2-HLA interactions influence HIV-1 disease outcomes by regulating functional properties of DCs and their ability to generate antigen-specific T cell responses . Such effects are likely to be amplified by upregulation of LILRB2 surface expression on DCs in peripheral blood [35] , [36] and lymph nodes [37] during progressive HIV-1 infection . We have recently demonstrated a correlation between HLA-C expression level and HIV-1 control [38] . Analyses of in vivo CTL responses indicated that differential HLA-C expression influences CTL responses to HIV-1 peptides despite its lower overall cell surface expression relative to that of HLA-A and -B . The ability of HLA molecules , even those expressed at low levels , to trigger CTL killing of target cells is supported by in vitro data showing that as few as three HLA/peptide complexes can trigger CTL killing [39] . The mechanism of differential immune responses suggested in the current work is distinct from allotype-restricted CTL killing and involves regulation of DCs through engagement of LILRB2 with all allotypes of HLA-A , -B and -C ( i . e . it is not allotype specific , distinguishing it from CTL killing ) . Due to the relatively low amount of HLA-C on the cell surface , the variation in its expression level would contribute minimally to the diversity of LILRB2 binding to HLA class I as a whole . Thus , differential HLA-C expression level has a significant effect on HLA-C-restricted CTL responses [38] , but the overall low expression of HLA-C compared to HLA-A and -B limits its relative importance in mediating a response through LILRB2 , which binds ( at various levels ) to all class I molecules . The data presented herein underscore the complexity of HLA class I involvement in control of HIV-1 that goes beyond peptide presentation to CD8+ T cells . We propose that the LILRB2-HLA class I interaction may contribute to the effect of class I on HIV/AIDS through regulation of DC function . The relative size of this effect compared to the CTL or NK cell responses requires further investigation .
We used data from a total of 5126 HIV-1-infected individuals from eight US and one European cohorts: the AIDS Linked to Intravenous Experience ( ALIVE ) , the U . S . military HIV Natural History Study ( DoD HIV NHS ) , the DC Gay Cohort Study ( DCG ) , the Multicenter AIDS Cohort Study ( MACS ) , the Multicenter Hemophilia Cohort Study ( MHCS ) , the Massachusetts General Hospital Controller Cohort ( MGH ) , the San Francisco City Clinic Cohort ( SFCCC ) , the Study on the Consequences of Protease Inhibitor Era ( SCOPE ) and the Swiss HIV Cohort Study ( SHCS ) . Patients from MACS , MGH , SCOPE and SHCS , including 2685 white and 1306 black patients , were categorized in controller and noncontroller groups for the analysis of HLA class I impact on HIV-1 immune control . Longitudinal viral load data were available for 2900 white and 1490 black patients from ALIVE , MACS , MGH , DoD HIV NHS , SCOPE and SHCS . Seroconversion time and AIDS progression data were known for 780 white and 287 black patients from ALIVE , DCG , MHCS and SFCCC . This study was approved by the protocol review office of the US National Cancer Institute institutional review board , as well as by the institutional review board of Massachusetts General Hospital . Informed consent was obtained at the study sites from all individuals . Patients' ethnicities were defined based on self-report . We performed genotyping of the HLA-A/B/C following the PCR-SSOP ( sequence-specific oligonucleotide probing ) typing protocol and PCR-SBT ( sequence based typing ) recommended by the 13th International Histocompatibility Workshop ( http://www . ihwg . org ) . All HLA class I genotypes were defined to 4-digit resolution with the exceptions of A*74:01/2 , C*17 and C*18 , which were determined to 2-digits . LILRB2-HLA binding scores were defined in a previously described experiment [22] . Briefly , a set of LILR-Fc fusion proteins was tested for binding with LABScreen HLA class I SABs at a concentration of 0 . 5 , 1 and 2 µM . The level of binding was assessed by measuring the median fluorescence intensity ( MFI ) of the LILR-Fc bound to the beads using appropriate normalizations , which included subtraction of the Fc-negative control MFI and division of the result by the MFI of W6/32 ( monoclonal anti-HLA class I antibodies recognizing β2m-associated HLA molecules ) . The normalized values were assigned to each HLA allotype as binding scores . Each binding score is a function of avidity of bivalent LILRB2-Fc for HLA , which in turn depends on the affinity of monomeric LILRB to HLA . Therefore , the binding score can be used as a quantitative characteristic of the strength of LILR-HLA interactions . The relative LILR binding to different HLA allotypes was similar at each of the LILR concentrations tested ( Figure S4 ) . This consistency between LILR concentrations assured us that the difference in MFIs between the allotypes is mainly due to difference in binding strength , and is not an experimental artifact . The 1 µM concentration results were chosen as a representative dataset . Among the HIV-1-infected patients used for the analyses , frequencies of the HLA-A/B/C alleles with unknown binding scores were 2/11/24% in white and 8/13/39% in black patients . To avoid power loss , we used mean values for the corresponding locus for each genotype with unknown score . The pairs of alleles A*74:01/2 and B*81:01 differ only at the signal peptide , therefore , they were treated as individual alleles in the context of LILRB2 binding . Monocyte Derived Dendritic Cells ( MDDC ) were prepared as described previously . Briefly , 2 × 108 PBMCs were plated in 5% pooled human serum medium and incubated during 60 min at 37°C to adhere monocytes . After discarding non-adherent cells , monocytes were differentiated into MDDC in the presence of RPMI 1640 medium supplemented with 50 µg/ml of GM-CSF ( Amgen ) and 10% fetal bovine serum . On day 5 , immature MDDC were gently detached using PBS with 0 . 5% BSA and 2 mM EDTA , harvested and plated at 4x105 cell/well in a round-bottom 96-well plate ( Costar ) . Next , cells were incubated with beads coated with selected HLA-B allotypes , or uncoated control beads ( One Lambda ) for 30 min at 37°C , washed , and subsequently matured in the presence of a previously described cytokine cocktail containing 5 ng/ml IL-1β , 5 ng/ml TNFα , 1 µg/ml PGE-2 and 0 . 15 µg/ml IL-6 . After 16 hours , mature MDDC were mixed with negatively-isolated CFSE-labeled allogeneic T cells at a DC:T cell ratio of 1∶100 for mixed lymphocyte reactions . Allogeneic T cell proliferation was determined after 6 days in culture by investigators blinded towards the added HLA class I molecules , using an LSRFortessa flow cytometer ( Becton Dickinson ) . To analyze cytokine secretion , immature MDDC were prepared and treated with HLA class I molecules as described above and then matured using 5 µg/ml CL097 ( InvivoGen ) in the presence of 5 µg/ml brefeldin A . After 20 hours , cells were fixed and permeabilized , stained with antibodies recognizing intracellular IL-12p70 , TNFα and IL-6 , and processed to flow cytometric acquisition by investigators blinded towards the added HLA class I allotypes . 106 MDDC were suspended in 300 µl Optimem ( Gibco ) in the presence of 2 nmol of either LILRB2-specific ( LILRB2-siRNA ) or scramble control siRNA ( sc-siRNA ) pools ( On-TARGET plus SMARTpool , Dharmacon ) and transferred to a 4-mm electroporation cuvette ( Bio-Rad Laboratories ) . Cells were left on ice for 10 min , electroporated ( 900 V , 0 . 75 msec square wave; Genepulser Xcell; Bio-Rad Laboratories ) , and transferred back to culture medium for another 24 to 48 hours . Efficiency of specific siRNA-mediated LILRB2 knockdown was determined by flow cytometry using an anti-LILRB2 antibody ( clone 42D1 , Biolegend ) . We used SAS 9 . 1 ( SAS Institute ) for data management and statistical analyses . The effect of HLA alleles on viral control was determined by categorical analysis of the allelic frequencies in HIV-1 controllers and noncontrollers . Corresponding ORs were calculated using logistic regression model with SAS procedure PROC LOGISTIC . Relationships between viral loads and LILRB2-HLA binding scores were analyzed by the Spearman correlation test using PROC CORR . Permutation analysis was done by random assignment of binding scores to HLA-B alleles ( 10 , 000 times ) and testing the probability of significant Spearman correlation of the binding scores with ORs with p<0 . 05 . LILRB2 binding scores as continuous variables and presence versus absence of all individual HLA class I alleles of frequency ≥2% were included with stepwise selection in all regression models . Results in the tables are for the models using a threshold of a two-sided p value <0 . 05 for inclusion of a covariate as a significant independent effect . The stability of regression models was tested using more stringent thresholds of p<0 . 01 and p<0 . 001 for inclusion in the model . The results for the binding scores are indicated in the footnotes to the tables . Cox proportional hazards model was applied to perform AIDS progression analysis by using PROC PHREG . For this , we estimated the seroconversion date as the midpoint between the first positive and the last negative HIV-1 antibody test ( mean interval , 0 . 79 years; range , 0 . 07 to 3 . 0 years ) . Four end points reflecting disease progression ( AIDS outcomes ) were evaluated: time to CD4<200 cells/ml; progression to AIDS according to the 1987 definition by the Centers for Disease Control and Prevention ( CDC , [40] ) ; progression to AIDS according to the 1993 definition by CDC; and AIDS-related death [41] . Data of in vitro experiments were presented as Box and Whisker plots , reflecting the median , minimum , maximum and the 25th and 75th percentiles . Significance was tested using one-way ANOVA followed by post-hoc analysis with the Tukey multiple comparison test , or using paired t-tests , as appropriate .
|
Leukocyte immunoglobulin-like receptors B1 and B2 ( LILRB1 and LILRB2 ) bind HLA class I allotypes with variable affinities . Here , we show that the binding strength of LILRB2 to HLA class I positively associates with level of viremia in a large cohort of untreated HIV-1-infected patients . This effect appears to be driven by HLA-B polymorphism and demonstrates independence from class I allelic effects on viral load . Our in vitro experiments suggest that strong LILRB2-HLA binding negatively affects antigen-presenting properties of dendritic cells ( DCs ) . Thus , we propose an impact of LILRB2 on HIV-1 immune control through altered regulation of DCs by LILRB2-HLA engagement .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"infectious",
"diseases",
"hiv",
"retrovirology",
"and",
"hiv",
"immunopathogenesis",
"viral",
"diseases"
] |
2014
|
LILRB2 Interaction with HLA Class I Correlates with Control of HIV-1 Infection
|
Locomotor activity rhythms are controlled by a network of ~150 circadian neurons within the adult Drosophila brain . They are subdivided based on their anatomical locations and properties . We profiled transcripts “around the clock” from three key groups of circadian neurons with different functions . We also profiled a non-circadian outgroup , dopaminergic ( TH ) neurons . They have cycling transcripts but fewer than clock neurons as well as low expression and poor cycling of clock gene transcripts . This suggests that TH neurons do not have a canonical circadian clock and that their gene expression cycling is driven by brain systemic cues . The three circadian groups are surprisingly diverse in their cycling transcripts and overall gene expression patterns , which include known and putative novel neuropeptides . Even the overall phase distributions of cycling transcripts are distinct , indicating that different regulatory principles govern transcript oscillations . This surprising cell-type diversity parallels the functional heterogeneity of the different neurons .
Nearly all organisms possess a circadian clock , which allows for the adaptation and anticipation of the daily oscillations of day ( light ) and night ( dark ) . The circadian clock of Drosophila melanogaster drives a 24-hour locomotor activity rhythm , which includes bouts of morning and evening activity . This rhythmic behavior is controlled by a molecular clock , which includes transcriptional negative feedback loops that are conserved from insects to mammals . Clock ( CLK ) and Cycle ( CYC ) form a heterodimeric transcription factor that functions as the central circadian transcriptional activator . CLK/CYC activates the expression of two transcription factor genes , timeless ( tim ) and period ( per ) in the late morning . TIM and PER enter the nucleus in the early night , inhibit CLK/CYC driven transcription , and sequester CLK/CYC until morning . Once released , CLK/CYC start the cycle over again by activating tim and per . This negative feedback leads to oscillating gene expression for per and tim as well as many other CLK/CYC controlled genes . Two other CLK/CYC transcriptional target genes , Vrille ( vri ) and par domain protein 1 ( pdp1 ) encode transcription factors that form a second circadian feedback loop . The cyclical expression of many different genes provides temporal control of different behaviors or outputs of the clock; they include for example feeding and sleep ( reviewed in [1] , [2] ) . The molecular clock is expressed in ~150 clock neurons in the Drosophila brain , which function together to regulate many of these circadian behaviors . These neurons are classified based upon their anatomical location ( reviewed in [3 , 4] ) . There are dorsal neurons that are divided into three groups: DN1s , DN2s and DN3s . There are also lateral neurons ( LNs ) , which can be subdivided into 4 groups . They include the lateral posterior neurons ( LPN; 3 neurons ) , dorsal lateral neurons ( LNds; 6 neurons ) , and two groups of ventral lateral neurons: the small ventral lateral neurons ( s-LNvs; 5 neurons ) and the large ventral lateral neurons ( l-LNvs; 4 neurons ) . The LNs can also be subdivided based on expression of the neuropeptide , PDF ( pigment dispersing factor ) . The PDF+ lateral neurons consist of all of the LNvs except the 5th small LNv . PDF- lateral neurons consist of all the LNds plus the 5th small LNv . The PDF+ LNvs are considered to be the major fly pacemaker neurons as they are sufficient to drive rhythmic locomotor behavior in the absence of light cues [5 , 6] . Like in flies , an anatomically restricted region of the mammalian brain serves as the circadian central pacemaker . This is the suprachiasmatic nucleus ( SCN ) , a subregion of the hypothalamus that contains ~15 , 000 neurons ( in mouse ) . The SCN has two main regions: the ventrolateral “core , ” which expresses vasoactive intestinal polypeptide ( VIP ) ; and the dorsolateral “shell , ” which expresses arginine vasopressin ( AVP ) . Although the core and shell provide a simple anatomical framework , the SCN is complicated: different regions oscillate in different phases , express scores of different neuropeptides and project to unique target areas of the brain [7–9] . A key question in both systems is how brain circadian neurons work together to drive complex circadian behaviors . Due to the relative simplicity of the Drosophila system , much more is known about the fly circadian network . The PDF neurons , the l-LNvs and s-LNvs , are probably part of the primary light-input pathway to the clock . They receive light information directly via the intracellular presence of the blue-light photoreceptor Cryptochrome ( CRY ) as well as indirectly via photoreceptors of both the compound eyes and the H-B eyelets [10–13] . PDF release by the LNvs is critical for communicating time of day signals to the LNds and DN1s as well as to the non-circadian LK/LK-R neurons [14–16] . A subset of the LNds , the 3 Cry+ LNds well as the 5th small PDF- LNv , are important for controlling evening anticipatory behavior and are therefore referred to as evening cells [17–19] . However , their role is not limited to driving evening activity as they can also modulate morning activity [20] . This is because silencing them leads to a strong decrease in both morning and evening locomotor activity , and other experiments from our lab indicate that the LNds are general activity-promoting neurons [19] . The DN1s are intriguing . A recent study illustrates that the circadian clock controls daily changes in DN1 membrane excitability [21] . This cell-autonomous control is then modulated by effects from the circadian network . For example , PDF signaling from the LNvs to the DN1s is important for arousal in the morning [22–24] . The DN1s then release the neuropeptide , Dh31 , to promote awakening at dawn [25] . Later in the day however , DN1s send inhibitory signals to the LNds and LNvs to promote the siesta and nighttime sleep [26–28] . Not surprisingly , the DN1s contact several groups of neurons to carry out these multiple functions: the pars intercerebralis ( PI ) , the LNds and the LNvs [19 , 23 , 24 , 29] . To learn more about these three important groups of circadian neurons and what molecules may be important for their functions , we used RNA-sequencing ( RNA-seq ) to profile the transcriptomes of isolated PDF+ lateral neurons ( referred to subsequently as LNvs ) , PDF- lateral neurons ( LNds plus including 5th small PDF- LNv; referred to subsequently as LNds ) and DN1s . We also assayed dopaminergic neurons ( referred to as TH; tyrosine hydroxylase ) as a non-circadian outgroup . This profiling was done “around the clock” to address the temporal ( circadian ) regulation of gene expression . First , we identified both common as well as group-specific transcripts and then identified among them known and putative neuropeptides . They could facilitate intra-circadian network communication and/or communicate with neurons outside of the circadian network to drive output behaviors . Second , we identified cycling transcripts in each neuronal group . The low level of core clock gene expression in dopaminergic neurons indicates that cell-autonomous clock function may not be ubiquitous in the fly brain . Nonetheless , a small number of cycling transcripts are identified in TH neurons . In the four different circadian neuronal groups , cycling gene expression was almost completely distinct , which resembles what has been reported for mammalian cells and tissues . In addition the phase distribution of these cycling clock neuron transcripts was strikingly different in the LNvs , suggesting that distinct mechanisms determine the phase of transcript cycling within different clock neurons .
To compare specific Drosophila circadian neuron subsets , we sequenced the transcriptomes of 3 well-described groups of circadian neurons: the LNds ( LNds plus 5th PDF- LNv ) , LNvs ( small and large PDF+ cells ) and a subset of the DN1s . We also sequenced a non-circadian outgroup , dopaminergic cells ( TH cells; ~120 neurons per brain ) . Neuron groups were labeled with GFP using specific GAL4 drivers and manually isolated with 3 rounds of fluorescent cell sorting ( see Methods and [30] ) . Every sample contained 50–100 cells , which yielded approximately 200-500pg of total RNA; this was amplified to make RNA libraries for deep sequencing ( see Methods and [30] ) . Two independent sets of circadian time courses were performed for each group at 4 hour intervals to identify cycling transcripts ( see below ) . Experiments were performed in light:dark ( LD ) conditions to maximize comparisons between cycling gene expression and circadian behavior , which is more robust in LD than in DD . We also combined data from both replicates of 6 circadian time points , pooling all 12 samples from each neuron group , to address cell type-specific gene expression without regard to circadian time . As expected , the circadian genes timeless ( tim ) and cryptochrome ( cry ) are strongly expressed in all three groups of circadian neurons but poorly expressed in non-circadian dopaminergic neurons ( Fig 1 ) . Also as expected , a control gene , actin ( Act5c ) , is expressed approximately equally in all 4 groups of neurons . Previous studies have shown that the neuropeptides PDF and ITP , the dopamine biosynthesis gene ple , and transcription factor gl are expressed in LNvs , LNds , TH cells and DN1s , respectively [31–34] . Our deep sequencing results confirm these observations: PDF is expressed solely in the LNvs , Itp mRNA is highly expressed only in the LNds , ple transcripts are enriched in TH cells and gl mRNA is found exclusively in the DN1s ( Fig 1 ) . As evidenced by the presence of the s-LNv-specific transcript sNPF ( Fig 1; [32] , the LNv samples contains the large cells ( l-LNvs ) as well as the harder to isolate small cell population ( s-LNvs ) . Some transcripts on the other hand show unexpected profiles . For example , the neuropeptide sNPF is expressed not only in the small LNvs but also the LNds and TH cells , but it has not been detected in DN1s [35 , 36] . However , sequencing data indicate that sNPF transcripts are present in DN1s as well as the 3 expected locations ( Fig 1 ) . There are a few other discrepancies between transcript detection in sorted cells and previous immunostaining results ( see Discussion ) . Nonetheless , the good correlation with previously defined neuron-specific factors suggests that the RNA sequencing libraries reflect the transcriptomes of these four neuronal groups . We then used the transcriptional profiling results to address two questions . First , do the three groups of circadian neuron have shared transcripts beyond the core clock mRNAs ? These additional transcripts may play some common role in the different clock neurons , for example circadian timekeeping like the core clock mRNAs . Second , are there transcripts enriched in a single circadian group , which could provide insight into the more specialized functions of that group [28] ? To address the first question , we identified transcripts that are more highly expressed in at least two of the three circadian groups relative to TH neurons ( Table 1; see below for an explanation of why we did not require enrichment in all 3 groups ) . 18 transcripts are enriched by this criterion ( greater than 5-fold enrichment and p-value <0 . 05 in Anova Tukey HSD post-hoc test with Benjamini Hochberg correction; see Methods ) . As expected , almost all core clock genes are present among these 18 genes . cry , vri , and tim are enriched in all three groups , but per and Clk are only enriched in two . This is because per and Clk mRNAs are not sufficiently expressed in LNds to reach the required threshold of 10 average reads/million ( Table 1 , LR low reads ) . Several other enriched genes have also been implicated in circadian processes , e . g . , the neuropeptide Dh31 [25] as well as the transcription factor unfulfilled or HR51[37 , 38] , whereas others function in a variety of different processes . They include the neuropeptide ( npf ) as well as two genes involved in octopamine synthesis ( Tdc2 and Tbh ) . In addition , they could also contribute to aspects of circadian function carried out similarly by all three circadian groups . To address neuron-specific functions , we identified transcripts that are more abundant in one group of circadian neurons relative to the other 2 groups ( Fig 2A; greater than 2-fold enrichment , p-value <0 . 05 in Anova Tukey HSD post-hoc test with Benjamini Hochberg correction; see Methods ) . About 2 . 5% of the LNv transcriptome , 113 genes , meet this criterion ( S2 File ) . They include genes previously shown to be preferentially expressed in LNvs , e . g . , the neuropeptide PDF , the transcription factor dimmed , the translation factor twenty-four ( tyf ) as well as the octopamine receptor , oamb [39–43] . In addition , three highly significant LNv-enriched transcripts include two of unknown function ( CG12947 and CG43117 ) as well as a transcript encoding the putative Clk coactivator opa ( Fig 2B; [44] ) . Gene ontology analysis ( GO; see Methods ) indicates that genes encoding G-protein coupled receptors ( 4 GPCR genes; Dh31-R , MsR1 , AstC-R2 , CG13229 ) , genes involved in cyclic nucleotide biogenesis ( 3 genes ) , and genes encoding members of the Peptidase M13 , neprilysin family ( 3 genes; Nep1 , Nep2 and Nep3 ) are overrepresented among the list of LNv mRNAs . These functions are consistent with our proposal that the LNvs integrate environmental information ( GPCRs ) and transmit that information to the rest of the circadian network [19] , i . e . , requiring signal transduction ( cyclic nucleotide biogenesis ) and neuropeptide processing ( peptidases ) . Although a much smaller fraction of the LNd transcriptome is enriched compared to the other two groups of circadian neurons ( <1% , 29 transcripts; S2 File ) , several known LNd-specific transcripts were identified . They include the acetylcholine biosynthetic transcript ChAT and the neuropeptide ITP; their products have both been identified in LNds by immunohistochemistry [32] . In addition , the transcripts encoding the bHLH transcription factor CG34367 and a component of the integrator complex that processes snRNAs , IntS12 , are significantly LNd-enriched ( Fig 2C; S2 File ) . GO analysis indicated that transcripts involved in neuropeptide hormone activity are enriched in LNds ( 3 genes; hug , ITP , Dh44 ) . The expression of Dh44 is surprising as this neuropeptide is reported to be absent from these neurons ( [24]; see Discussion ) . About 5% of the DN1 transcriptome , 264 transcripts , is enriched compared to LNvs and LNds ( S2 File ) . The neuropeptide hormone Dh31 was recently reported to be strongly expressed in DN1s [25] , and its transcript is indeed expressed much more highly in the DN1s than in the other two circadian neuron groups ( ~100-fold; Fig 2D ) . In addition , two transcripts known to be expressed in the DN1s , the transcription activator , gl , and the glutamate vesicular transporter , Vglut , are enriched ( [34] [45]; Fig 1 , Fig 2D; see Discussion ) . GO analysis indicates that genes in the cytochrome p450 family ( 15 genes ) , genes encoding proteins involved in hormone binding ( 7 genes ) , and genes encoding S1 and S6 peptidases ( 10 genes ) are overrepresented in the more highly expressed DN1 genes ( Fig 2A ) . Other highly enriched DN1 transcripts include the neuropeptide CNMa ( Fig 2D ) . Because neuropeptides feature prominently in this analysis , we examined this class of genes in more detail . Transcripts encoding neuropeptides known to be expressed in the circadian network were identified ( Fig 3A ) , and the localization of these neuropeptides within the circadian neurons is summarized in a cartoon ( Fig 3B ) . In addition , transcripts for several neuropeptides not known to be expressed in circadian neurons were identified in LNds ( hugin , Dms , Trissin , and Ast-C ) and DN1s ( Ast-C , CCHA1 , and CNMa ) . Receptor mRNAs for some neuropeptides were also expressed in the circadian network ( Fig 3A ) , suggesting that they may act in part within this network . In addition to known neuropeptides , we also noticed a number of short , non-intron-containing transcripts that are enriched in circadian neurons; these features are common in neuropeptide genes . To further explore this possibility , NeuroPID was used to analyze the predicted proteins encoded by these mRNAs [46] . NeuroPID examines a peptide sequence for signal peptides and cleavage sites characteristic of pro-neuropeptides ( see Methods ) . As proof of principle , NeuroPID successfully identified many known neuropeptide precursors among the large number of transcripts enriched in circadian neurons ( Fig 3A; bold ) . Prominent exceptions are mRNAs for sNPF and Dh44 , which were not identified . NeuroPID also identified putative novel pro-neuropeptides , some of which scored similarly to well-characterized neuropeptides ( Fig 3A; first column ) . For example , CG17777 is a putative signal peptide-containing pro-neuropeptide identified by NeuroPID . It is expressed in all three circadian neuron groups and is also enriched ( Table 1 ) . Two of the most abundant transcripts enriched in LNvs , CG43117 and CG4577 , also encode putative proneuropeptides by these criteria ( Fig 2B and Fig 3A ) . Transcriptome profiling of Drosophila heads has identified many cycling transcripts [47–50] . However , there may be additional genes under circadian control only within individual neuron groups . To address this possibility in a comprehensive manner , cycling transcripts were identified in the four groups of neurons . As mentioned above , 2 independent 6 time point circadian RNA samples were purified and sequenced from each group , and they were analyzed using both JTK cycle ( p-value < 0 . 05 ) and fourier analysis ( F24 score > 0 . 5; see Methods for additional criteria ) . Genes that encode cycling transcripts with both methods were defined as high confidence cyclers ( HC cyclers ) , and genes that cycle with only one method were low confidence cyclers ( LC cyclers ) . The two methods identified between ~150–300 HC cyclers in each circadian group and many fewer in TH neurons , i . e . , 249 , 303 , 185 , and 31 HC cyclers in LNvs , LNds , DN1s , and TH neurons , respectively; S3 File ) . As ~30% of the cyclers identified by fourier analysis are also identified with JTK cycle ( S1 Fig ) , the HC criterion has a much lower false positive rate . The stringent HC criterion may explain why we observe so few cyclers compared to mammalian studies , especially compared to a recent SCN study [51] . However , it is generally the case that flies have fewer cycling transcripts than mammals [52 , 53] . Only 4% of these HC cyclers were previously identified as cycling head transcripts using similar methods and fly lines [50] . This comparison suggests that many cycling circadian neuron transcripts are indeed invisible in studies of more heterogeneous tissues like the fly head and fly brain because they are neuron-specific . We first examined the known CLK/CYC core clock target genes: tim , per , vri , and pdp1 . To identify all 4 transcripts as cyclers in the three groups of circadian neurons required the LC criterion . This emphasizes the stringent nature of the HC criterion , which is useful for numerical comparisons but not necessarily for identifying individual cycling transcripts because of false negatives ( Fig 4A and Table 2 ) . The TH neurons express low levels of these core clock genes , and only tim was identified there as a cycling ( LC ) transcript . Even using LC as well as HC criteria , only 12 cycling mRNAs ( the 4 core clock genes plus 8 others ) are common among all 3 circadian neurons . These transcripts are involved in diverse processes , from histone methylation to neuron morphogenesis ( Table 2 ) . A total of 30 additional transcripts ( ~6% of the total ) were identified as HC cyclers in two circadian groups ( Table 2 and Fig 4B ) . The list includes several genes previously shown to impact circadian rhythms and/or sleep ( e . g . cwo , Usp8 , and Dh31 ) . Although the levels of Dh31 mRNAs are much higher in DN1s than in LNvs , these neuropeptide-encoding transcripts were identified as HC cyclers in both sets of neurons . Interestingly , Dh31 expresses a different transcript isoform in LNvs , where it has a much shorter 3’-UTR ( Fig 4C; see Discussion ) . Most cycling mRNAs are specific to a single group of circadian neurons ( Fig 4B and S2 Fig ) . This is partially explained by differential gene expression: ~15% of neuron-specific cycling transcripts are not expressed in the other two neuronal groups . For example , CCHa1r mRNA encodes a GPCR and is one of the top LNv cyclers; it peaks in the morning ( ZT2 ) , disappears at night ( Fig 5A ) and is not expressed in either LNds or DN1s ( definition: average expression >5 reads/million ) . The remainder , ~85% of the clock neuron cycling transcripts , oscillate specifically in one group despite being expressed in one or both of the other circadian groups . For example , the metabotropic glutamate receptor ( mGluRA ) mRNA is one of the highest amplitude cyclers in LNds , ~60-fold [28] . Although it only cycles in LNds ( Fig 5B ) , mGluRA mRNA is also expressed at comparable levels in LNvs and DN1s as previously reported [54] . This suggests that LNvs and DN1s also respond to glutamate but only LNds temporally modulate their response to this neurotransmitter [28] . There are also a few examples in which a transcript may cycle in two groups of circadian neurons but with a different phase . CG17777 mRNA encodes a putative novel neuropeptide and is a HC cycler in LNvs and DN1s ( Fig 5C ) . These transcripts peak in the early morning and are at their nadir during the mid-day in LNvs , but they are lowest in early morning and peak at mid-day in DN1s . It is also expressed in LNds , where it does not cycle . A comparison of cycling phase between neuron groups was also done genome-wide despite the fact that most transcript cycling is restricted to a single group . To this end , the three HC cycling transcript phase distributions were plotted as histograms , i . e . , % of all cyclers with a particular phase ( Fig 6A ) . Cycling transcripts in LNds and DN1s have similar unimodal phase distributions centered around mid-day ( Fig 6A , red and blue ) . It is difficult to determine whether these two distributions are truly different . This is because the time points used for the DN1 and LNd purifications were somewhat different , which could modestly affect phase determination ( see Methods ) . However , the cycling LNv transcripts have a very different distribution; it is bimodal with one peak shortly after lights on and the other shortly after lights off ( Fig 6A; green ) . This LNv bimodal phase distribution is maintained with the inclusion of LC as well as HC cyclers ( 900 transcripts in total; S3 Fig ) . In contrast , the CLK/CYC direct target core clock transcripts ( tim , per , vri , pdp1 ) have a similar unimodal phase of ~ZT12-ZT15 in all three groups . ( Fig 6A; red , blue , and green dotted lines ) . This phase is also similar to what has been observed in heads for these same transcripts [55] . As ZT12-15 is very different from the overall cycling transcript phase distribution from LNvs , it indicates that this different profile is unlikely due to a technical or analytical artifact . This suggests in turn that the different phase distributions reflect at least two different mechanisms operating in the three clock neuron groups ( see Discussion ) . One simple hypothesis is that the bimodal phase distribution comes from the two functionally distinct subclasses of LNv neurons: the s-LNvs and the l-LNvs . As the s-LNvs have been shown to track lights-on [56] , we considered that those transcripts peaking in the morning might come predominantly from s-LNvs ( Fig 6B ) . To address this possibility , we used previous microarray data that separately profiled the s-LNvs and l-LNvs to calculate the expression ratio of each transcript in the two clock neuron groups [40] . The distribution ( a value of <1 indicates more expression in s-LNvs; Fig 6B ) indeed indicates that transcripts peaking in the morning show higher expression in the s-LNvs than transcripts peaking in the evening ( t-test; p-value 3x10-5; see Discussion ) . Given the low number of TH neuron HC cyclers ( 31 ) , they may be false positives . Interestingly however , their phase distribution is not random but is centered around mid-day ( Fig 6C ) . As this distribution anticipates the light-dark transition , it suggests a regulated mechanism of transcriptional regulation within TH neurons rather than a response to the light-dark cycle ( see Discussion ) . In addition , gene ontology analysis indicates that TH cyclers are enriched in interesting functions including the mitochondrial inner membrane ( HC cyclers as well as all cyclers ( HC and LC ) ) , cytochrome-c oxidase activity ( all cyclers ) , and the chaperone tailless complex polypeptide 1 ( TCP1; all cyclers ) .
We have profiled the transcripts from three circadian neuron groups as well as from TH neurons . The profiling was also done as a function of time so that transcript oscillations as well as relative transcript abundance could be assessed in the four groups . In TH neurons , some clock gene mRNAs like Clk are difficult to detect , and all clock gene expression is at least ~5x lower than in the circadian neurons . This is also the case for tim mRNA , which is the only core clock gene ( LC ) cycler in TH neurons . The low expression levels and weak cycling of clock gene mRNAs suggest a substantial quantitative and probably qualitative difference in gene expression between the 75 pairs of clock neurons and TH neurons . The low core clock gene RNA levels in TH neurons could also reflect some clock neuron contamination or that a small percentage of the TH neurons contain a molecular clock . However , we have evidence that the default state of general brain neuron chromatin is permissive for a low level of non-cycling core clock gene transcription ( S4 Fig ) , which is mechanistically distinct from clock gene expression in circadian cells . All of these data suggest that TH neurons do not express a functional circadian clock . Indeed , this notion is consistent with previous immunostaining experiments showing that CLK is detectable only in the ~150 circadian neurons in the adult brain [57] . The absence of an endogenous clock can also explain why TH neurons have many fewer cycling transcripts . Although they too could be due to contamination or false positives , these cyclers are enriched in specific functions . Their striking phase distribution ( Fig 6C ) also suggests that they are genuine cyclers and further suggests that this cycling is governed by a single prominent mechanism . These rhythms could be an indirect consequence of behavioral or physiological rhythms ( homeostatic regulation ) or be modulated by the 150 circadian neurons . This interpretation recalls results from mammals in which 10% of cycling liver transcripts are driven by systemic cues and continue to cycle even in the absence of a clock in the liver [58] . Interesting , the percentage of TH cyclers compared to those in the circadian neurons is not dissimilar , i . e . , 5–10% ( Fig 4B ) . It is possible that this view extends to mammals , i . e . , that not all mammalian cells and most importantly not all brain neurons contain a functional circadian clock . Mammalian transcripts important for dopamine synthesis such as tyrosine hydrolyase ( TH ) oscillate throughout the day in dopaminergic neurons [59 , 60] . Although mammalian dopaminergic neurons in the ventral tagmental area ( VTA ) have been reported to express CLK [59] , recent neuron-specific transcriptome studies report little or no core clock gene expression in dopaminergic neurons ( personal communication , S . Nelson; Neuro-seq project ) . It is therefore possible that transcript cycling in mammalian dopaminergic neurons is also driven by signals from elsewhere in the brain , perhaps from the SCN , and that a functional clock may not exist in all mammalian cells . There is substantial overlap between this RNA-seq profiling of clock neurons and previous microarray experiments [37 , 40]; PDF neurons , s-LNvs and l-LNvs , as well as the overall clock neuron population were separately characterized in these studies . Moreover , several brain proteins , including neuropeptides and neurotransmitter systems , have been previously identified within different clock neurons by immunohistochemistry . These results also agree for the most part with our transcript profiling , suggesting that the neuron purification and RNA sequencing libraries properly describe the transcriptomes of the 3 clock neuron groups . The few discrepancies indicate clock neurons with specific mRNAs but no detectable protein . Although this could reflect contamination , a more positive interpretation is that some cell-type specific protein expression may rely on post-transcriptional regulation . The neuropeptide Dh31 is a good example: its transcript has a shorter 3’UTR in LNvs than in DN1s , which argues strongly against contamination with DN1 RNA ( Fig 4C ) . As the neuropeptide is detectable in DN1s but not LNvs , the longer 3’UTR may be necessary for the binding of required RNA binding proteins , for example positive translation factors . In contrast to LNvs , LNd and DN1 genome-wide profiling has not been previously reported . These groups include cells that promote evening activity and sleep , respectively [19 , 28] . Consistent with these behavioral roles , LNds cells have enriched levels of acetylcholine enzyme mRNAs , whereas DN1s have enriched levels of the glutamate vesicular transporter ( Vglut; Fig 2D ) mRNA . ( Glutamate acts as an inhibitory neurotransmitter in the circadian network; [28] These transcripts are functional: RNAi of the acetylcholine vesicular transporter mRNA in LNds cells increases sleep ( S5 Fig ) , whereas RNAi of the glutamate vesicular transporter within DN1s reduces sleep [28] The LNds and DN1s also contain neuropeptide transcripts not previously implicated in the circadian system . Although the functions of most of these neuropeptides are not understood , PK-2 ( encoded by the propeptide hugin ) is implicated in feeding control [61] . This suggests that the circadian system may use PK-2 to convey time of day information to neurons modulating feeding . PK-2 and several other identified peptides ( Dms and CNMa ) , have mammalian homologs that may have a role in the mammalian circadian system . For example , the PK-2 homolog Neuromedin-U ( NmU ) is regulated by the circadian clock in the SCN [62] . Some of these neuropeptides as well as additional neurotransmitters and neuropeptide receptors ( Fig 3A ) may also contribute to cell-specific circadian functions in Drosophila . These neuropeptides are representative of most differential and cycling gene expression; the three different circadian neuron groups are largely distinct . We expected the profiles to be more shared , but only the core clock genes and a handful of additional genes are regulated similarly in the three clock neuron groups . Even with relaxed criteria to include a greater number of cycling transcripts from each cell group , there was no change to the conclusion , nor was it changed by including LC as well as HC cyclers . However , the limited overlap in cycling transcripts could be influenced by cell heterogeneity , which exists within each group . The DN1s have the greatest number of enriched mRNAs and may be the most variable of the 3 groups ( S2 File ) . They likely contain sleep-promoting cells expressing glutamate as well as arousal-promoting cells containing Dh31 [25 , 28] . These cell-type specific cycling gene expression results recall similar comparisons between mammalian tissues: cycling gene expression is predominantly tissue-specific with only modest shared gene expression beyond the core clock genes [51 , 63] . This interpretation also offers a simple explanation of why most of these cycling transcripts were absent from those previously reported in fly heads [47–50]: most neuron-specific cycling transcripts are obscured by the same non-cycling transcript from many other neurons in head mRNA . Despite the different cycling mRNAs , those in LNds , DN1s and TH cells have similar phase distributions; they peak at about ZT11-13 . This is similar to the well-described phase of CLK/CYC-controlled gene expression from heads [55] . In contrast , the LNv phase distribution is dramatically different: it has two peaks , one shortly after lights-on at ZT0 and the other shortly after lights-off at ZT12 . The striking difference between the LNvs and the other circadian neurons is unlikely due to technical or analytical difficulties , as the CLK/CYC controlled core clock mRNAs in LNvs have a similar unimodal phase at around ZT14 like the other 2 circadian groups . In addition , we observe the same bimodal phase distribution when including LC as well as HC cyclers ( S3 Fig ) . It is tempting to assign the two peaks of cycling transcripts to the s-LNvs and the l-LNvs , respectively . Indeed , transcripts with the later phase show a statistically significant bias toward higher expression in the l-LNvs rather than the s-LNvs ( Fig 6B ) . However , there are multiple exceptions , e . g . , genes in the evening peak that were previously found to be more highly expressed in s-LNvs . Although intragroup heterogeneity complicates the interpretation , there are probably two major peaks of expression/day even within a single cell type , which is similar to data from mammalian liver and SCN [51 , 64] . However , these are different transcripts , i . e . , we do not reliably detect a group of transcripts with two peaks/day comparable to the light sensing pathway recently reported from the SCN [51] . This difference may reflect the major differences in light sensing and light input pathways between flies and mammals . Stronger conclusions will require more experiments . The mechanisms that underlie these cell type-specific phase distributions are unknown . The similarity between LNds and DN1s suggests that they share common mechanisms if not molecules , which likely differ in PDF cells . We can imagine two possibilities to explain the similar LNd and DN1 patterns . One is that these two groups share a circadian firing pattern , which results in a common circadian pattern in calcium and calcium-dependent gene expression . However , recent results suggest that the calcium activity patterns of LNds and DN1 are quite different , with the LNds firing in late morning and the DN1s at late night and early morning [21 , 56] . The other is that the two groups receive similar circadian input , for example from light or from PDF activation of the PDF receptor ( PDFR ) . As PDF signaling is under circadian control [5 , 65 , 66] , it should result in a similar circadian signal transduction pathway downstream of PDFR [67] . This could give rise to a common phase of cycling gene expression in LNds and DN1s despite substantial differences in responsive ( accessible ) genes . The connection of the LNvs with light input suggests that its very different phase distribution might reflect a gene expression response to the lights-on and lights-off stimuli characteristic of the entrainment protocol . All of these possibilities require experimental support and still do not address the mechanisms or molecules that underlie the phase distributions within PDF neurons .
In order to visualize neurons for sorting the following fly lines were used: Pdf-GAL4 , UAS-mCD8::GFP for LNvs , Dv-Pdf-GAL4 , UAS-EGFP , PDF-RFP for LNds , yw; CLK4 . 1m-GAL4 , UAS-EGFP for DN1s and yw; UAS-EGFP; TH-GAL4 for dopaminergic or TH cells . Pdf-RFP flies were a gift of J . Blau . CLKout flies were a gift of P . Hardin [68] . ChAT RNAi flies were obtained from Bloomington Drosophila Stock Center ( BL25856 , [69] ) . Flies were entrained for 4 days in 12:12 LD cycles . Fly brains were isolated every 4 hours for two independent sets of six circadian timepoints for each neuron group . Samples for LNvs , LNds and TH were collected at ZT2 , ZT6 , ZT10 , ZT14 , ZT18 and ZT22 . Samples for DN1s were collected at ZT3 , ZT7 , ZT11 , ZT15 , ZT19 and ZT23 . Brains were dissociated and the neurons of interest were isolated using three rounds of manual sorting using a fluorescent microscope . PolyA+ RNA was isolated from approximately 50–100 isolated neurons and subjected to one round of linear amplification prior to making libraries for deep sequencing [30] . Libraries were sequenced on a Hi-Seq 2000 ( Illumina ) using 50bp single end reads . RNA for whole brain RNA-seq was extracted from brains collected at ZT2 and ZT14 using standard Trizol methods ( Invitrogen ) . Libraries were made following the standard protocol of the TruSeq RNA Sample Prep Kit ( v2; Illumina ) . The resulting sequencing files were mapped to the Drosophila genome ( dm3 ) using Tophat [70 , 71] . On average ~50% of the reads mapped to the genome . Lower mapping frequencies were due to a number of factors including the presence of rRNA and contamination of the libraries with non-Drosophila nucleic acid . The total number of reads in each library is summarized in S1 File . The libraries generated from small numbers of purified neurons show 3’-bias ( Fig 1 ) . Although sometimes recommended , we did not remove identical sequencing reads ( often called removing PCR duplicates ) from our sequencing libraries since oligo-dT amplification lead to an abundance similar 3’-reads in the libraries that would be removed . After mapping , gene expression was quantified using End Sequencing Analysis Toolkit ( http://garberlab . umassmed . edu/software/esat/; [72] . ESAT quantitates gene expression by examining reads in a sliding 300bp window at 3’-end all isoforms of a gene and prevents any bias introduced by differences in gene length using more standard methods such as Cufflinks [73] . To ensure that gene expression is quantified similarly and is comparable , all 48 libraries were analyzed simultaneously using ESAT . Gene expression values were normalized and are expressed as reads per one million reads . As noted in earlier studies , the low amount of starting material isolated from purified neurons leads to lower sample reproducibility than observed with typical RNA-seq experiments [74] . The mean values for Pearrson coefficients for pair-wise sample comparisons for LNvs , LNds , DN1s and TH neurons were 0 . 9 , 0 . 83 , 0 . 87 and 0 . 81 , respectively . There was typically more variation from libraries from dopaminergic cells perhaps due to the larger amount of heterogeneity in that population ( ~120 TH cells in the brain; [75] ) . To visualize transcript levels in the sequencing libraries , bigwig visualization files were made from bam files and were visualized using the integrated genome browser ( IGV; Whitehead Institute ) . In Fig 1 , the images represent the sum of all 12 samples made from each cell type . To illustrate cycling , replicate timepoints were combined to more concisely show cycling transcript levels ( Fig 4C ) . Heatmaps were produced from normalized expression data using heatmap . 2 in gplots package for R . Sequencing data is available at Gene Expression Omnibus ( Accession number GSE77451 ) . To identify differentially expressed genes , the average transcript levels was calculated for each set of 12 samples in each neuronal group ( two 6 timepoint circadian experiments ) . Transcripts expressed at low levels were removed by requiring an average of at least 10 reads/million in each of the two independent six timepoint experiments . The relative difference in transcript level in each of the four neuronal groups was calculated by taking the ratio of the averages . Transcripts that showed a 2-fold change in levels were analyzed further for statistical significance . An Anova analysis was performed with a p-value cutoff of 0 . 05 . A Tukey HSD post-hoc analysis was used to identify statistically significant groups and a Benjamini Hochberg correction ( p-value <0 . 05 ) was used to account for the complications of multiple comparisons . Transcripts were considered to be enriched in circadian neurons if they showed a 5-fold enrichment when compared to TH neurons in at least 2 of the circadian neuron groups and met the statistical cutoffs . Transcripts were considered to be specifically enriched in one group of circadian neurons if they met all statistical cutoffs and were > 5-fold higher relative to one circadian neuron group and >2-fold higher relative to the other . Gene Ontology analysis was performed using DAVID bioinformatics resources [76 , 77] . In all cases analyses were performed using a list of neuron-specific genes as a background comparison . A p-value of less than 0 . 05 was required in order for a gene ontology classification to be considered enriched . To identify potential neuropeptides that play a role in the circadian system , we identified those genes that were more highly expressed in each group of circadian neurons relative to the whole brain ( data for brains from [53] . The peptide sequence of all genes with higher expression in circadian neurons was obtained using FlyMine [78] and submitted to NeuroPID [46] http://neuropid . cs . huji . ac . il . NeuroPID was used to identify putative neuropeptide precursors that contained candidate signal peptides and were identified as high confidence predictions . NeuroPred ( http://stagbeetle . animal . uiuc . edu/cgi-bin/neuropred . py ) was then used to explore the cleavage sites of these neuropeptide precursors . A subset of those novel neuropeptide precursors that were identified were included in Fig 3A . To identify cycling transcripts , normalized transcript levels for two independent experiments ( 6 timepoints each ) generated by ESAT were used as input . Transcript expression values were normalized relative to the maximum signal in each set of 6 timepoints as previously described [79] . Cycling transcripts were identified using both fourier transformation [79] and JTK_cycle [80] . To be considered cycling using fourier transformation the following cutoffs were used: F24 score greater than 0 . 5 , >2 fold amplitude of transcript cycling , and the average transcript reads greater than 5 . JTK_cycle identified transcripts as cycling that had a >2 fold amplitude of transcript cycling , average transcript reads greater than 5 , and a p-value cutoff of less than 0 . 05 . The overlap of these two approaches ranged from 30–80% ( Sup Fig 1 ) . Those cycling transcripts identified by both methods were considered high-confidence cyclers ( HC cyclers ) and those identified by only one method were considered low-confidence cyclers ( LC cyclers ) . To examine whether LNv cycling transcript expression came primarily from the s-LNvs or l-LNvs , we utilized data from previously microarray studies [40] . The ratio of expression in l-LNvs versus s-LNvs was calculated and used as a metric for expression in the two neuronal subtypes .
|
Organisms ranging from bacteria to humans contain circadian clocks . They keep internal time and also integrate environmental cues such as light to provide external time information for entrainment . In the fruit fly Drosophila melanogaster , ~150 brain neurons contain the circadian machinery and are critical for controlling behavior . Several subgroups of these clock neurons have been identified by their anatomical locations and specific functions . Our work aims to profile these neurons and to characterize their molecular contents: what to they contain and how do they differ ? To this end , we have purified 3 important subgroups of clock neurons and identified their expressed genes at different times of day . Some are expressed at all times , whereas others are “cycling , ” i . e . , expressed more strongly at a particular time of day like the morning . Interestingly , each circadian subgroup is quite different . The data provide hints about what functions each group of neurons carries out and how they may work together to keep time . In addition , even a non-circadian group of neurons has cycling genes and has implications for the extent to which all cells have or do not have a functional circadian clock .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"invertebrates",
"neurochemistry",
"messenger",
"rna",
"neuroscience",
"animals",
"circadian",
"oscillators",
"hormones",
"animal",
"models",
"drosophila",
"melanogaster",
"model",
"organisms",
"experimental",
"organism",
"systems",
"chronobiology",
"neuropeptides",
"drosophila",
"research",
"and",
"analysis",
"methods",
"animal",
"cells",
"neurochemicals",
"gene",
"expression",
"insects",
"peptide",
"hormones",
"arthropoda",
"biochemistry",
"circadian",
"rhythms",
"rna",
"cellular",
"neuroscience",
"cell",
"biology",
"nucleic",
"acids",
"neurons",
"genetics",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"organisms"
] |
2017
|
RNA-seq analysis of Drosophila clock and non-clock neurons reveals neuron-specific cycling and novel candidate neuropeptides
|
Envenoming by viper snakes constitutes an important public health problem in Brazil and other developing countries . Local hemorrhage is an important symptom of these accidents and is correlated with the action of snake venom metalloproteinases ( SVMPs ) . The degradation of vascular basement membrane has been proposed as a key event for the capillary vessel disruption . However , SVMPs that present similar catalytic activity towards extracellular matrix proteins differ in their hemorrhagic activity , suggesting that other mechanisms might be contributing to the accumulation of SVMPs at the snakebite area allowing capillary disruption . In this work , we compared the tissue distribution and degradation of extracellular matrix proteins induced by jararhagin ( highly hemorrhagic SVMP ) and BnP1 ( weakly hemorrhagic SVMP ) using the mouse skin as experimental model . Jararhagin induced strong hemorrhage accompanied by hydrolysis of collagen fibers in the hypodermis and a marked degradation of type IV collagen at the vascular basement membrane . In contrast , BnP1 induced only a mild hemorrhage and did not disrupt collagen fibers or type IV collagen . Injection of Alexa488-labeled jararhagin revealed fluorescent staining around capillary vessels and co-localization with basement membrane type IV collagen . The same distribution pattern was detected with jararhagin-C ( disintegrin-like/cysteine-rich domains of jararhagin ) . In opposition , BnP1 did not accumulate in the tissues . These results show a particular tissue distribution of hemorrhagic toxins accumulating at the basement membrane . This probably occurs through binding to collagens , which are drastically hydrolyzed at the sites of hemorrhagic lesions . Toxin accumulation near blood vessels explains enhanced catalysis of basement membrane components , resulting in the strong hemorrhagic activity of SVMPs . This is a novel mechanism that underlies the difference between hemorrhagic and non-hemorrhagic SVMPs , improving the understanding of snakebite pathology .
Snakebite envenoming is an important neglected disease in many tropical and subtropical developing countries . As recently reviewed , globally , venomous snakebite is estimated to affect more than 421 , 000 humans per year , with 20 , 000 of fatalities . However , if we take into account the non-reported accidents , these data may be as high as 1 , 841 , 000 envenomings and 94 , 000 deaths [1] . Antivenom therapy was set at the end of 19th century and is still the only efficient approach to treat snakebites . It cures systemic symptoms of envenoming while the local effects are not covered and usually leads to temporary or permanent disability observed in many patients [2] , [3] . In Brazil , the majority of the accidents reported to the Ministry of Health are caused by viper snakes [4] . The victims of viper envenoming frequently present systemic disturbances in hemostasis including spontaneous bleeding and blood incoagulability , and strong local effects characterized by edema , ecchymoses , blisters and extensive hemorrhage [2] . Hemorrhagic toxins play an important role in vascular damage and subsequent generation of ischemic areas that largely contribute to the onset of local tissue necrosis that may result in amputation of affected limbs [5] , [6] . The pathogenesis of venom-induced hemorrhage involves direct damage of microvessels by the snake venom metalloproteinases ( SVMPs ) . They are multidomain Zn2+-dependent proteinases that share structural and functional motifs with other metalloproteinases , such as MMPs ( Matrix Metalloproteinases ) and ADAMs ( A Disintegrin And Metalloproteinase ) [7] , [8] . SVMPs are classified from PI to PIII according to their domains constitution ( Reviewed by Fox and Serrano [9] ) . The mature form of the PI class is composed only of the metalloproteinase domain with the characteristic zinc-binding site present in all classes of SVMPs , MMPs and some ADAMs . P-II and P-III SVMPs exhibit additional non-catalytic domains , such as disintegrin , disintegrin-like and cysteine-rich domains , similar to those found in ADAMs , which are related to adhesive properties [9] . Despite sharing similar catalytic activity , not all SVMPs induce hemorrhage in in vivo models . In general , P-III SVMPs that include disintegrin-like and cysteine-rich domains are potent hemorrhagic toxins while P-I SVMPs show reduced hemorrhagic activity . There are also a number of non-hemorrhagic SVMPs that may be found preferentially in the P-I class and rarely in P-III class , which often function as pro-coagulant enzymes [10] , [11] , [12] . The mechanism of hemorrhage induced by SVMPs has been investigated in several studies [13] , [14] , [15] , [16] , [17] . However , the precise molecular and cellular events associated with microvessel disruption remain unknown . The degradation of vascular basement membrane components has been proposed as a key event for the onset of capillary vessel disruption . The four major components of basement membranes are type IV collagen , laminin , nidogen/entactin and perlecan . Type IV collagen and laminin individually self-assemble into supramolecular structures , and both networks are crucial for basement membrane stability [18] . Nidogen/entactin and perlecan bridge the laminin and type IV collagen networks , increasing their stability , influencing the structural integrity of basement membranes [18] , [19] . Thus , the hydrolysis of vascular basement membrane components by SVMPs could profoundly affect the stability of the endothelium , resulting in bleeding . In this regard , in vitro , SVMPs efficiently degrade basement membrane components such as laminin , type IV collagen , nidogen/entactin , presenting minor effects on interstitial collagens [20] . However , catalytic activity is apparently similar in hemorrhagic and non-hemorrhagic SVMPs , indicating that the hydrolysis of basement membrane components is not the only mechanism acting on vascular damage induced by the hemorrhagic toxins . Endothelial cells have also been investigated as potential targets of hemorrhagic toxins . The survival signals promoted by endothelial cell anchorage can be disrupted by SVMPs using mechanisms dependent or independent of their proteolytic activity . Both P-I and P-III SVMPs interfere with adhesion components involved in focal adhesion between endothelial cells and the extracellular matrix , affecting the organization of actin filaments and stress fibers , which culminates in cell death by apoptosis [21] , [22] . However , apoptosis of endothelial cells shows little correlation with the hemorrhage induced by SVMPs . Hemorrhagic and non-hemorrhagic SVMPs induce comparable rates of apoptosis in endothelial cells in culture [23] and the onset of hemorrhage induced by SVMPs occurs much earlier than the induction of apoptosis of endothelial cells in vitro . In addition , no apoptosis of endothelial cells was observed in SVMP-induced hemorrhage in the dermis of mouse ear skin , in vivo [24] . An additional mechanism involved in SVMPs hemorrhagic activity could be related to the accumulation of hemorrhagic SVMPs close to capillary vessels , through the adhesive properties of the non-catalytic domains , allowing the hydrolysis of basement membrane components and disruption of the blood vessels . This mechanism has been suggested previously [14] , [16] , [25] , but up to the moment , there are no experimental evidences that this occurs in vivo . In this study , we evaluated this hypothesis analyzing the tissue distribution of jararhagin , a highly hemorrhagic P-III SVMP , and BnP1 , a weakly hemorrhagic toxin from P-I class , and the hydrolysis of basement membrane collagen and laminin within the hemorrhagic lesions , using the mouse skin as experimental model . We clearly showed a correlation between the binding of toxins to basement membrane and their ability to induce hemorrhage . Moreover , we showed the in situ degradation of basement membrane collagen IV , but not laminin , suggesting that collagen is an important target for the tissue accumulation of hemorrhagic SVMPs .
The conducts and procedures involving animal experiments were approved by the Butantan Institute Committee for Ethics in Animal Experiments ( License number CEUAIB 191/2004 ) . BALB/c mice ( 18–22 g body weight ) were used throughout the study . As a model for class P-I SVMP with low hemorrhagic activity , we used BnP1 ( GI:172044591 ) , a 25 kDa metalloproteinase , isolated from Bothrops neuwiedi venom according to Baldo et al . , [23] , which contains only the catalytic domain . Jararhagin ( GI:62468 ) was used as a model of highly hemorrhagic P-III SVMP , comprised of catalytic , disintegrin-like and cysteine-rich domains . It was isolated from Bothrops jararaca venom , as previously described [26] . Jararhagin-C was isolated from Bothrops jararaca venom as described [27] . It is devoid of catalytic activity and does not induce hemorrhage , as it contains only the non-catalytic domains of jararhagin – disintegrin-like and cysteine-rich . BSA ( Bovine serum albumin ) was used as control of non-toxic protein in different set of experiments . Toxins were used in the native form or labeled with Alexa Fluor 488® ( Molecular Probes , USA ) , following the instructions of the manufacturer . Groups of three BALB/c mice were intradermically injected with jararhagin ( 10 µg ) and BnP1 ( 5 µg or 50 µg ) , dissolved in 20 µL of PBS ( phosphate buffered saline ) . The control group received only 20 µL of PBS . After 15 minutes , the animals were sacrificed by CO2 inhalation and the dorsal skin corresponding to the site of the injection was carefully dissected out and fixed in methacarn solution ( 60% methanol , 30% chloroform , 10% glacial acetic acid ) for 3 hour at 4°C and thereafter dehydrated in ethanol and embedded in Paraplast ( Merck , Germany ) . Sections of 5 µm were adhered to glass slides using 0 . 1% poly-L-Lysine ( Sigma , UK ) and dried at room temperature . Sections were dewaxed in xylene and hydrated in distilled water . Each of the succeeding steps was followed by washing with PBS . Some sections were stained with hematoxylin and eosin for histological analysis . The detection of collagen fibers was performed by staining the sections using the picrossirus method , according to a previously described protocol [28] . The staining of basement membrane components was performed by immunofluorescence assays . After dewaxing and hydrating , the slides were submitted to antigen retrieval using enzymatic treatment of the sections with 4 mg/mL solution of pig pepsin ( 1 , 120 units/mg protein ) ( Sigma , UK ) in acid buffer ( pH 2 . 2 ) , for 10 minutes at room temperature , followed by incubation with blocking solution ( PBS/BSA 10% and goat serum 1∶1 ) for 1 hour at room temperature . Afterwards , the sections were incubated with goat anti-rabbit type IV collagen polyclonal antibody ( Chemicon , USA ) , at a 1∶40 dilution , or goat anti-rabbit laminin polyclonal antibody ( Chemicon , USA ) , at a 1∶40 dilution , for 18 hours at 4°C . After washing with PBS , the sections were incubated with Alexa fluor 488 goat anti-rabbit IgG ( Molecular Probes , USA ) , at a 1∶1000 dilution , for 90 minutes at room temperature . The nuclear staining was performed with DAPI ( 4′ , 6′-diamino-2-phenylindole , Sigma , UK ) , at a 1∶1000 dilution . Negative control consisted of omitting the primary antibody step from the protocol . The sections were examined with a Confocal Microscope LSM 510 Meta ( Zeiss , Germany ) . The distribution of toxins in the skin vasculature was evaluated after intradermically injection of 10 µg of Alexa Fluor 488-labeled jararhagin ( Alexa488-Jar ) , 5 µg of Alexa Fluor 488-labeled jararhagin-C ( Alexa488-Jar-C ) or 50 µg of Alexa Fluor 488-labeled BnP1 ( Alexa488-BnP1 ) in BALB/c mice . The control group received 20 µL of Alexa Fluor 488-labeled bovine serum albumin ( Alexa488-BSA ) . After 15 minutes , the animals were sacrificed by CO2 inhalation and the piece of dorsal skin on the site of injection was carefully dissected , frozen in OCT ( Optimal cutting temperature ) and sectioned 5 µm thick in the cryostat ( Leica , CM1510 ) . The sections were then fixed in 3 . 7% formaldehyde for 10 minutes at room temperature . Nonspecific staining was blocked by incubating the sections for 1 hour at room temperature with PBS containing 1% triton X-100 , 5% normal goat serum , 1% BSA , 0 . 5% glycine and 0 . 5% fish skin gelatin . Then , the sections were incubated with donkey anti-rat CD-31 polyclonal antibody ( BD Bioscience , USA ) , at a 1∶40 dilution , or goat anti-rabbit type IV collagen polyclonal antibody ( Chemicon , USA ) , at a 1∶40 dilution , for 18 hours at 4°C . The sections were washed with PBS and incubated with TRITC-labeled ( Tetramethyl Rhodamine Isothiocyanate ) goat anti-rat IgG ( Jackson ImmunoResearch , USA ) , at a 1∶500 dilution , or TRITC goat anti-rabbit IgG ( Jackson ImmunoResearch , USA ) , at a 1∶100 dilution , for 2 hours at room temperature . The distribution of toxins in the tissues was analyzed searching in at least 10 different fields of each section . Alternately , the distribution of jararhagin was analyzed 45 minutes after its injection under the same experimental condition . The sections were examined with a Confocal Microscope LSM 510 Meta ( Zeiss , Germany ) .
In order to investigate the mechanisms involved in the hemorrhage induced by SVMPs , we initially compared the pathological alterations induced by jararhagin , a highly hemorrhagic P-III SVMP and BnP1 , a weakly hemorrhagic P-I SVMP , using the mouse skin as experimental model . Tissues were inspected 15 minutes after injection of toxins in order to evaluate the first events involved in the hemorrhagic lesions . At this period , we focused on the direct action of venom toxins , avoiding interference of secondary effects of endogenous components released by the local reaction . After 15 minutes , macroscopic analysis of the skin injected with doses adjusted at the same molar basis ( 10 µg jararhagin and 5 µg BnP1 ) , revealed that jararhagin induced intense hemorrhage , whereas only a small hemorrhagic spot at the site of the injection was observed in the samples injected with BnP1 or PBS , used as injection control ( Fig . 1A ) . Only high doses of BnP1 ( 50 µg ) was able to induce hemorrhage , but less intense than jararhagin ( Fig . 1A ) . Morphological analysis under light microscopy showed drastic hemorrhage in the hypodermis and also in the skeletal muscle adjacent to hypodermis in mice injected with jararhagin ( Fig . 1B ) . The equivalent dose of BnP1 induced an enlargement of skin thickness probably due to its edema-forming activity but did not induce hemorrhagic alterations . When a hemorrhagic doses of BnP1 ( 50 µg ) was injected , edema was persistent and only sparse spots of hemorrhage were detected in the hypodermis ( Fig . 1B ) . We also evaluated the action of jararhagin and BnP1 on dermal-epidermal junctions by staining with antibodies anti-laminin and anti-β4 integrin and no alteration of these structures were observed after injection of toxins ( data not shown ) . These results confirm the differences in hemorrhagic activity of jararhagin and BnP1 and show that most of the hemorrhagic incidence occurs in the hypodermis . Next , we investigated the integrity of extracellular matrix components , mainly collagens and laminin , after injection of hemorrhagic doses of jararhagin ( 10 µg ) or BnP1 ( 50 µg ) . After 15 minutes , the control skin injected with PBS , showed a dense network of collagen fibers stained by picrossirus . The bundles of collagen fibers were closely packed characterizing a dense connective tissue ( Fig . 2 A , B ) . In contrast , mice injected with jararhagin showed a clear loosening of the bundles of collagen fibers in the dermis ( Fig . 2C ) . In the hypodermis , where the hemorrhagic lesion occurs , only a few weakly stained fibers were observed , indicating a massive degradation of fibrillar collagen ( Fig . 2D ) . BnP1 induced only a discrete disorganization of collagen fibers throughout the dermis ( Fig . 2E ) and the hypodermis ( Fig . 2F ) . The effect of toxins on the distribution of type IV collagen and laminin at the basement membrane was then evaluated by immunofluorescence . In the control skin , type IV collagen ( Fig . 3A ) and laminin ( Fig . 3B ) were observed as linear and continuous lines surrounding small blood vessels and in the basement membrane of skeletal muscle cells . In contrast , jararhagin induced a remarkable alteration in the immunostaining of type IV collagen in the basement membrane of blood vessels of the hypodermis , where only traces of type IV collagen deposition were detected . In addition , jararhagin also promoted a notable reduction of type IV collagen immunostaining in skeletal muscle basement membrane ( Fig . 3A ) . After injection of 50 µg of BnP1 , only a slight alteration in the immunostaining of type IV collagen was observed . The immunoreaction in the basement membrane of blood vessels and skeletal muscle was more diffuse than the pattern observed in the control , with the presence of some spots . However , the alterations induced by BnP1 are not comparable to the extensive disruption induced by jararhagin on type IV collagen , whose immunoreaction was practically abolished in blood vessels and skeletal muscle basement membrane ( Fig . 3A ) . The effect of jararhagin on laminin distribution was less intense than that observed on type IV collagen . The presence of laminin was detected in tissues injected with jararhagin , but its distribution on basement membrane was not as homogeneous as observed in control tissues suggesting punctual disruptions of basement membrane integrity , probably as a result of collagen degradation . Similar effects were induced by BnP1 ( Fig . 3B ) . The next step was to analyze the distribution of SVMPs in the skin tissue . After 15 minutes of injection , when the hemorrhagic lesion has already been set , jararhagin was located close to small blood vessels stained by anti-CD31 , in the hypodermis region ( Fig . 4B ) . It is interesting to note that co-localization with CD-31 was not observed , suggesting the accumulation of the toxin near the blood vessels . In contrast , after injection of BnP1 ( Fig . 4C ) , only a weak and diffuse fluorescence was observed , slightly higher than in control tissues ( Fig . 4A ) . Similar deposition of the toxins was observed in the skeletal muscle adjacent to the hypodermis . High fluorescence was observed after jararhagin injection ( Fig . 4E ) in the basement membrane of skeletal muscle and capillaries suggesting its accumulation in these areas . After BnP1 injection ( Fig . 4F ) , only a weak fluorescence was observed . No fluorescence was detected in control tissues ( Fig . 4D ) . Similar pattern of toxin distribution was detected up to 45 minutes after jararhagin injection in areas adjacent to the main focus of the injection . This toxin was detected around hypodermis blood vessels ( Fig . 5A ) , and close to the capillaries in the skeletal muscle ( Fig . 5B ) . In order to verify the binding of hemorrhagic toxins to the basement membrane , we carried out a double-staining protocol using Alexa488-labeled SVMPs and type IV collagen antibody . According to figure 6 , jararhagin showed co-localization with type IV collagen in the basement membrane of venules and capillaries , 15 minutes after injection . Contrarily , no co-localization with type IV collagen was observed in mouse skin injected with BnP1 , which showed a similar staining pattern to control samples ( Fig . 6 ) . These results confirm the particular binding of hemorrhagic toxins to basement membrane components , explaining their accumulation near blood vessels . We next addressed the role of non-catalytic domains of SVMPs in the distribution of jararhagin on tissues . For that , mice were injected with Alexa488- labeled jararhagin-C , which consists of jararhagin disintegrin-like and cystein-rich domains , and its distribution was observed in mouse skin . The distribution of jararhagin-C in skin was the same for jararhagin: jararhagin-C was detected close to the CD-31 marker around blood vessels in the hypodermis ( Fig . 7A ) , and close to capillaries in the skeletal muscle ( Fig . 7B ) , and co-localized with basement membrane type IV collagen in the hypodermis venules ( Fig . 7C ) . These results strongly suggest that the non-catalytic domains are determinant to hemorrhagic activity of SVMPs from P-III class , locating the catalytic site specifically to the microvascular wall .
In this work , we unveil an important step for understanding the mechanisms involved in the expressive hemorrhage induced by SVMPs , by comparing the in vivo degradation of extracellular matrix proteins and the tissue distribution of jararhagin , a highly hemorrhagic P-III SVMP , and BnP1 , a weakly hemorrhagic P-I SVMP , using the mouse skin as experimental model . This comparison revealed that tissue localization and in vivo degradation of collagens are key events in SVMPs induced hemorrhage . Jararhagin induced a massive degradation of fibrillar collagen in the hypodermis , where the hemorrhagic lesion was concentrated . In the vascular basement membrane , type IV collagen was the major substrate for jararhagin . In contrast , BnP1 was not able to efficiently degrade these substrates or to induce hemorrhage . Instead , a remarkable edema and dermal alteration were observed , consistent with the dermonecrotic activity already described for BaP1 , a SVMP class P-I isolated from B . asper venom [29] , . Considering that the macromolecular organization and the biomechanical stability of basement membrane are mainly determined by the type IV collagen network [31] , its cleavage would alter the structural stabilization of the other related basement membrane components . Consistent with that , we observed that jararhagin induced only slight alterations in laminin distribution , suggesting that its epitopes are conserved after treatment with the toxin . This characteristic is not restricted to snake venom pathology . Selective degradation of type IV collagen has been associated with other pathologies involving toxic metalloproteinases . The metalloproteinase from Vibrio vulnificus ( VVP ) is a major determinant for skin lesions of this microorganism , which also causes hypodermic hemorrhage [32] . According to the literature , both P-I and P-III SVMPs are similarly able to hydrolyze extracellular matrix components in vitro , such as matrigel [33] and isolated components , such as type IV collagen , laminin and fibronectin [20] , [33]–[37] . Escalante et al . [17] showed that jararhagin and BaP1 had a similar proteolytic activity on matrigel with a slightly different cleavage pattern , since BaP1 , exerted a limited proteolysis of both laminin and nidogen , whereas jararhagin predominantly degraded nidogen . However , the hydrolysis of extracellular matrix components in vitro occurs only after long incubation periods , suggesting that distinct mechanisms are involved in the basement membrane digestion in vivo . These authors also analyzed the immunostaining of laminin , nidogen , type IV collagen and the endothelial cell marker VEGFR-2 ( vascular endothelial cell growth factor receptor 2 ) in mouse gastrocnemius muscle injected with hemorrhagic doses of jararhagin and BaP1 , observing reduction in the number of capillary vessels and a similar pattern of immunostaining for the basement membrane components laminin , nidogen and type IV collagen in muscular fibers after injection of BaP1 or jararhagin , showing a disorganization of extracellular matrix [17] . Although they showed the first evidence of the catalytic action of SVMPs in vivo and morphological alterations in muscular tissues , the authors failed to detect any difference between weakly and highly hemorrhagic SVMPs . A parameter not yet explored in the literature consisted of eventual differences in the distribution of toxins in the damaged tissues . In this study , jararhagin , but not BnP1 , concentrated in the vicinity of small venules and in capillaries of skeletal muscle . This effect was correlated to the non-catalytic domains of jararhagin . In vitro , SVMPs bind to extracellular matrix proteins , such as type I collagen [15] , [38] , [39] , [40] , type IV collagen [40] , collagen XII and XIV and the matrilins 1 , 3 and 4 [16] . The high affinity for these extracellular matrix proteins could contribute for the accumulation of the toxin in the damaged tissue , enhancing the catalytic action of SVMPs towards basement membrane components . A fine correlation between collagen binding and hemorrhagic activity has been shown by our group . An anti-SVMP monoclonal antibody neutralizes hemorrhagic activity and collagen binding of jararhagin without interfering with its catalytic activity [25] . Recently , it was shown that jararhagin binds with high affinity to type I collagen and type IV collagen , whereas berythrativase , a non-hemorrhagic P-III SVMP isolated from B . erythromelas venom , failed to bind to these substrates [40] . Molecular modeling of the putative epitopes binding to this monoclonal antibody pinpointed a motif present in the hemorrhagic toxin jararhagin and absent in the pro-coagulant enzyme berythractivase , located at the Da-subdomain of disintegrin-like domain [40] . However , it is important to consider that a collagen-binding motif was also detected in the cysteine-rich domain of Atrolysin-A , a hemorrhagic P-III SVMP [15] . One aspect still unclear is the apparent contradiction between the results of collagen hydrolysis in vitro and in vivo . In vitro , most SVMPs ( class P-I or P-III ) hydrolyse type IV collagen but not fibrillar collagens [20] . Here , an almost complete disassembly of hypodermal fibrillar collagen was observed in tissues treated with jararhagin , but not BnP1 . Escalante and co-workers [41] also observed collagen hydrolysis in vivo . Several fragments of collagens were detected in the exudates of muscle injected with BaP1 , a class P-I SVMP . Most of the fragments corresponded to non-fibrillar collagen . However , a fragment corresponding to the fibrillar collagen V was also identified . According to our results , class P-I SVMPs induced minor alterations to hypodermal fibrillar collagen . Thus , it is possible to predict that skin homogenates of P-III SVMP lesions would contain more degradation fragments of fibrillar collagens . A possible explanation for the differences observed in vivo could be the disorganization of fibrils due to the toxin binding allowing the degradation of fibrillar collagen on hypodermis . However , data explaining why this occurs only in vivo is still lacking . Other alternative would be that the injection of SVMPs would induce tissue secretion of MMPs , able to digest fibrillar collagen . However , this hypothesis is not consistent with the fast onset of the reaction . Also , P-I SVMPs are very efficient to induce over-expression of MMPs [29] , [42] and they appear to be much less efficient to hydrolyze collagen in vivo . Our results confirm previous suppositions that the non-catalytic domains play a crucial role in the expression of hemorrhagic activity of P-III SVMPs , implying that non-enzymatic mechanisms are also involved in bleeding . The hypervariable region of the cysteine-rich domain is attributed to the binding of SVMPs to a series of substrates containing von Willebrand factor A domains , allowing the catalysis of a specific substrate region [16] . Here we suggest that SVMPs may present additional functional motifs related to their binding to collagens . The cysteine-rich domain exosite would be essential for the enzymatic selectivity of the SVMPs while a disintegrin-like domain collagen binding motif would be responsible for high affinity binding to collagens and tissue concentration of the toxin . The data presented herein are particularly important to understand the mechanisms involved in the onset of hemorrhage and could contribute to the rational of alternative treatments for snakebites victims . Since accumulation of SVMPs at the site of the bite allows hemorrhage enhancing the local venom effects , the local administration of metalloproteinases inhibitors could represent an interesting approach in order to improve the neutralization of toxins responsible for the local damage . Indeed , it has already been shown that the local injection of batimastat , a peptidomimetic matrix metalloproteinase inhibitor , totally neutralized the proteolytic , hemorrhagic and dermonecrotic effects induced by Bothrops asper venom [43] . Moreover , it has recently been shown that local administration of tetracycline prevented the dermonecrosis induced by Loxosceles spider venom . In addition to its antimicrobial properties , tetracycline was also able to inhibit MMPs , which are important for the progression of dermonecrotic lesions [44] . In summary , we showed that the strong hemorrhage induced by class P-III hemorrhagic SVMPs is related to their accumulation at basement membrane , reaching enzyme concentrations sufficient for its rapid degradation . This mechanism may serve as a rational for the design of alternatives in which local administration of metalloproteinase inhibitors may complement antivenoms in the neutralization of local tissue damage .
|
Snakebite accidents by vipers cause a massive disturbance in hemostasis and tissue damage at the snakebite area . The systemic effects are often prevented by antivenom therapy . However , the local symptoms are not neutralized by antivenoms and are related to the temporary or permanent disability observed in many patients . Although the mechanisms involved in coagulation or necrotic disturbances induced by snake venoms are well known , the disruption of capillary vessels by SVMPs leading to hemorrhage and consequent local tissue damage is not fully understood . In our study , we reveal the mechanisms involved in hemorrhage induced by SVMPs by comparing the action of high and low hemorrhagic toxins isolated from Bothrops venoms , in mouse skin . We show remarkable differences in the tissue distribution and hydrolysis of collagen within the hemorrhagic lesions induced by high and low hemorrhagic metalloproteinases . According to our data , tissue accumulation of hemorrhagic toxins near blood vessel walls allowing the hydrolysis of basement membrane components , preferably collagen IV . These observations unveil new mechanistic insights supporting the local administration of metalloproteinases inhibitors as an alternative to improve snakebite treatment besides antivenom therapy .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"cell",
"biology/extra-cellular",
"matrix",
"cell",
"biology/cell",
"adhesion",
"biochemistry/protein",
"chemistry"
] |
2010
|
Mechanisms of Vascular Damage by Hemorrhagic Snake Venom Metalloproteinases: Tissue Distribution and In Situ Hydrolysis
|
Reciprocal interactions between neurons and oligodendrocytes are not only crucial for myelination , but also for long-term survival of axons . Degeneration of axons occurs in several human myelin diseases , however the molecular mechanisms of axon-glia communication maintaining axon integrity are poorly understood . Here , we describe the signal-mediated transfer of exosomes from oligodendrocytes to neurons . These endosome-derived vesicles are secreted by oligodendrocytes and carry specific protein and RNA cargo . We show that activity-dependent release of the neurotransmitter glutamate triggers oligodendroglial exosome secretion mediated by Ca2+ entry through oligodendroglial NMDA and AMPA receptors . In turn , neurons internalize the released exosomes by endocytosis . Injection of oligodendroglia-derived exosomes into the mouse brain results in functional retrieval of exosome cargo in neurons . Supply of cultured neurons with oligodendroglial exosomes improves neuronal viability under conditions of cell stress . These findings indicate that oligodendroglial exosomes participate in a novel mode of bidirectional neuron-glia communication contributing to neuronal integrity .
In the CNS , oligodendrocytes insulate axons with a multilayered myelin sheath enabling rapid impulse conduction . Formation of functional axon-myelin units depends on bidirectional axon-glia interaction [1] , [2] . During nervous system development neuronal signals including activity-dependent neurotransmitter release control the differentiation of oligodendrocytes and myelination [3]–[5] . Axon-glia communication remains important throughout life . In addition to axon ensheathment , oligodendrocytes provide trophic support to neurons critical for long-term axonal integrity [6] . Glial support has been suggested to represent an ancestral function independent of myelination [7] . The mechanisms of neuron-glia communication essential to sustainably maintain and protect the highly specialized axon-glial entity over a lifetime are not well understood . Recent studies indicate that glycolytic oligodendrocytes provide axons with external energy substrates such as lactate [8] , [9] . These studies reveal new insights into axonal energy supply , although it remains still open how other resources ( such as enzymes of a certain half-life ) reach distal sites of axons . Oligodendrocytes release membrane vesicles with the characteristics of exosomes , which include specific myelin proteins such as proteolipid protein ( PLP ) [10] , [11] . Since exosomes have the capacity to affect neighboring cells , they have been generally implicated in intercellular communication [12] , [13] Exosomes of 50–100 nm in size are generated within the endosomal system and secreted upon fusion of multivesicular bodies ( MVBs ) with the plasma membrane . The exosomal membrane exhibits the topology of the plasma membrane and encloses cytoplasmic cargo . Most if not all cell types secrete exosomes and other microvesicles , budding from the plasma membrane . Consequently , body fluids such as serum , urine , and CSF contain significant amounts of mixed microvesicles , including exosomes [14] . Exosomes carry cell-type-specific components as well as common cargo , including proteins involved in MVB biogenesis , heat shock proteins , and integral membrane proteins such as integrins and tetraspanins . Furthermore , exosomes contain mRNA and miRNA , which upon horizontal transfer can alter protein expression , thus modulating the properties of recipient cells [15]–[17] . They have been described to contribute to immune responses , to spread pathogens such as viruses and prions , to modulate the tumor cell micro-environment , and furthermore to educate the phenotype of bone marrow cells [18]–[20] . Although cells exhibit a basal level of release , secretion of exosomes is a regulated process . Increase in cytoplasmic Ca2+ triggers exosome release from several cell types , including neurons and oligodendrocytes [10] , [21] , [22] . In this study , we analyze the role of exosomes in axon-glia communication . We demonstrate that neuronal activity-mediated release of the neurotransmitter glutamate regulates oligodendroglial exosome secretion by activation of glial ionotropic glutamate receptors . In turn , neurons internalize exosomes released from oligodendrocytes and retrieve their cargo . Furthermore , our results indicate that oligodendrocyte-derived exosomes mediate neuroprotective functions . These findings reveal a novel mode of cell communication among cells of the CNS that may be employed by oligodendrocytes to support axons .
Expression of Cre recombinase under control of a cell-type-specific promoter is utilized to drive the recombination of floxed target genes in a defined subset of cells within a tissue . MOGi-Cre mice carry Cre as a knock-in allele under control of the endogenous MOG promoter , which is described to be specifically active in the late stage of oligodendrocyte maturation [23] driving Cre expression in oligodendrocytes exclusively [24] , [25] . However , analysis of double transgenic MOGi-Cre/Rosa26-lacZ mice revealed reporter gene expression not only in oligodendrocytes but also in a subset of neurons in several brain regions ( Figure 1 ) . In the cerebellar granule cell layer , 17% of NeuN-labeled cells were positive for LacZ , while a lower number of recombined cells carrying neuronal markers were present in the cortex ( 3 . 8% ) , hippocampus ( 1 . 2% ) , and brainstem ( 2 . 9% ) . This finding may be explained by ( 1 ) activity of the MOG promoter in individual neurons or their precursors or ( 2 ) the horizontal transfer of Cre recombinase from oligodendrocytes to neurons . By q-PCR , MOG transcripts were either undetectable or at the detection limit in the embryonic brain and turned up during the first postnatal week coinciding with the appearance of mature oligodendrocytes ( Figure 1E ) . Therefore , it is unlikely that MOG-promoter activity in early embryonic progenitor cells is responsible for recombination persisting through neuronal differentiation into adulthood . The present study explores the possibility of horizontal transfer of molecules from oligodendrocytes to neurons by vesicles secreted from MVBs via the exosome pathway . Exosomes are generated by inward budding of the endosomal limiting membrane and stored in MVBs before release . Ultrastructural examination of optic nerve or spinal cord myelinated fibers by electron microscopy revealed that MVBs are present in the cytoplasm of oligodendrocytes , including the innermost uncompacted wrapping of the myelin membrane ( adaxonal loop ) in close proximity to the axon ( Figure 2A ) . In rare cases , we detected fusion profiles of MVBs indicating the release of intraluminal vesicles ( exosomes ) into the periaxonal space ( Figure 2B ) . The MVBs occasionally carried immunolabeling of LAMP1 ( not shown ) or PLP in the MVB limiting membrane and the intraluminal vesicles . The quantification revealed that MVBs are most prominent in the adaxonal loop at periaxonal sites , compared to their localization in the outer abaxonal loop or in channels between myelin lamellae ( Figure 2C ) . Among adaxonal MVBs , 29±8 . 8% carried immunolabeling for PLP . Taken together , these in vivo observations support the concept of exosomes playing a role in oligodendrocyte-neuron communication . We first asked whether oligodendroglial exosome secretion is regulated by neuronal signals . Our previous work revealed that elevation of cytosolic Ca2+ levels stimulates exosome release [10] . Since neurotransmitters such as glutamate mediate Ca2+ signaling in oligodendrocytes [26] , we investigated whether glutamate triggers oligodendroglial exosome release . Primary cultured oligodendrocytes were treated with glutamate for 5 h and exosomes were isolated from the supernatant by differential centrifugation . Oligodendroglial exosomes are known to include the myelin tetraspan protein PLP and its splice variant DM20 [10] , which was utilized to identify oligodendrocyte-derived exosomes by Western blotting . The amount of PLP/DM20 detected in 100 , 000× g pellets obtained from supernatants of glutamate-treated cells was significantly increased compared to untreated cells , while total PLP expression in the cells was unaffected . Consistently , we observed higher levels of the exosomal marker proteins Alix and Hsc/Hsp70 in the 100 , 000× g pellet obtained from glutamate stimulated cells , indicating that more exosomes were secreted ( Figure 2D ) . Previous studies showed that oligodendrocyte viability is affected by glutamate-mediated toxicity depending on intensity and duration of glutamate exposure as well as other side effects such as oxidative stress [27] , [28] . We observed no apparent damage of cultured oligodendrocytes 5 h after glutamate administration ( Figure S1A ) . To finally rule out that membrane fragments released from dying cells contaminated the exosome preparation , we investigated the integrity of the plasma membrane by evaluating lactate dehydrogenase ( LDH ) release and propidium iodide exclusion and found no detrimental influence of glutamate within the exosome collection period ( Figure S1B and S1C ) . Moreover , we utilized the ER-localized protein calnexin to determine contamination with nonexosomal membranes , which were virtually absent in most preparations ( Figure 2D ) . We further purified exosomes by sucrose density gradient centrifugation and detected PLP , Alix , and Hsc/Hsp70 in fractions of the typical density range reported for exosomes . Compared to untreated cells , the amounts of PLP , Alix , and Hsc/Hsp70 were increased , indicating that more exosomes had been isolated from glutamate-treated cultures ( Figure 2E ) . The slightly different position of the exosomal marker proteins in the gradient may reflect different exosome subpopulations , consistent with the current view of exosome heterogeneity [12] . Nanoparticle tracking analysis revealed that glutamate-stimulated cells release significantly more particles with a mean size of 95 nm , reflecting the expected size of exosomes ( Figure S1D ) . Notably , the size distribution of the released particles was not influenced by glutamate treatment . Examination of 100 , 000× g pellets derived from supernatants of glutamate-stimulated cells by electron microscopy identified larger aggregates of vesicles with the characteristic size and structure of exosomes as compared to pellets obtained from control cells , which released only few particles within the 5 h collection period ( Figure 2F ) . To obtain further proof that glutamate acts on exosome secretion , we performed siRNA silencing of the small GTPase Rab35 . Previous work has demonstrated that Rab35 regulates exosome secretion from oligodendrocytes [29] . Silencing of Rab35 in primary oligodendrocytes ( by 60±6 . 8% on the protein level ) interfered with the glutamate-dependent release of PLP and Alix demonstrating a specific effect of glutamate on exosome secretion ( Figures 2G and S1E ) . Thus , the particles released from oligodendrocytes in response to glutamate are released in a Rab35-dependent manner , and match the marker profile , size , and the density of exosomes . Next , we determined the dose dependence of glutamate-mediated exosome release . A concentration of 50 µM was sufficient to stimulate exosome release and saturation was reached at the dose of 100 µM ( Figure 2H ) . Moreover , we asked if extracellular Ca2+ is mobilized to stimulate glutamate-dependent exosome release and pre-incubated primary oligodendrocytes with EDTA , a nonmembrane permeable chelator of divalent cations , before exposure to glutamate . Depletion of divalent ions from the medium completely prevented stimulation of exosome release by glutamate , indicating that entry of extracellular Ca2+ through ionotropic glutamate channels is essential to trigger exosome release ( Figure 2I ) . Analysis of intracellular Ca2+ levels using a fluorescent calcium indicator showed that the cells responded to glutamate with a rise in intracellular calcium ( Figure S1F ) . Taken together , these results demonstrate that the neurotransmitter glutamate triggers Ca2+-dependent exosome release from oligodendrocytes . Glutamate-mediated Ca2+ influx across the oligodendroglial plasma membrane occurs through ligand-operated Ca2+ channels , such as NMDA and AMPA receptors [26] . Expression of functional receptors and their subunits by mature oligodendrocytes and their precursors has been described previously [30]–[33] . We confirmed their presence in differentiated oligodendrocytes in vitro by immunocytochemistry , with the expected localization of NMDA receptors in the membrane sheets , while AMPA receptors appear preferentially on cell bodies and primary processes ( Figure S2A , B ) [34] . To investigate if these receptors regulate oligodendroglial exosome release , we stimulated cells with receptor-specific agonists to provoke Ca2+ entry . NMDA as well as AMPA application induced a significant increase in exosome secretion . Moreover , administration of both agonists had a synergistic effect ( Figure 3A ) . Next , we co-applied glutamate with selective NMDA and AMPA receptor antagonists . MK801 and NBQX block oligodendroglial NMDA and AMPA receptors , respectively [32] . Both antagonists impaired glutamate-dependent exosome release ( Figure 3B ) , though NMDA receptor inhibition affected exosome release more potently . When we applied the known NMDA receptor co-agonist D-serine [35] together with glutamate , D-serine potentiated glutamate-dependent exosome release over glutamate alone ( Figure 3C ) . To obtain further evidence for the role of NMDA receptors in oligodendroglial exosome secretion , we used conditional knock-out mice lacking the essential NMDA receptor subunit NR1 selectively in oligodendrocytes ( CNP+/Cre*NR1flox/flox ) [36] , [37] . The floxed NR1-locus is recombined by Cre , which is expressed under control of the CNP promoter . Conditional NR1-null mice were heterozygous for ROSA26-flox-stop-EYFP , which we used as reporter for Cre expressing cells . As expected , primary cultured oligodendrocytes derived from these mice expressed EYFP confirming efficient recombination in oligodendrocytes ( unpublished data ) . Furthermore , NR1 was absent from mature oligodendrocytes expressing PLP ( Figure S2A ) . We compared the relative exosome release in response to glutamate in cells lacking NR1 ( CNP+/Cre*NR1flox/flox ) and in control cells ( CNP+/Cre*NR1+/+ ) and found that exosome release was not stimulated by glutamate in the absence of NR1 ( Figure 3D ) . These results suggest that NMDA receptors are essential for the glutamate-dependent secretion of exosomes from oligodendrocytes . Electrically active axons releasing glutamate evoke oligodendroglial Ca2+ signals [38]–[40] . To test if oligodendroglial exosome secretion is linked to neuronal electrical activity , we made use of a transwell device ( Boyden chamber ) , allowing contact-free co-culture of cells and exchange of metabolites by diffusion through pores ( Figure 4A ) . Cortical neurons grown for 7 d in vitro and placed on top of oligodendrocytes were depolarized with 20 mM potassium to trigger neurotransmitter release . Depolarization induced a significant increase in exosome secretion from oligodendrocytes ( Figure 4B ) . Exposure of oligodendrocytes to potassium in the absence of neurons was ineffective . In addition , we treated cortical neurons grown for 14 d to allow synapse formation with the GABAA-receptor antagonist bicuculline to block inhibitory activity and enhance spontaneous glutamatergic activity [22] . Again , oligodendroglial exosome release was strongly increased in response to enhanced neuronal electrical activity ( Figure 4C ) . Although bicuculline provoked a response of isolated oligodendrocytes , the stimulation of exosome release was 3 times more potent in the presence of neurons . To investigate if glutamate released by neurons acts on oligodendroglial NMDA receptors , we interfered with NR1 expression using RNAi . Silencing of NR1 in oligodendrocytes reduced the amount of secreted exosomes upon neuronal depolarization ( Figures 4D and S2C ) . We obtained similar results using oligodendrocytes derived from conditional NR1 null mice ( Figure S2D ) . These findings suggest that glutamate released by neurons in response to depolarization activates oligodendroglial glutamate receptors ( mainly of the NMDA receptor subtype ) stimulating exosome secretion . Thus , oligodendroglial exosome release is linked to neuronal electrical activity . We further studied the fate of the released exosomes and asked whether exosomes can be internalized by other neural cells . We utilized the transwell co-culture system and cultured oligodendrocytes labeled with the fluorescent dye PKH67 on porous filters , allowing the passage of exosomes . PKH67 is a lipophilic dye that is released in association with exosomes [41] . Cortical neurons or glial cultures were placed in the bottom chamber and imaged to visualize the uptake of fluorescent membrane particles in different neural cell types . We observed internalization of oligodendroglia-derived particles by 20 . 7±7% of the cortical neurons and by 93 . 4±3 . 5% of the microglia , while uptake by astrocytes and oligodendrocytes was detected only in 3±1% and 2 . 2±0 . 4% of the cells , respectively ( Figure 5A–D , see Figure S3 for overview ) . These data suggest that extracellular vesicles including exosomes released by oligodendrocytes are internalized preferentially by microglia and neurons . To confirm exosome uptake by neurons , we exposed primary cortical neurons to density gradient purified exosomes derived from the oligodendroglial cell line Oli-neu ectopically expressing SIRT2-EYFP and PLP-EGFP . Both proteins are sorted to oligodendroglial exosomes ( Figure 5E ) [10] . PLP-EGFP is located in the exosomal membrane , while SIRT2-EYFP is intraluminal . Incubation of neurons with these exosomes led to an uptake of fluorescent particles ( Figure 5E ) . Furthermore , neurons incubated with purified exosomes accumulated glial-specific , exosome-associated PLP and SIRT2 with time ( Figure 5F ) . Exosomes were indeed internalized , since removal of surface-bound exosomes by protease treatment before the lysis did not prevent the detection of PLP-EGFP/SIRT2-EYFP in neurons ( Figure 5G ) . Transfer of exosomes can also be visualized after co-culture of primary cortical neurons with Oli-neu cells in Boyden chambers . We utilized the neutral sphingomyelinase inhibitor GW4869 , which inhibits the release of exosomes [42] . Application of GW4869 to Oli-neu cells reduced the amount of secreted and thus internalized exosomes significantly , verifying that exosomes were responsible for the horizontal transfer of proteins and not other cell-derived particles ( Figure 5H ) . Moreover , the transfer appears to be selective , since untagged EGFP overexpressed in Oli-neu cells is not delivered to neurons ( Figure S4 ) . We further employed primary oligodendrocytes to examine exosome transfer to neurons . Exosome-associated PLP/DM20 as well as SIRT2 were detectable in neurons co-cultured with oligodendrocytes in Boyden chambers ( Figure 5I ) . Stimulation of oligodendroglial exosome release with glutamate significantly increased the level of PLP/DM20 and SIRT2 associated with neurons , demonstrating that neuronal uptake correlates with the amount of exosomes . To identify the subcellular destination of internalized exosomes , we performed double-labeling experiments of PKH67-stained glial exosomes together with endocytic markers . PKH67-stained exosomes accumulated within the neurons in endosome-like structures partly overlapping with LAMP1-positive late endosomes/lysosomes ( Figure 5J ) . A 3D reconstruction of confocal images confirmed that exosomes were located inside the neurons ( Figure S5A ) . To exclude that excess dye instead of stained exosomes was taken up by neurons , we pelleted PKH67-labelled exosomes and incubated neurons with exosomes and with exosome-depleted supernatant , respectively . PKH67-positive particles only associated with neurons when the exosome pellet was used , ruling out a transfer of excess dye ( unpublished data ) . To determine whether neurons internalize oligodendroglial exosomes by endocytosis , we pretreated neurons with Dynasore and Pitstop-2 inhibiting dynamin and clathrin-dependent endocytosis , respectively [43] , [44] . Dynasore and Pitstop2 application , as well as inhibition of actin polymerization by Cytochalasin D treatment , reduced the neuronal uptake of oligodendroglial exosomes ( Figures 5K and S5B , C ) . Impeding cholesterol-dependent ( clathrin-independent ) endocytosis by Methyl-β-Cyclodextrin application , which sequesters cholesterol from the plasma membrane , did not affect the uptake ( Figure S5C ) . Importantly , oligodendroglial cells secreted normal levels of exosomes in the presence of the inhibitors ( Figure S5D ) . Similar results were obtained utilizing the neuronal cell line HT22 ( Figure S5E–G ) . Expression of dominant negative dynamin K44A in HT22 cells inhibited endocytosis of Transferrin-Alexa568 and uptake of exosomes ( Figure S5H–J ) . These results indicate that neurons internalize oligodendroglial exosomes via an endocytic pathway that requires dynamin , clathrin , and actin polymerization . Targeting of oligodendroglial exosomes to endosomes and late endosomes in neurons may result in lysosomal degradation or the recovery of exosomal components . In addition to protein cargo , oligodendroglial exosomes include distinct RNA species ( Figure S6 ) . To analyze , if the exosomal cargo is functionally retrieved by neurons , we employed Cre-mediated recombination as a reporter . Primary oligodendrocytes were infected with a replication-deficient recombinant Adeno-associated virus ( AAV ) vector encoding Cre-recombinase under control of the oligodendrocyte-specific MBP promoter ( AAV/MBP-Cre ) prior to co-culture with neurons in transwells [45] . These neurons had been transduced with reporter virus delivering the ubiquitous chicken β-actin promoter , followed by a floxed transcriptional termination element and the GFP coding region ( AAV/CBA-floxstop-GFP ) . Reporter protein expression in neurons is specifically induced upon Cre-mediated excision of the floxed sequence [46] . Indeed , neurons acquired GFP reporter expression upon co-culture with AAV/MBP-Cre–infected oligodendrocytes ( Figure 6A , B ) . Inhibition of glial exosome secretion by the sphingomyelinase inhibitor GW4869 ( Figure 6C , F ) or siRNA-mediated knockdown of Rab35 ( Figure 6E ) interfered with neuronal GFP expression , while stimulation of oligodendroglial exosome release with glutamate enhanced reporter expression ( Figure 6D ) . Neurons incubated with oligodendroglial culture supernatant depleted from exosomes by 100 , 000× g centrifugation did not acquire GFP expression ( Figure 6F ) . Furthermore , co-culture of reporter-virus–infected neurons and AAV/MBP-Cre–infected HEK cells or Oli-neu cells , which do not synthesize MBP promoter-driven Cre , did not result in GFP expression ( Figure 6G ) , demonstrating that Cre expression in the donor cells is required and excluding a potential leakiness of the viral expression system . Consistent with these results , Cre protein , as well as Cre mRNA , were detectable in exosomes pelleted from supernatants of AAV/MBP-Cre–infected oligodendrocytes ( Figure 6H , I ) . Taken together , these results demonstrate that the cargo of glial exosomes internalized by neurons is functionally retrieved within the target cells . To provide the proof of principle that exosome transfer to neurons can occur in vivo , exosomes derived from AAV/MBP-Cre–infected oligodendrocytes were stereotactically injected into the cerebellum and hippocampus of adult Rosa26-lacZ reporter mice . Exosome internalization and cargo retrieval labels target cells as β-galactosidase-positive cells . In the injected hippocampus , individual neurons within the pyramidal cell layer of the CA3 region were positive for β-gal and neuron-specific enolase ( eight cells per injection , n = 5 , Figure 6J , K ) . In the cerebellum , we observed single recombined cells , which carried the GABA-Rα6 subunit , a marker of cerebellar granule cells ( five cells per injection , n = 5 , Figure 6L ) . The number of detected recombined neurons may be limited by spatial restraints of the injected exosomes and by the fact that the retrieval of Cre from exosomes requires several steps before recombination can take place . Recombined cells were only present in animals injected with Cre-exosomes and never detected in brains of control mice injected with glial exosomes lacking Cre or at the contra-lateral side of injection . These experiments demonstrate that oligodendroglial exosomes can be internalized by neurons in vivo and validate the concept that exosome-mediated transfer of Cre from oligodendrocytes to neurons may underlie the neuronal reporter gene recombination observed in oligodendrocyte-specific Cre-driver mice ( Figure 1 ) . We utilized microfluidic chambers to examine whether exosomes are internalized at the somatodendritic or the axonal domain . Cortical neurons cultured with their cell bodies in one compartment of the chamber grow axons through microgrooves into the other compartment of the chamber . Isolated exosomes labeled with PKH67 dye or containing Cre recombinase were added either to the somatodendritic or the axonal compartment and neuronal exosome uptake was monitored by imaging of exosomes or Cre reporter detection ( Figure 7 ) . Internalization of fluorescent exosomes was visible at the somatodendritic as well as at the axonal domain of the neurons . After application of Cre-bearing exosomes , we quantified the number of recombined neurons located within an area 100 µm from the microgrooves , thus focusing on neurons having the chance to project axons through the microgrooves into the axonal compartment ( not all neuronal cell bodies in this area will successfully grow axons through the microgrooves ) . The number of recombined neurons did not differ significantly upon somatodendritic or axonal addition of exosomes . Thus , uptake of exosomes is possible at both sites . To investigate if exosomes convey bioactivity to neurons , we performed viability assays on cultured neurons exposed to oligodendroglial exosomes . Oligodendrocytes and neurons were co-cultured in Boyden-chambers for 48 h , allowing a constant supply of the neurons with oligodendrocyte-derived factors including exosomes . Control cells were cultured without oligodendrocytes in oligodendrocyte pre-conditioned medium , deprived of exosomes but still containing secreted soluble factors . Subsequently , neurons were subjected to a MTT assay to measure their metabolic activity . Neurons grown under optimal conditions were not significantly affected by the presence of exosome-producing oligodendrocytes ( Figure 8A , unstressed ) . Intriguingly , when neurons were subjected to stress conditions such as oxidative stress ( 25 µM H2O2 for 1 h ) or nutrient deprivation ( culture in absence of B27 supplement ) , their metabolic activity was significantly increased in the presence of exosome-secreting oligodendrocytes ( Figure 8A ) . Neuronal metabolic activity was increased by 23 . 6±7 . 4% and 31 . 2±7% after oxidative stress exposure and nutrient deprivation , respectively . Oxidative stress was not alleviated when neurons were challenged by oxidative stress prior to co-culture ( not shown ) , indicating that exosome supply is protective but is not sufficient to rescue damaged neurons . MitoCapture staining demonstrates that co-culture of nutrient-deprived neurons with exosome-producing oligodendrocytes prevents the breakdown of the mitochondrial membrane potential ( Figure 8B ) . In a second approach , we directly incubated neurons with isolated exosomes or exosome-containing supernatants and determined their metabolic activity compared to untreated neurons ( Figure 8C , D ) . A single administration of isolated oligodendroglial exosomes 12–14 h prior to oxidative stress resulted in a significant increase in neuronal metabolic activity by 25 . 4±9 . 5% . In contrast , pretreatment of neurons with HEK293T-derived exosomes or artificial liposomes was ineffective , indicating a specific function of oligodendroglial exosomes . In the nutrient-deprivation paradigm , the metabolic activity of starving neurons increased by 13 . 8±3 . 1% upon incubation with oligodendrocyte-conditioned culture supernatant containing exosomes compared to supernatants deficient of exosomes ( Figure 8D ) , while exosome-containing HEK cell supernatant was ineffective ( not shown ) . A single application of isolated exosomes had only a minor supportive effect on the metabolic activity of starving neurons ( unpublished data ) . However , the level of metabolic activity in starving neurons was clearly dependent on the presence of oligodendroglial exosomes . The results of these experiments demonstrate that oligodendrocyte-derived exosomes support the neuronal metabolism under conditions of cell stress , suggesting a role in neuroprotection .
Evidence exists for activity-dependent release of glutamate along axons of white matter tracts occuring by vesicular means [47] , [48] and by reversal of Na+-dependent glutamate transporters [34] , [49] . On the glial side , glutamate receptors of the NMDA subtype appear to be preferentially localized to the oligodendrocyte periphery including myelin and the periaxonal membrane [32] , [33] , where they are ideally positioned to sense axonal glutamate release . Our results demonstrate that glutamate triggers exosome release from differentiated oligodendrocytes by activating Ca2+ permeable glial NMDA and AMPA receptors . While the physiological role of glutamate signaling to oligodendrocytes is still under debate , overstimulation is known to damage oligodendrocytes in pathological conditions due to excessive Ca2+ signaling [34] , [50] , [51] . Thus , a concern associated with our findings was that glutamate exposure could induce cell death and lead to the unspecific release of membrane fragments . We had no indication of cell disintegration or cell death within the period of exosome collection . Moreover , particle release was also induced by glutamate receptor agonists , which alone are not sufficient to induce oligodendroglial cell death [52] . Several studies show that glutamate-mediated cell death depends on receptor-independent secondary parameters such as the action of glutamate transporters , complement exposure , or oxidative stress [27] , [53] , [54] . In addition , the particles released in response to glutamate meet all criteria of exosomes ( homogeneous size below 100 nm , presence of ubiquitous and cell-type-dependent marker proteins , density ) and particle release was dependent on extracellular Ca2+ and the GTPase Rab35 , which have been shown to control oligodendroglial exosome secretion . Notably , exosomes released in constitutive and glutamate-stimulated fashion share similar biophysical characteristics , although it is possible that individual components differ . Our results suggest that exosome release occurs in response to neuronal electrical activity . Neuronal impulses have been implicated in the control of oligodendrocyte differentiation and myelination [55]–[57] . Oligodendrocyte precursor cells as well as mature oligodendrocytes respond to the firing of neighboring neurons with Ca2+ signals , which is mediated by glutamatergic and purinergic signaling [4] , [26] , [40] , [58] . It has been demonstrated that the release of glutamate from electrically active axons initiates translation of the major myelin protein MBP , thus promoting myelination [5] . Activity-stimulated exosome release may well accompany the myelination-promoting effects of neuronal impulses . We observed a dominant role of the NMDA receptor subtype in the control of glutamate-dependent exosome release , while AMPA receptor activation also played a role . Oligodendrocytes lacking the obligatory NMDA receptor subunit NR1 could not be stimulated by glutamate to release exosomes . Glial NMDA receptors are characterized by a low degree of voltage-dependent Mg2+ block [31] , which in neurons is released by AMPA receptor activation . It is possible that the role of AMPA-receptor signaling in exosome secretion is to reinforce Ca2+ influx through NMDA receptors by removal of Mg2+ ions from the channel pore . This could explain why AMPA receptor activation is sufficient to induce exosome release , while NMDA receptor activation appears necessary . Two recent studies have shown that the conditional deletion of oligodendroglial NR1 in mice does not affect oligodendrocyte differentiation and myelination [59] , [60] . Assuming the transfer of exosomes to neurons and the contribution of their cargo to the neuronal metabolism , one might anticipate that NMDA receptor deficiency in oligodendrocytes results in axonal degeneration . However , since exosome secretion occurs in a constitutive mode , a basal level of release will be maintained in these mice . Thus , an influence on axonal integrity would probably only become apparent when these mice are exposed to conditions of stress . Neurons and microglia internalize oligodendroglial exosomes effectively , whereas uptake by oligodendrocytes and astrocytes occurs only sporadically . Microglia take up exosomes unspecifically by macropinocytosis [61] , while internalization of oligodendroglial exosomes by neurons is mediated by selective clathrin- and dynamin-dependent endocytosis . As yet unknown receptors , mediating uptake of oligodendroglial exosomes , are likely to be present on neurons ( or even on a subset of neurons ) . In contrast to neurons , which recover the exosomal cargo functionally , microglia appear to degrade the exosomes . It is unclear how the cargo is retrieved from the endosome to reach its site of action . Injection of Cre-containing exosomes into the mouse brain provokes recombination in neurons , which involves functional Cre enzyme . This finding is consistent with the recombination of the reporter in neurons upon crossing of MOGi-Cre mice with Rosa26-lacZ mice , since Cre recombinase may be transferred via exosomes from neighboring oligodendrocytes . Only a minority of all neurons exhibits reporter activity in both cases , indicating that Cre packaging into exosomes or its retrieval in neurons is not very efficient . In the MOGi-Cre mice , however , we cannot fully exclude minimal MOG-promoter activity in early progenitors or mature neurons . Exosomes carry a multitude of molecules including proteins , lipids , and also RNA that potentially can affect target cells . In particular , the delivery of mRNAs and miRNAs by exosomes ( and other types of extracellular vesicles ) has been shown to modulate target cell functions [15]–[17] , [41] , [62] , [63] . Our results suggest that oligodendroglial exosomes support the neuronal metabolism and exhibit neuroprotective functions . Neurons treated with oligodendroglial exosomes were less sensitive to oxidative stress or starvation . At present , we can only speculate about the nature of the neuroprotective activity . Since exosomes are complex molecular entities and carry multiple enzymes with functions in metabolism or relieve of oxidative stress ( [10] and Table S1 ) , it is possible that several components contribute to this function . One of these factors may be Hsc/Hsp70 , which has a well-established role in neuroprotection and is known to be supplemented from adjacent glial cells [64] . Hsc/Hsp70 is sorted to oligodendroglial exosomes and also transferred to neurons . There is evidence in the literature that a transfer of molecules from glia to neurons takes place , but the mechanisms remain unknown . It has been shown that the squid giant axon receives Hsc/Hsp70 produced by periaxonal glial cells [65] and newly synthesized glial RNAs are delivered to the giant axon as a result of axonal depolarization [66] . Moreover , a classic study hypothesizes that voltage-dependent sodium channels are shuttled from Schwann cells to axons to replenish channels during their life cycle [67] . It is tempting to speculate that the exosome-dependent delivery of bioactive molecules contributes to the glial support of axons , which has been discovered in mice deficient in the myelin proteins PLP and 2′ , 3′-cyclic nucleotide 3′-phosphodiesterase ( CNP ) . These mice develop a secondary axonal degeneration despite the presence of morphologically normal myelin [2] , [36] , [68] . Two recent studies provide compelling evidence that after myelination , glycolytic oligodendrocytes support axons with energy metabolites such as lactate and furthermore , interference with oligodendroglial lactate export results in axonal degeneration [8] , [9] . However , it remains less clear how lactate supply is conceptionally linked to the lack of PLP and CNP in the mouse models of glial support . Exosomes carry both PLP and CNP as well as a heterogeneous group of enzymes and their secretion occurs independent of myelination ( as does glial support ) . It is possible that exosome transfer complements the energy supply of neurons by delivering metabolizing enzymes ( see Table S1 ) . The coupling of exosome transfer to axonal electrical activity would link this process to neuronal energy consumption . It will be interesting to analyze exosomes derived from PLP- and CNP-deficient oligodendrocytes with respect to their components , neuronal transfer capacity , and protective function . A vesicle-dependent pathway of axonal support may also exist in the PNS . It has been shown that myelinating Schwann cells supply injured and regenerating axons with ribosomal subunits in association with vesicles [69] , [70] . Exosome-mediated intercellular communication may be a more widespread phenomenon in the CNS involved in diverse physiological functions and pathological conditions [71]–[73] . Neurons also have the capability to secrete exosomes in an activity-dependent fashion , which has been suggested to modulate synaptic plasticity [22] . Microglia release plasma-membrane-derived vesicles that modulate synaptic activity by increasing neurotransmission [74] . Under brain inflammatory conditions such as EAE or multiple sclerosis , the amount of these vesicles in the CSF is increased [75] . Exosomes may also have detrimental effects under disease conditions and it has been proposed that they propagate pathogenic proteins through the CNS [76] . Several pathogenic proteins involved in CNS diseases , such as prions [18] , β-amyloid peptide [77] , superoxide-dismutase [78] , and α-synuclein [79] are released from cells in association with exosomes . Moreover , cellular cholesterol homeostasis appears to be regulated by exosome release [80] . Notably , there is accumulating evidence that exosomes can pass the blood brain barrier and thus are promising tools for diagnostic and therapeutic applications in the future [16] , [81] . In conclusion , this study reveals a novel mode of bidirectional neuron-glia interaction that couples neuronal activity to the external supply of neurons with glia-derived bioactive molecules . We suggest a role of exosome transfer from glia to neurons in metabolic support of neurons and thus neuroprotection .
Experiments were in compliance with the animal policies of the University of Mainz , approved by the German Federal State of Rheinland Pfalz , in accordance with the European Community Council Directive of November 24 , 1986 ( 86_609_EEC ) . Mouse strains: ( 1 ) wild-type C57Bl/6N , ( 2 ) CNPCre/+ [36] , ( 3 ) oligodendroglial NR1 knockout mice CNP+/Cre*NR1flox/flox [37] . Conditional NR1-null mice were also heterozygous for ROSA26-flox-stop-EYFP [82] , which we used as reporter for Cre expressing cells . ( 4 ) MOG+/Cre*R26R+/lacZ [24] , [83] . Mice of either sex of the strain C57Bl/6N were used for preparation of primary oligodendrocyte ( pOL ) cultures . Primary oligodendrocytes , mixed glial cells , and the cell line Oli-neu were prepared and cultured in Sato 1% HS as described [84] . Purity of the cultures was assessed by regular quality controls ( immunostaining ) and varied between 80% and 95% . Contaminating cells were of astroglial ( 3%–15% ) , microglial ( 1%–5% ) , and neuronal ( 5%–10% ) origin . Cortical neurons were prepared from E14 embryonic mice as described [85] and were essentially glia free with less than 1% of contaminating astrocytes . HEK293T cells were cultured in DMEM+10% FCS and 1 mM sodium pyruvate . HT22 cells were cultured in DMEM+10% FCS . rAAV plasmids contained the chicken β-actin promoter ( CBA ) followed by a transcriptional termination sequence and the cDNA encoding hrGFP [45] , [46] , or the 1 . 3 kb mouse MBP-promoter driving Cre-recombinase . The AAV cassettes containing the woodchuck hepatitis virus posttranscriptional regulatory element ( WPRE ) and the bovine growth hormone polyadenylation sequence ( bGHpA ) were flanked by AAV2 inverted terminal repeats . Mosaic AAV1/2 vectors were produced by transfection of HEK cells as described and purified by discontinuous iodixanol centrifugation [86] . Genomic titers were determined by PCR using primers against WPRE and vectors were adjusted at 5×1011 vector genomes/ml . Primary oligodendrocytes and neurons were infected by adding 1 µl ( 5×107–8 vg ) to the cell culture medium . Antibodies used were as follows: rat PLP ( clone aa3 ) , mouse AIP1/Alix ( 49; BD ) , mouse Hsc/Hsp70 ( Santa Cruz ) , mouse CD9 ( BD Transduction labs ) , rabbit calnexin ( Stressgen ) , mouse NR1 ( Millipore ) , mouse NR2B ( N59/20; Neuromab ) , rabbit NR3A ( Millipore ) , rabbit GluR3 ( Alamone Labs ) , rabbit GluR4 ( Millipore ) , mouse GFAP ( 1B4; BD ) , rat F4/80 [87] , mouse NeuN ( Chemicon ) , mouse O4 and mouse O10 [88] , mouse Tsg101 ( 4A10; GeneTex ) , rabbit GFP ( Abcam ) , mouse αTub ( DM1A; Sigma ) , rabbit TUJ1 ( 1-15-79; Covance ) , rat LAMP1 ( 1D4B; BD ) , rabbit SIRT2 ( Abcam ) , rabbit hrGFP ( Stratagene ) , rabbit Rab35 ( Proteintech ) , Carbocyanin and HRP secondary antibodies ( Dianova ) , and Alexa secondary antibodies ( Invitrogen ) . Expression plasmids pPLP-EGFP [42] and pSIRT2-EYFP ( this study ) were used . SIRT2 sequence was amplified from genomic DNA with PCR ( primers 5′-CCGCTCGAGCGATGGACTTCCTGAGGAATTTATTC-3′ and 5′-GGCGAATTCTCTGCTGTTCCTCTTTCTCTTTG-3′ ) restriction digested and inserted into pEYFP-N1 ( Clonetech ) between XhoI and EcoRI restriction sites . pEGFP-DynK44A was a kind gift from Luise Florin ( University Medical Center Mainz , Department of Medical Microbiology ) . Reagents used were as follows: NMDA , AMPA , MK801 , NBQX , bicuculline ( Tocris ) , GW4869 , Dynasore , Methyl-β-Cyclodextrin , Cytochalasin D , PKH67 , PKH26 , glutamate , D-serine ( Sigma ) , Pitstop2 ( Ascent Scientific ) , and Alexa568-Transferrin ( Invitrogen ) . Transient transfections of Oli-neu cells were performed by electroporation using the Bio-Rad Gene Pulser X-Cell [89] . pOL were transfected with Rab35 siRNA ( Dharmacon L-042604-01 ) using Amaxa technology ( Lonza ) and the Amaxa Nucleofector Kit , Primary Neurons ( program O-005 ) . Transfection of pOL with NR1 siRNA ( Dharmacon L-045931-01 ) was carried out using Lipofectamine RNAiMAX ( Invitrogen ) . Exosomal RNA was isolated from 100 , 000× g pellets using miRNA high affinity kit ( Roche ) . cDNA was synthesized using the Quantitect reverse transcription kit ( Qiagen ) , and Cre and Pgk1 were amplified by PCR . Total brains derived from E10 , E14 , P0 , P7 , and adult C57Bl/6N mice were homogenized using TissueRuptor ( Qiagen ) . Total RNA was prepared with the miRNeasy kit ( Qiagen ) , cDNA was synthesized , and qPCR was performed with a StepOne instrument utilizing TaqMan assays recognizing MOG , NG2 , and normalizers Pgk1 and β-actin ( Applied Biosystems ) . Culture supernatants were collected from Oli-neu cells ( between div 1–3 ) and pOL ( between div 5–8 ) and exosomes were isolated by differential centrifugation and flotation density gradient centrifugation as described before [10] with modifications . Briefly , culture supernatants of pOL or Oli-neu cells were collected and cleared from debris by successive centrifugation for 10 min at 60× g and for 20 min at 10 000× g ( 4°C ) . Membrane particles remaining in the supernatant were pelleted by ultracentrifugation for 1 h at 100 , 000× g and 4°C using the SW40 rotor ( Beckman ) . For analysis , the pellet was resuspended in SDS-PAGE sample buffer and subjected to 12% or 10% SDS-PAGE and Western blotting . For nanoparticle tracking analysis , 100 , 000× g pelleted exosomes were resuspended in PBS and analyzed using the Nanosight LM10 system . To prevent aggregation of exosomes and loss of material due to tube adherence , membrane particles were centrifuged onto a 200 µl sucrose cushion ( 1 . 8 M in TBS ) for 1 h at 100 , 000× g and 4°C . For neuronal uptake studies , the cushion including floating exosomes was diluted to a final concentration of 250 µM sucrose and applied to CN . For further purification of exosomes , the diluted cushion was loaded on top of a continuous gradient ( 0 . 3–1 . 8 M sucrose in TBS ) followed by centrifugation for 16 h at 100 , 000× g and 4°C . pOL ( 5×106 ) differentiated in vitro for 7 d were washed with HBSS . Glutamate ( 50 , 100 , 200 µM ) , NMDA ( 100 µM ) , or AMPA ( 100 µM ) was added to the cells in Sato 1% HS and reapplied for further 1 h . To inhibit or potentiate glutamate-mediated exosome release pOL were preincubated with MK801 ( 5 µM ) , NBQX ( 25 µM ) , D-serine ( 100 µM ) , or EDTA ( 5 mM ) for 5 min . Exosome pellets and cell lysates were prepared and analyzed as described above . For determination of relative exosome release , x-ray films were scanned and PLP Western blot signals were quantified using ImageJ software ( National Institutes of Health ) . The amount of exosomal PLP was normalized to total cellular PLP . If ER contaminations occurred , the values obtained were further normalized to levels of the ER protein calnexin . Cells were scraped in 10 mM Tris pH 7 . 4 , 150 mM NaCl , 1 mM EDTA , 1% Triton X-100 , and protease inhibitor cocktail ( Roche complete ) or PMSF on ice . Nuclei were pelleted by centrifugation for 10 min at 300× g . Cell lysates and exosome samples were subjected to 10% or 12% SDS-PAGE and were blotted onto a PVDF membrane , which was blocked with 4% milk powder/0 . 1% Tween in PBS . Proteins were detected by sequential incubation of the membrane with primary and HRP-coupled secondary antibodies and developed with enhanced chemiluminescence reagents ( Pierce ) . Immunocytochemical staining of cells was performed as described [85] . For immunohistochemical analysis , adult mice were fixed by transcardial perfusion with 4% formaldehyde ( Serva ) , and brains were cut into 30 µm slices on a vibratome and stained according to standard protocols . Fluorescence images were acquired using a fluorescence microscope ( DM6000 , Leica ) or a confocal laser scanning microscope ( Zeiss , Axiovert LSM 710 ) and processed with ImageJ software ( National Institutes of Health ) . Immunoelectron microscopy was performed as described [85] . Briefly , adult mice were fixed by perfusion with 4% formaldehyde ( Serva ) and 0 . 2% glutaraldehyde ( Science Services ) in 0 . 1 M phosphate buffer containing 0 . 5% NaCl . Slices of spinal cord and gelatin-embedded pieces of optic nerves were infiltrated in 2 . 3 M sucrose in 0 . 1 M phosphate buffer overnight; pieces from the area of the dorsal column and optic nerve samples were mounted onto aluminum pins for ultramicrotomy and frozen in liquid nitrogen . Ultrathin cryosections were prepared using a cryo-ultramicrotome ( UC6 equipped with a FC6 cryobox , Leica ) . Sections were incubated with primary antibodies followed by protein A-gold ( 15 nm ) and analyzed with a LEO EM912 Omega ( Zeiss ) . Digital micrographs were obtained with an on-axis 2048×2048-CCD camera ( TRS ) . For quantitation of multivesicluar bodies , cryosections of optic nerves from three different animals were immunolabeled with anti-PLP and protein A-gold ( 10 nm ) . Ten large images were taken from every sample at 8 , 000× magnification each covering an area of 11 . 5 µm×11 . 5 µm , summing up to a total area of 1 , 322 . 5 µm2 per sample . On every image , the number , location , and labeling of MVBs were analyzed and the number of axons determined . The number of MVBs is expressed relative to the number of axons in the imaged field . After differential centrifugation exosomes derived from 12×106 cells were pelleted directly on a 75 meshes Ni-Grid , fixed with 4% PFA , and analyzed by EM after washes with a . dest . and embedding in 2% methylcellulose with 0 . 4% uranyl acetate . Oligodendrocyte/neuron co-cultures were performed in Boyden-Chambers , which allow contact-free culture while permitting exchange of particles through 1 µm pores in a filter membrane ( six-well companion plates ( 353502 ) , six-well cell culture inserts ( 353102 ) , BD Falcon ) . Cells growing on filter membranes in the culture inserts ( top well ) were manipulated , while effects on exosome release or exosome uptake were studied on cells growing in the companion plates ( bottom well ) . To study the effect of neuronal activity on oligodendroglial exosome release , CN were placed on top of pOL and 20 mM KCl or 60 µM bicuculline was added to the top well . After 20 min ( bicuculline ) or 3 times 2 h ( KCl ) , exosomes were isolated from the supernatant of the bottom well by differential centrifugation and analyzed by Western blotting . For investigation of exosomal transfer between oligodendrocytes and neurons Oli-neu cells expressing PLP-EGFP and Sirt2-EYFP from plasmids or pOL were grown on top of CN for 1–3 d . pOL were stained with the dye PKH67 according to the manufacturer's instructions and extensively washed , or infected with AAV/MBP-Cre . In the latter case , CN were infected with a recombinant AAV1/2 virus carrying a flox-stop-hrGFP cassette ( AAV/CBA-floxstop-GFP ) as reporter construct for the detection of Cre activity [46] . PKH67 fluorescence or reporter gene expression in neurons was visualized by fluorescence microscopy or Western blotting and quantified using ImageJ . To examine neuronal uptake of exosomes by endocytosis , CN and HT22 cells were pre-treated with inhibitors for 30 min ( Dynasore , 50 or 100 µM; Pitstop2 , 30 µM; CytochalasinD , 10 µM; Methyl-β-Cyclodextrin , 500 µM ) and subsequently co-cultured for 24 h with pOL and Oli-neu cells , respectively . Exosome transfer was assessed by Western blotting or fluorescence microscopy . HT22 transfected with pEGFP-DynK44A were co-cultured with Oli-neu cells ectopically expressing PLP-EGFP and Sirt2-EYFP or stained with PKH26 for 24 h . As positive control for endocytosis Alexa568-transferrin ( Alexa568-Tf ) was used . HT22 cells expressing pEGFP-DynK44A were incubated with Alexa568-Tf for 24 h and analyzed by fluorescence microscopy . To discriminate between axonal or somatodendritic uptake of exosomes , microfluidic chambers with a microgroove length of 150 µm were used ( Xona , standard neuron device 150 µm , Cat# SND150 ) . CN were plated into the device according to the manufacturer's protocol . After 7 div exosomes of PKH67 stained or AAV/MBP-Cre–infected pOLs were either applied to the somatodendritic or axonal compartment of the device . In each case the opposite compartment was filled with a larger volume creating a hydrostatic pressure , thereby fluidically isolating each chamber . In case of the addition of Cre exosomes , CN were infected with a reporter virus as described above . For quantification of uptake efficiency , recombined hrGFP-positive neurons located with their cell bodies in an area of 100 µm above the microgrooves were counted . Two- to five-month-old male and female ROSA26-lacZ reporter mice were anesthetized with ketamine and xylazin and placed in a stereotactic frame . Exosomes prepared from AAV/MBP-Cre–infected pOL by differential centrifugation and dissolved in PBS were injected in the right hippocampus ( coordinates in relation to bregma: anteroposterior 1 . 8 mm , mediolateral 1 . 8 mm , dorsoventral 1 . 8 mm ) and cerebellum ( coordinates: anteroposterior 6 . 2 mm , mediolateral 2 . 0 mm , dorsoventral 2 . 0 mm ) . As a control , exosomes prepared from uninfected pOL were injected . A volume of 2 µl was injected using a glass microcapillar and a motorized injection pump ( World Precision Instruments ) at a constant flow rate of 250 nl/min starting 120 s after injection . To limit fluid reflux along the injection track , the needle was kept in place for an additional 5 min after injection . Mice were sacrificed 14 d after injection and perfused with PFA . Brains were vibratome cut ( 30 µm slices ) . LacZ and immunofluorescence staining was performed according to standard protocols . In co-culture assays , cortical neurons ( CN , 0 . 7×106/6-well ) were cultured for 48 h in neurobasal/B27 medium with primary oligodendrocytes ( pOL , 1 . 25×106/insert ) in Boyden-Chambers . After or during the co-culture period , neurons were challenged by oxidative stress or nutrient deprivation , respectively . Oxidative stress was induced by incubation with 25 µM H2O2 for 1 h , nutrient deprivation by exposure to culture medium lacking B27 supplement for 48 h . In the controls , neurons were grown in the presence of blank inserts containing oligodendrocyte-conditioned medium deprived of exosomes by centrifugation at 100 , 000× g . Neuronal viability was assessed by the MTT assay: 0 . 75 mg/ml 3- ( 4 , 5-dimethylthiazol-2-yl ) -2 , 5-diphenyltetrazolium bromide ( MTT , Sigma ) was added to the medium for 2 h . Formazan crystals formed were solubilized in a buffer containing 40% [v/v] dimethyl-formamide ( Sigma ) , 10% [w/v] SDS , and 2% [v/v] acetic acid overnight . The absorbance was measured at 562 nm using a plate reader ( Tecan Infinite 200 ) . Cells were stained with MitoCapture-dye to visualize the breakdown of the mitochondrial membrane potential utilizing the MitoCapture Apoptosis Staining Kit ( PromoCell ) . Cells were imaged by fluorescence microscopy ( DM6000 , Leica ) . To incubate neurons with exosomes , exosome-containing supernatants or exosome pellets were prepared . Briefly , supernatants were collected over 24 h from pOL ( 6×106 ) cultured in neurobasal medium and depleted from cellular debris by subsequent centrifugation for 10 min at 60× g and 20 min at 10 , 000× g to generate exosome-containing supernatants . A further round of centrifugation for 1 h at 100 , 000× g was performed to yield exosome-deprived supernatants that were used as control . Exosome-containing or exosome-deprived supernatants were transferred to CN ( 0 . 7×106/6-well ) and incubated for 12–14 h before assessment of viability by MTT assay . Exosomes from pOL and HEK293T cells as well as liposomes ( Liposome Kit by Sigma , preparation according to manufacturer's protocol ) were pelleted by differential centrifugation and resuspended in PBS . The number of containing particles was determined by the Nanosight LM10 system to normalize the number of exosomes and liposomes used for treatment ( aprox . 6 , 000 particles/neuron ) . Exosomes and liposomes were added to CN 12–14 h before the exposure to oxidative stress and subsequent viability assessment . To analyze membrane integrity , pOL ( 7 d in culture ) were incubated for 5 h with glutamate in different concentrations ( 50 , 100 , 200 µM , 5 mM ) and subjected to LDH assay ( Roche ) . The LDH assay was carried out as described in the manufacturer's protocol . For positive controls , cells were incubated with 10 mM NaN3 or with H2O2 prior to the experiment for 1 h . For PI exclusion , pOL growing on coverslips for 7 d were treated with 100 µM glutamate or 2 mM H2O2 as positive control for 5 h . Cells were incubated with PI ( 5 µg/ml , Sigma ) for 15 min , fixed for 10 min with 4% PFA in PBS , and treated with RNase ( 100 µg/ml ) for 20 min at 37°C . Cells were stained with O10 antibodies recognizing PLP followed by incubation with α-mouse-Alexa-488 and DAPI . DAPI and PI positive cells were counted and the proportion of PI stained cells was calculated . pOL were grown in a 96-well plate for 7 d . Cells were incubated with 10 µM Oregon Green 488 BAPTA 1 , AM ( Molecular Probes ) in Sato for 30 min at 37°C and stimulated with 100 µM glutamate or 2 µM ionomycin . Fluorescence was recorded over 1 h using a Tecan Infinite M1000 reader ( excitation 494 , emission 523 ) . Significance was calculated by the nonparametric Wilcoxon test ( paired , two-tailed ) in cases n>5 . Analysis was performed with the PASW statistics 18 software ( SPSS Statistics , IBM ) .
|
Brain function largely depends on the communication between electrically excitable neurons and surrounding glial cells . Myelinating oligodendrocytes are a type of brain cell that insulate major neuronal processes ( axons ) and help to sustainably maintain axonal health , which is poorly understood in molecular terms . Several cell types release microvesicles termed exosomes that include genetic information ( primarily RNA ) and can act as vehicles transferring specific cargo to target cells . Here , we demonstrate that exosomes secreted by oligodendrocytes in response to neuronal signals enter neurons to make their cargo functionally available to the neuronal metabolism . We revealed in cultured cells that exosome release from oligodendrocytes is triggered by the neurotransmitter glutamate through activation of ionotropic glutamate receptors . We also show that glial exosomes are internalized by neurons via an endocytic pathway . By modifying oligodendroglial exosomes with a reporter enzyme , we could demonstrate that the exosome cargo is recovered by target neurons in culture as well as in vivo after injection of exosomes into the mouse brain . Neurons challenged with stressful growth conditions were protected when treated with oligodendroglial exosomes . The study introduces a new concept of reciprocal cell communication in the nervous system and identifies the signal-mediated transfer of exosomes from oligodendrocytes to neurons contributing to the preservation of axonal health .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"immunocytochemistry",
"medicine",
"cellular",
"structures",
"subcellular",
"organelles",
"molecular",
"neuroscience",
"neurochemistry",
"multiple",
"sclerosis",
"macromolecular",
"assemblies",
"neuroscience",
"demyelinating",
"disorders",
"biological",
"transport",
"neurotransmitters",
"cytochemistry",
"membranes",
"and",
"sorting",
"biology",
"biochemistry",
"organelles",
"cellular",
"neuroscience",
"neuromodulation",
"neurology",
"neurons",
"cellular",
"types",
"molecular",
"cell",
"biology",
"metabolism"
] |
2013
|
Neurotransmitter-Triggered Transfer of Exosomes Mediates Oligodendrocyte–Neuron Communication
|
Protein domains shallowly inserting into the membrane matrix are ubiquitous in peripheral membrane proteins involved in various processes of intracellular membrane shaping and remodeling . It has been suggested that these domains sense membrane curvature through their preferable binding to strongly curved membranes , the binding mechanism being mediated by lipid packing defects . Here we make an alternative statement that shallow protein insertions are universal sensors of the intra-membrane stresses existing in the region of the insertion embedding rather than sensors of the curvature per se . We substantiate this proposal computationally by considering different independent ways of the membrane stress generation among which some include changes of the membrane curvature whereas others do not alter the membrane shape . Our computations show that the membrane-binding coefficient of shallow protein insertions is determined by the resultant stress independently of the way this stress has been produced . By contrast , consideration of the correlation between the insertion binding and the membrane curvature demonstrates that the binding coefficient either increases or decreases with curvature depending on the factors leading to the curvature generation . To validate our computational model , we treat quantitatively the experimental results on membrane binding by ALPS1 and ALPS2 motifs of ArfGAP1 .
Lipid bilayers serving as matrices of biological membranes bear internal elastic stresses . These stresses can be generated by external forces applied to the membrane surface and driving overall membrane deformations such as generation of membrane curvature and stretching-compression of the membrane area [1] , and/or by internal factors such as elastic frustrations , which are intrinsic to the membrane structure [2] . Insertion into the membrane matrix of protein domains spanning completely or partially the lipid bilayer interior must interfere with the intra-membrane stresses . This has to result , on one hand , in the stress-dependence of the energy of the protein insertion into the membrane and , on the other , in alteration of the intra-membrane stresses . The former phenomenon results in the stress sensing by these protein domains , which can be manifested as stress-dependence of the protein partitioning between the membrane and the surrounding aqueous solution [3] and/or as regulation by the stresses of the protein conformational transitions and the related protein activity within the membrane ( see e . g . [4] ) . Alteration of the membrane stress caused by the protein embedding can affect the membrane conformation , e . g . by changing membrane curvature [5] , [6] . During the last decade , one of the hot topics discussed in the biophysical literature , and referred to as the curvature sensing by proteins , has been the ability of a number of peripheral membrane proteins to bind preferentially to small liposomes with radii of several tens of nanometers . Commonly for most of these curvature sensing proteins , their binding to membranes has been mediated by shallow insertion into the membrane matrix of an amphipathic or hydrophobic domain [7] . In most cases , such domain is an amphipathic helix [8] , but can also be a short hydrophobic loop [9] . The curvature sensing has been demonstrated for numerous proteins involved in intracellular membrane shaping and remodeling such as the N-BAR ( Bin-amphiphysin-Rvs ) domain-containing protein amphiphysin playing a key role in endocytosis [10]; the GTPase dynamin driving membrane fission [9]; synaptotagmin implicated in membrane fusion [11]; α-synuclein [12]; the lipid droplet enzyme CTP∶phosphocholine cytidylyltransferase ( CCT ) [13] , synapsin I [14] , and the autophagosomal protein Barkor/Atg14 ( L ) [15] . A most thorough study of curvature sensing has been performed for a particular kind of amphipathic helices contained in proteins such as Arf1 GTPase-activating protein ( ArfGAP1 ) , responsible for the disassembly of the COPI coat [16]–[19]; the golgin GMAP-210; the sterol sensor/transporter Osh4p/Kes1p; and the nucleoporin Nup133 [18] . These helices , which are characterized by bulky amino acids in the non-polar face and small uncharged amino acids in the polar face ( mainly serine and threonine ) , have been demonstrated not only to bind , selectively , to highly curved membranes of small liposomes [16] , but also to sense mismatch between the actual membrane curvature and the curvature preferred by the specific lipids composing the outer membrane monolayer [17] . This led to the suggestion that these helices sense membrane curvature by recognizing the curvature dependent defects in lipid packing ( see [7] , [20] , [21] and references therein ) and to calling them the amphipathic lipid-packing sensors ( ALPS ) . Since generation and alterations of membrane curvature as well as formation of the lipid packing defects are intimately related to the intra-membrane stresses , it is reasonable to expect that the observed apparent curvature sensing by the insertion-containing proteins is a manifestation of a more general phenomenon of intra-membrane stress sensing . The goal of the present work is to analyze quantitatively the interplay between the protein insertions imitating amphipathic helices and the membrane stresses produced either by the intra-membrane elastic frustrations or by the external forces leading to different kinds of overall membrane deformations including curvature generation . The major statement of the work is that binding of the insertion-containing proteins to the membrane depends primarily on the local intra-membrane stresses existing within the region of the protein embedding , rather than on the way these stresses have been generated . Particularly , concerning the suggested curvature sensing , we predict that the insertion-containing proteins can exhibit similar binding to membranes of different curvature provided that the membrane stresses in the protein-embedding region are similar . Conversely , these proteins are predicted to bind differently to membranes of similar curvature provided that this curvature is achieved by diverse combinations of the intra- and extra-membrane forces and , hence , corresponds to different intra-membrane stresses . Hence the shallow insertions have to be seen as sensors of the intra-membrane stress rather than the membrane curvature . We substantiate our conclusions by demonstrating that the computational approach we use provides quantitative description of the experimental results on differential binding of ALPS domains to liposomes of various diameters and diverse lipid compositions .
We consider amphipathic helix-like protein domains shallowly embedded into the membrane matrix and refer to these domains as the protein insertions . We model such insertions as cylindrical rods of about one nanometer cross-sectional diameter embedded into the outer membrane monolayer such that the rod axis lies parallel to the membrane plane . The typical embedding depth is about 40% of the monolayer thickness [22] ( Fig . 1 , left cartoon ) . We define as the membrane stress sensing by protein insertions the dependence of the insertion binding to the membrane on the membrane stress . To formalize this definition , we consider a system consisting of Np protein insertions partitioning between the aqueous solution and the outer monolayers of lipid membranes , which are subject to elastic stresses and can have curved shape . We quantify the insertion binding to the membranes by the binding constant , KB , defined as a ratio between the number of the insertions remaining in the aqueous solution , , and the number of the membrane bound insertions , , ( 1 ) Since the binding constant is measurable experimentally [18] , the dependence of KB on the membrane stress is a convenient quantitative measure of the stress sensing . The physical reason for the stress sensing is the dependence of the total free energy of the insertion binding , εbind , on the membrane stress . The binding energy εbind , which is determined as the change of the free energy of the whole system resulting from one insertion binding , has a major contribution , ε0 , from a number of essential membrane-insertion interactions such as the hydrophobic , hydrogen bonding and electrostatic interactions . On top of that , embedding of the insertion generates intra-membrane strains and the related change of the membrane elastic energy , εel , such that , ( 2 ) Below , εel will be referred to as the elastic binding energy . It follows from thermodynamics of the insertion binding ( see Text S1 ) that for the insertion number , Np , much smaller than the numbers of water , Nw , and lipid , Nl , molecules , , , the binding constant can be presented as ( 3 ) where , is the stress-independent part of the binding constant , R is the radius of the membrane mid-plane and δ is the monolayer thickness . The correction is relevant only if the radius is so small that the difference between the amounts of the lipid molecules in the outer and inner membrane monolayers becomes considerable . Eq . 3 determines a strong exponential dependence of the binding constant on the elastic binding energy , εel , which , in turn , depends on the values of the intra-membrane stresses and their distribution over the membrane thickness . To reveal the factors that determine the elastic binding energy , εel , we dissect the embedding event into two steps . The first step is embedding of the insertion into the membrane while keeping the initial membrane shape unchanged . The variation of the membrane elastic energy at this stage , εV , is related to creation of a void in the membrane matrix necessary for the insertion accommodation ( Fig . 1 ) . This is accompanied by perturbation of the strains and stresses within the membrane . The second step is a partial relaxation of the stress perturbation due to the change of the membrane shape , which is accompanied by another change of the elastic energy , εR ( Fig . 1 ) . The energy of the first step , εV , can be seen as thermodynamic work performed against the membrane stresses in the course of the void generation , which can be presented as a sum of two contributions , ( 4 ) In essence , is the work of the void formation performed against the initial stresses existing in the membrane before insertion , while accounts for the energy of the stress perturbation . Summarizing , the elastic binding energy can be presented as consisting of three contributions ( 5 ) As shown below , in all relevant cases represents the major part of the energy of void formation . In addition , turns out to be the most convenient value accounting for the distribution of the initial unperturbed stresses in the context of the void formation . Therefore , we will refer to as the void energy and use it as a variable characterizing the stressed state of the membrane before insertion . Since the protein insertions we are considering do not span the whole membrane but rather get shallowly embedded into the membrane matrix , the total energy of the void formation , εV , and hence the elastic binding energy εel , depend on the character of the stress distribution through the membrane thickness . This distribution can be described by the trans-membrane stress profile σ ( z ) [2] ( Fig . S1 ) . In Text S1 we discuss the model assumptions concerning the properties of the trans-membrane stress profile and the relationships between σ ( z ) and the overall force factors determining the membrane stressed state , namely , the lateral tension γ and the bending moment τ . These determine the ways of the stress profile generation by application to the membrane of external forces or by changing the monolayer spontaneous curvatures through variations of their lipid compositions ( see for review [23] , [24] and references therein ) . Specifically , the void energy representing , as mentioned above , the thermodynamic work against the initial stresses needed for the void formation , can be related to the initial stress profile existing within the outer membrane monolayer before the insertion embedding , ( 6 ) where the integration is performed over the volume of the void . Whereas , according to Eq . 6 , the void energy will be calculated by a direct integration of the initial stress profile , the additions to the energy related to the emerging strains , , and the energy of relaxation εR , require a more involved numerical computation including determination of the relaxed membrane shape and will be performed based on the relationships Eqs . S10–S13 using Comsol Multiphysics [5] . We address the sensing by the protein insertions of the intra-membrane stress generated by several specific ways that are experimentally feasible and biologically relevant . First , we consider the stress resulting from the spontaneous curvature Js , which is produced in the membrane monolayers by changing the monolayer lipid composition ( see Text S1 ) and consider three different situations: Second , we consider application of an external torque to an initially flat bilayer , which results in generation of a bilayer curvature J ( see Text S1 ) . This corresponds to the experimental procedures of generation of small liposomes in vitro by means of sonication or extrusion . The external torque produces in the outer monolayer a trans-monolayer stress profile with a bending moment ( 10 ) while the stress-profile in the inner monolayer corresponds to a bending moment ( Fig . 2D ) . ( 11 ) Finally , we analyze the case where the membrane stress is produced by applying a stretching force to the flat monolayer , which generates an overall lateral tension , γ , related to the trans-membrane stress profile ( see Text S1 ) ( Fig . 2E ) . Such a force can be produced as a result of , e . g . , osmotic stretching of the liposomal membrane . We model the membrane as consisting of two monolayers each characterized by a bending modulus of [25] . The monolayers can be laterally uncoupled , meaning that there is a reservoir of material for each monolayer with which the lipid molecules can be exchanged . This is the case in most of the biologically relevant situations where the insertions are restricted to a small membrane patch for which the surrounding membrane plays a role of a lipid reservoir . Alternatively , the monolayers can be laterally coupled if , e . g . , some rigid barrier restricts the lipid exchange between the membrane patch accommodating the insertions and the surrounding membrane , or if the proteins are recruited to the entire area of a closed membrane , so that there are no lipid reservoirs to exchange with , as it occurs in common in vitro assays . An insertion is modeled as a rigid cylindrical rod with a radius of 0 . 5 nm that partially embeds into the outer membrane to a depth of 0 . 8 nm , which imitates the typical size and insertion depth of amphipathic helices [22] . In general , we consider the length of the insertion along the membrane plane to be 2 nm , characteristic for some amphipathic helices [22] , [26] . Our goal is to describe quantitatively the membrane stress sensing by insertions in all five above-mentioned cases of membrane stress generation . We will compute the dependence of the insertion binding constant , KB , Eq . 3 , on the void energy , , Eq . 6 . The absolute value of the binding constant depends , according to Eq . 3 , on the stress-independent factor B , whose value is unknown since it accounts for a combination of the electrostatic , hydrophobic and hydrogen-bonding interactions between the membrane and the insertion , , ( Eq . 2 ) . We will therefore compute the relative binding constant , where characterizes the insertion embedding into the initial unstressed flat membrane . The relation between and the elastic binding energy is ( 12 ) where is the elastic binding energy prior to the stress generation and R is the curvature radius of the membrane in the stressed state under assumption that the membrane shape is spherical . For the cases where generation of the membrane stress is accompanied by membrane curvature variations , we will illustrate the relationship between the stress sensing and the earlier suggested curvature sensing by presenting the binding constant , , as a function of the curvature . The equilibrium distributions of the membrane stresses and strains before and after insertion embedding have been found by solving the set of partial differential equations for the displacement field , and , as explained in [5] . Briefly , for the case of two-dimensional deformations where the membrane adopts a tubular shape with the y-axis laying along the tube , the equations to be solved are ( 13 ) where , , and are the trans-monolayer profiles of the monolayer elastic moduli [5] , [27] . This approach accounts in a continuous manner for variations of the local pressures and elasticities at distances of sub-nanometer scale . This is equivalent to usage of intra-membrane force field for modeling membrane processes by molecular dynamic simulations , which proved to provide a quantitative description of the membrane behavior . The implemented trans-membrane pressure and elasticity profiles represent a simplified version of those computed recently by the state-of-the-art molecular dynamic simulation using Martini force field [27] . Therefore , our predictions are expected to be of at least semi-quantitative accuracy . For a further discussion about the advantages and disadvantages of both continuum and simulation approaches see Ref . [28] . The equations ( Eq . 13 ) were solved for a membrane element of length L and thickness 2h , where h is the monolayer thickness , with the following boundary conditions . First , the insertion is assumed to be much more rigid than the lipid bilayer and hence imposes a horizontal displacement that corresponds to the insertion shape . Second , the top and bottom surfaces of the bilayer are set free , implying that the stresses vanish there . Finally , the right boundary for each monolayer is a symmetry plane , which remains straight but can rotate with respect to the left boundary and can also have a certain constant displacement in both the horizontal and vertical directions . The rotation angle and the displacements are found from minimization of the elastic free energy change upon insertion . The set of equations ( Eq . 13 ) was solved by a finite element method scheme using the commercial software Comsol Multiphysics , allowing one to represent the membrane deformation , calculate the elastic free energy change upon insertion , as well as the void energy . The membrane shape was discretized for the finite element method using a triangular mesh starting with at least 1908 elements , and refined using and adaptive mesh refinement to at least 5514 elements . For simulation of an initial membrane stress created by a combination of different monolayer spontaneous curvatures , the lateral stress profile has been taken according to Eq . S17 . For simulation of the application of an external torque , a constant torque has been applied to the right boundary of the bilayer for both laterally coupled and uncoupled monolayers . Finally , a constant force perpendicular to the right boundary has been applied to simulate the case of a stretched or compressed membrane . In all these cases , the free energy minimization has been acquired by taking into account the work of deformation produced by the externally applied forces .
The computed dependences of the elastic binding energy , , and the relative binding constant , , on the void energy , , for all five aforementioned scenarios of the stress generation are presented in Fig . 3 . Remarkably , the results for obtained for different ways of the stress generation collapse to a single straight line with a slope equal to one ( Fig . 3A ) . This infers , based on Eq . 5 , that the contributions from the stresses emerging in the course of insertion , , and the shape relaxation , , have a negligibly small dependence on . While the void energy is determined solely by the intra-membrane stresses preceding the insertion , and are expected to depend on the scenario of the stress generation . Although is expected to be a small correction to the binding energy , the shape relaxation part of the binding energy , , is of the same order of magnitude as the void energy . Our results show that the shape relaxation is independent of the stress distribution along the membrane in the initial state . As a result , the elastic binding energy is practically independent of the way the stress is produced . Also the dependences of the relative binding constant , , on the void energy , , computed for the five scenarios of the stress generation collapse to a unique curve described by the exponential function ( Fig . 3B ) . According to Eqs . 12 and 5 , this is the result of the above-obtained negligibility of the dependences of the energies and on the void energy , , and of the smallness in most cases of the ratio between the membrane thickness and the curvature radius , . Hence , also the relative binding constant , , quantifying the stress sensing does not depend on the scenario of stress generation . Amphipathic helices of different proteins have various dimensions and could , potentially , get embedded to different depths into the lipid monolayer matrix . Insertion induced curvature depends substantially on the insertion size and the embedding depth [5] , which indicates that also the elastic binding energy , , and , hence , the relative binding constant , , may depend on these parameters . Fig . 3C presents a comparison of the computed dependences of and on the void energy , , for insertions with cross-sectional radii of 0 . 75 nm and 0 . 5 nm for the five ways of stress generation . Both types of insertions are assumed to be embedded to the same depth of 0 . 8 nm , and have the same length of 2 nm . Fig . 3D shows the results obtained for different embedding depths . In both cases , whereas the values of the elastic binding energy do depend on the insertion radius ( Fig . 3C , left panel ) and embedding depth ( Fig . 3D , left panel ) , the variation of as a function of is always represented by a straight line with slope equal one . This means that although the stress-independent part of the elastic binding energy varies with the insertion size and the embedding depth , the stress-dependent part does not and is , practically , equal to . As a result , the dependence of the relative binding constant , , on the void energy , , and , hence , the stress sensitivity are independent of the cross-sectional radius of the insertion ( Fig . 3C , right panel ) and of the embedding depth ( Fig . 3D , right panel ) . Summarizing , the protein insertions are predicted to be universal sensors of membrane stresses existing in the region of the insertion embedding . In three out of five considered scenarios of the stress generation , building up of the stress is accompanied by emergence of membrane curvature . Fig . 4 presents examples of the computed shapes , which are adopted by an initially flat bilayer as a result of the stress generation ( left panels ) followed by the insertion embedding ( right panels ) . If the stress is produced as a consequence of inducing the spontaneous curvature in the outer ( Fig . 4A ) or inner ( Fig . 4B ) monolayer , or by application of an external torque ( Fig . 4C ) , the membrane acquires curvature prior to the insertion embedding such that the insertion interacts with a bent membrane . In case the stresses result from the spontaneous curvature induced symmetrically in both monolayers ( Fig . 4D ) , or from an overall membrane stretching ( Fig . 4E ) , the insertions get embedded into a flat membrane , whereas the curvature builds up only at the latest stage as a result of the shape relaxation . Since , as emphasized in the Introduction , an extended literature has been devoted to the curvature sensing by proteins , we show here the correlation between the strength of the insertion binding and the membrane curvature existing prior to the insertion embedding for the three relevant scenarios of the stress generation ( Fig . 4A–C ) . Fig . 5A , B presents the dependence of the elastic binding energy , , and the relative binding constant , , on the pre-insertion membrane curvature , J . We performed our analysis for a large range of membrane curvatures in order to take into account the highly curved membranes generated , e . g . , during transport carrier formation and endocytosis . If the stress is generated by inducing the spontaneous curvature of the inner monolayer or by application of an external torque , the elastic binding energy , , decreases and the binding constant , , increases with the curvature . Thus , the insertion binding is predicted to be stronger for small rather than large liposomes , in agreement with the experimental results of ALPS binding [17] . Opposite prediction corresponds to the case where the stresses are produced by the spontaneous curvature generation in the outer monolayer . In this situation , , increases and the binding constant , , decreases with growing curvature meaning that the insertions are expected to bind stronger to large rather than to small liposomes . These results can be qualitatively understood by considering the stresses in the external part of the outer monolayer for the different scenarios of curvature generation and their influence on the elastic binding energy . Hence , the dependence of the insertion binding on the membrane curvature and , therefore , the apparent curvature sensing , depend on the way the curvature is produced and can be opposite for different scenarios of curvature generation . It is convenient to quantify the apparent curvature sensitivity by the slope , , of the line representing , approximately , the elastic binding energy as a function of the membrane curvature J , so that . Following Eq . 12 , and under the assumption of smallness of the membrane curvature with respect to the inverse monolayer thickness , , the relative binding constant can be then expressed as . Positive values of the curvature sensitivity , , correspond to preferable insertion binding to membranes with larger curvature ( small liposomes ) , while negative curvature sensitivity , , means preferable binding to membranes with smaller curvature ( large liposomes ) . Based on the results above , the sign of the curvature sensitivity is determined by the way the membrane curvature is produced . The absolute value of the curvature sensitivity depends on the insertion cross-sectional radius and the embedding depths . These dependences are presented in Fig . 5E , F for the three ways of stress generation leading to membrane curvature . The model predicts that the absolute value of the curvature sensitivity increases with the insertion cross-sectional radius ( Fig . 5E ) . Interestingly , is predicted to change non-monotonously as a function the embedding depth ( Fig . 5F ) . It reaches a maximum for the positive , , and minimum for the negative , , values of the curvature sensitivity at some intermediate embedding depths , the latter varying between the different ways of curvature generation ( see Fig . 5F ) . Summarizing , the protein insertions cannot be considered as universal curvature sensors since the character of the curvature sensing depends on the specific curvature generating factors . To validate our proposal of the intra-membrane stress sensing by protein insertions , we used the suggested computational model to treat the quantitative experimental results on membrane binding by the two ALPS motifs of ArfGAP1 , ALPS1 and ALPS2 , which fold within membranes into amphipathic helices . These studies address the dependence of the ALPS binding on the liposome radius [19] and lipid composition [29] . The quantity measured in [19] was the percentage of ALPS1 and ALPS2 amphipathic helices bound to liposomes of 34 nm , 42 nm , and 90 nm radii . Based on these data , we first found the values that can be obtained from the experimental data and accessible to determination by our model . The absolute values of the binding constant , , are unaccessible , since they depend on the part of the binding energy , , accounted by the parameter B ( see Eq . 3 ) that is not stress-dependent but rather determined by a combination of strong interactions such and hydrophobic , electrostatic and hydrogen-bonding interactions . However , we can eliminate the unknown stress-independent parameter B by using the relative binding constants and , presented in Table 1 , and compare them with the experimental results . To obtain the corresponding values computationally , we took into account several aspects of the experimental system [19] . First , the liposomes were formed by extrusion or sonication meaning that the bending moment and the corresponding curvature , J , were generated by an external torque applied to the membranes . Second , the protein motifs were inserted along the whole membrane area rather than locally [19] . Therefore , the liposome monolayers must be seen as laterally coupled since they could not exchange lipid molecules with any lipid reservoir . The modifications of the computational method needed to account for the monolayer coupling were introduced in [5] . Finally , the length and , especially , the embedding depth of ALPS1 and ALPS2 amphipathic helices could be estimated based on structural data but have not been precisely determined and , therefore , had to be considered as fitting parameters . Fig . 6 presents the dependences of the computed binding constant ratio on the liposome radius R for different insertion lengths ( Fig . 6A ) and embedding depths ( Fig . 6B ) . The cross-sectional radii of the amphipathic helices were taken to be 0 . 5 nm for all computations . Fitting the computed values ( Fig . 6A , B ) to those derived from the experiments ( Table 1 ) , we find for each ALPS motif the relationship between the length and embedding depth guaranteeing a quantitative agreement between the experimental and theoretical results ( Fig . 6C ) . The expected lengths of the amphipathic helices , estimated based on the structural data , vary between 4–6 nm for ALPS1 and 3–4 nm for ALPS2 [19] as presented by the shaded region in Figure 6C . Comparison of the computed and the expected values ( Fig . 6C ) predicts that ALPS1 and ALPS2 amphipathic helices embed to a depth close to 0 . 4 nm with a tendency of ALPS1 to penetrate the membrane a little deeper than ALPS2 . In [29] the ArfGAP1 ALPS binding was studied in dependence on the membrane lipid composition , which was modified by symmetric addition to the two membrane monolayers of diacylglycerol ( DAG ) and phosphatidylethanolamine ( PE ) , the lipids generating a negative monolayer spontaneous curvature [23] . The percentage of the membrane bound ArfGAP1 was measured as a function of the mole fraction of these lipids within the membrane . As explained above ( see also Text S1 ) and according to Eq . 9 , symmetric generation of spontaneous curvature of the membrane monolayers leaves the membrane flat but produces stresses in each monolayer . These stresses are expected to modulate the amphipathic helix binding . To enable the comparison of the experimental results [29] with the model predictions , we first plot the measured fraction of bound protein as a function of the monolayer spontaneous curvature , , ( Fig . 7A ) . The latter is assumed to be related to the monolayer lipid composition by the relationship , , where and are , respectively , the spontaneous curvature and the intra-monolayer area fraction of the individual lipid components and the summation is performed over all lipid components [30] . The area fractions of the constituent lipids are taken from [29] , while the lipid spontaneous curvatures are taken to be ( see [23] and references therein ) . A convenient quantity to be derived from the experimental data is the ratio between the protein binding constants for certain monolayer spontaneous curvatures , and for the background spontaneous curvature of corresponding to 70% PC and 30% PS . The values of this ratio in dependence on are presented in Table 2 and Fig . 7B . The same ratio of the binding constants was computed based on our model using the insertion length and the embedding depth as fitting parameters and assuming , as mentioned above , that the monolayers are laterally coupled . The relationship between the ArfGAP1 insertion length and its embedding depth that best fits the experimental data in Fig . 7B is presented in Fig . 7E where the shaded region corresponds to the feasible values of these parameters . According to these results , for a realistic total insertion length of 4 nm , the required embedding depth is about 0 . 4 nm , which is consistent with the above estimations ( Fig . 6 ) .
It has been proposed and extensively discussed in the literature that some peripheral membrane proteins are able to sense large membrane curvatures [7] . In the experimental studies devoted to verification of this idea , the curvature sensing was manifested by a preferential binding of such proteins to small liposomes of few tens of nanometer radii [17] , [31] . The reason for the attractiveness of the concept of curvature sensing by proteins is a straightforward and , therefore , feasible mechanism it suggests for interplay between the geometry and protein composition of cell membrane patches . Such interplay including a positive feedback between the membrane bending and the local protein concentration may have far reaching consequences for the mechanisms of such intra-cellular processes as endocytosis [32] and generation of intra-cellular membrane carriers from the endoplasmic reticulum and the Golgi complex [33] , which involve membrane shaping and remodeling by proteins [34] . Two classes of protein domains have been proposed to sense membrane curvature: hydrophilic intrinsically curved domains , such as BAR domains , able to bind the membrane surface and referred to as the membrane scaffolds [10]; and small amphipathic or hydrophobic domains , such as amphipathic α-helices , which get shallowly embedded into the lipid monolayer matrix and are referred to as the hydrophobic insertions [16] . The potential importance of the curvature sensing by proteins raises a question about the mechanism of this phenomenon . The mechanism by which the protein scaffolds sense membrane curvature is straightforward and related merely to the membrane bending energy . The closer the membrane curvature is to that of the scaffolding protein domain , the less membrane bending deformation is required for the protein attachment to the scaffold and , hence , the less bending energy is consumed for the attachment event making it more energetically favorable . Hence , the scaffolding protein domains must sense the membrane curvature per se . The situation with the hydrophobic insertions , which are not characterized by a curved shape and penetrate the membrane interior rather than stick to the membrane surface , appears to be more complicated . The mechanism of curvature sensing by the insertions has to be related to the internal membrane stresses , which can arise from various membrane deformations rather than , solely , from the overall membrane bending . Here we analyzed numerically the changes of the membrane elastic energy related to the insertion embedding with a goal to understand whether the insertions sense , indeed , the membrane curvature per se , or , alternatively , they sense the intra-membrane stresses independently of the way the stresses are generated . A protein domain can be considered as a curvature sensor per se if its binding to the membrane is influenced by the membrane curvature , and the curvature-dependence of the binding coefficient is the same for different ways of the membrane curvature generation . Our calculations showed that this is not the case for the protein domains , such as amphipathic α-helices , which get shallowly inserted into the membrane matrix . As illustrated in Fig . 5A the binding constant of such domains increases with increasing curvature for the cases where the curvature is produced by an externally applied torque ( black asterisks ) , or by addition of lipids with negative spontaneous curvature to the inner monolayer ( blue tilted squares ) , but decreases if the curvature is produced by addition of lipids with a positive spontaneous curvature to the outer monolayer ( red squares ) . Hence the curvature sensing is not a universal property of the protein insertions . At the same time , according to our model , the protein insertions are universal sensors of the intra-membrane stresses within the region of the insertion embedding . The dependence of the insertion binding coefficient on these stresses does not depend on the way the stresses are generated ( Fig . 3A ) . The mechanism of this stress sensing is based on the elastic energy coming from formation of a void in the membrane matrix necessary to accommodate the insertion ( Fig . 1 ) . The thermodynamic work of the void formation is performed against the internal stress existing within the membrane matrix , which is equivalent to an intra-membrane pressure ( taken with opposite sign ) . As a result the mechanism of the stress sensing by hydrophobic insertions can be seen as a “pushing the walls” mechanism . It has to be noted that the stress-sensing mechanism must underlie also the curvature sensing by transmembrane proteins spanning the whole membrane thickness [35] . Distribution of transmembrane proteins between different regions of the same membrane must be determined by the thermodynamic work , which has to be performed against the intra-membrane stresses in order to create a void accommodating the protein . In case this work varies along the membrane , the transmembrane proteins must partition accordingly . A specific example of such situation is lateral partitioning of trans-membrane proteins characterized by asymmetric cone-like effective shapes along membranes with varying curvature [35] or surface concentration of non-bilayer lipids . The difference between trans-membrane proteins and shallow insertions is that the curvature sensitivity by the former cannot be related to the protein binding coefficient since such proteins are very hydrophobic and , therefore , insoluble in aqueous solutions . The suggested mechanism changes considerably the view on the potential role of proteins domains serving as hydrophobic insertions in the protein targeting and the mode of their action in membrane shaping processes . Further in vitro experimentation aimed at quantitative characterization by biochemical methods of binding of different proteins containing hydrophobic insertions to liposomes of different lipid composition , curvature , or membrane tension , would provide stronger evidence of the stress-sensing mechanism proposed here .
|
Selective targeting of soluble proteins to cellular membranes relies on different mechanisms such as receptor-mediated recruitment or direct binding to specific lipids . A new paradigm has been recently proposed , according to which membrane binding of some proteins is driven by the geometrical and physical properties of the membranes , namely the membrane curvature and lipid packing in the external membrane monolayer . Specifically , several proteins referred to as the membrane curvature sensors have been shown to preferentially bind strongly curved membranes . This mode of protein binding is especially relevant for such fundamental cell processes as endocytosis and carrier generation from ER and Golgi Complex , which involve shaping initially flat membranes into strongly curved ones . A subset of the curvature sensors contains amphipathic or hydrophobic domains that shallowly insert into the membrane . Here we explore computationally the detailed physical mechanism underlying the membrane binding by such proteins and demonstrate that their membrane affinity is not determined by the curvature per se but rather by the membrane stress , independently of the way the stress has been generated . Hence , the significance of our work is in elucidating the relationship between the membrane binding of peripheral proteins carrying shallowly inserting domains and the membrane stresses .
|
[
"Abstract",
"Introduction",
"Model",
"Results",
"Discussion"
] |
[
"physics",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences"
] |
2014
|
Sensing Membrane Stresses by Protein Insertions
|
Sleeping sickness ( human African trypanosomiasis [HAT] ) is a neglected tropical disease with limited treatment options that currently require parenteral administration . In previous studies , orally administered pafuramidine was well tolerated in healthy patients ( for up to 21 days ) and stage 1 HAT patients ( for up to 10 days ) , and demonstrated efficacy comparable to pentamidine . This was a Phase 3 , multi-center , randomized , open-label , parallel-group , active control study where 273 male and female patients with first stage Trypanosoma brucei gambiense HAT were treated at six sites: one trypanosomiasis reference center in Angola , one hospital in South Sudan , and four hospitals in the Democratic Republic of the Congo between August 2005 and September 2009 to support the registration of pafuramidine for treatment of first stage HAT in collaboration with the United States Food and Drug Administration . Patients were treated with either 100 mg of pafuramidine orally twice a day for 10 days or 4 mg/kg pentamidine intramuscularly once daily for 7 days to assess the efficacy and safety of pafuramidine versus pentamidine . Pregnant and lactating women as well as adolescents were included . The primary efficacy endpoint was the combined rate of clinical and parasitological cure at 12 months . The primary safety outcome was the frequency and severity of adverse events . The study was registered on the International Clinical Trials Registry Platform at www . clinicaltrials . gov with the number ISRCTN85534673 . The overall cure rate at 12 months was 89% in the pafuramidine group and 95% in the pentamidine group; pafuramidine was non-inferior to pentamidine as the upper bound of the 95% confidence interval did not exceed 15% . The safety profile of pafuramidine was superior to pentamidine; however , 3 patients in the pafuramidine group had glomerulonephritis or nephropathy approximately 8 weeks post-treatment . Two of these events were judged as possibly related to pafuramidine . Despite good tolerability observed in preceding studies , the development program for pafuramidine was discontinued due to delayed post-treatment toxicity .
Sleeping sickness ( human African trypanosomiasis [HAT] ) is a neglected tropical disease with limited treatment options that currently require parenteral administration . Trypanosoma brucei ( T . b . ) gambiense is found in 24 countries in west and central Africa and currently accounts for over 98% of reported cases [1] . Despite the long history of the disease ( first cases reported in 1373/1374 ) , the drugs available to treat it are toxic , difficult to administer , and stage-specific [2] . First stage symptoms entail bouts of fever , headaches , joint pains , and itching , and a person can be infected for months or even years without major signs or symptoms of the disease . When more evident symptoms emerge , the patient is often already in an advanced disease stage where the central nervous system is affected ( second stage ) . The majority of current HAT research is focused on stage 2 of the disease , which requires drugs that can cross the blood-brain barrier . Drugs for stage 2 HAT are either too toxic ( melarsoprol ) or have too complex a regimen ( nifurtimox-eflornithine combination treatment ) for use against the first stage of the disease . Only two drugs are approved for treatment of stage 1 HAT: pentamidine ( only for T . b . gambiense ) and suramin ( only for T . b . rhodesiense ) , which are both administered parenterally . Suramin , synthesized as a dye in 1916 [3] , has been used for the treatment of sleeping sickness since 1922 [4] [5] , but it can cause undesirable effects in the urinary tract and allergic reactions [1] . Pentamidine , introduced in 1937 , was developed as an analog of synthalin , a hypoglycemic agent with anti-trypanosomal activity [6] [7] . Pentamidine is administered by the intramuscular route and has a reported treatment failure rate after a course of five injections of approximately 7% [8] [9] [10] . Though this efficacy profile is encouraging , treatment with pentamidine has limitations . It requires injection , which hampers its use in rural treatment facilities , and though adverse reactions are usually reversible and its most serious long-term consequence , diabetes , is rare , the treatment is accompanied by a high frequency of adverse events , including hypotension , nephrotoxic effects , leukopenia , and hypo- and hyperglycemia [11] [12] . There is no vaccine for T . b . gambiense HAT and there is a great need for new safe and efficacious drugs that would be easy to use in rural health centers and affordable . In 2000 , the promising orally administered pro-drug pafuramidine was chosen for clinical development by the Consortium for Parasitic Drug Development , which was founded in 1999 to foster the development of compounds with antiprotozoal activity [13] . Pafuramidine is the dimethoxime prodrug of furamidine ( which has demonstrated excellent efficacy in vitro against T . b . rhodesiense ) [14] . Pafuramidine was shown to be effective in vivo in the acute model ( first stage disease ) in mice [15] [16] and in monkeys ( green vervet monkey [17] and rhesus monkey [18] ) . Preclinical evaluation in vitro as well as animal testing indicated no major safety concerns . In 2000 , pafuramidine was further evaluated in Phase 1 studies in healthy patients after single and multiple dosing ( up to 21 days ) and was well tolerated [19] . The subsequent Phase 2 studies ( conducted from 2001 to 2007 ) in patients with stage 1 sleeping sickness supported this finding [20] . Efficacy after 5 days of treatment was limited; therefore , to attain efficacy comparable to that of pentamidine , the treatment duration was prolonged to 10 days . The pharmacokinetic properties of pafuramidine ( in particular , the lack of proportional conversion of DB289 to DB75 at therapeutic doses ) precluded using a higher dose to improve efficacy [19] [20] [21] [22] . This single , confirmatory , pivotal Phase 3 study was developed to support the registration of pafuramidine for treatment of stage 1 HAT under a Special Protocol Assessment in collaboration with the United States ( US ) Food and Drug Administration ( FDA ) . The primary objective of the study was to demonstrate the non-inferiority of oral pafuramidine versus intramuscular pentamidine for treatment of first stage HAT caused by T . b . gambiense . Since safety and efficacy of a new drug should , if at all possible , be established in a study population representative of the target population , the secondary objective was to include pregnant and lactating women as well as adolescents in the study . Reproductive studies of pafuramidine in animals have not indicated any embryo or fetal toxicity or other effects on reproductive function of adult male and female rats or rabbits . Therefore , it was considered appropriate and was approved by the US FDA to proceed with studies including pregnant and lactating women .
This was a multi-center , multi-country , open-label ( sponsor-blinded ) , parallel-group , comparator-controlled , randomized Phase 3 study to compare the efficacy , safety , and tolerability of pafuramidine and pentamidine in 273 patients with first stage HAT caused by T . b . gambiense . The study was conducted at six African sites where T . b . gambiense sleeping sickness is endemic: two trypanosomiasis reference centers ( Angola and the Democratic Republic of the Congo [DRC] ) and four hospitals ( 1 in South Sudan and three in the DRC ) from August 2005 ( first patient enrolled ) to September 2009 ( last patient follow-up completed ) . The study was registered on the International Clinical Trials Registry Platform at www . clinicaltrials . gov with the number ISRCTN85534673 . International Protocol #289-C-006 . There was one protocol amendment that provided detailed microscopy instructions for examining blood and CSF for the presence of trypanosomes and determining WBC count in CSF . The amendment also detailed randomization of pregnant and lactating women in a separate strata , and provided additional clarifications and administrative changes . Male and female patients were eligible to participate if they were ≥12 years of age , weighed ≥30 kg , had first stage T . b . gambiense infection documented by the presence of trypanosomes in the blood and/or lymph , and had no evidence of second stage disease ( no trypanosomes detected in the cerebrospinal fluid [CSF] and ≤5 white blood cells [WBCs]/mm3 in CSF ) . Patients were also excluded if tested positive for malaria or helminth infections . Patients were not tested for HIV prior to treatment . Patients were treated at two trypanosomiasis reference centers ( Angola and the DRC ) and four hospitals ( 1 in South Sudan and three in the DRC ) . Written informed consent was obtained from each patient . If the patient was a minor or mentally impaired , a legal guardian also signed the consent form and if a patient was illiterate , an impartial witness assisted in the consent process . Pregnant and lactating female patients as well as adolescents 12 to 15 years could be enrolled . Adolescents underwent additional safety laboratory testing at the 3-month post-treatment visit . Eligible pregnant and lactating female patients could participate with the understanding that additional safety measurements regarding course and outcome of the pregnancy and/or the health of their infant would be performed . Patients were excluded if they had a possible or confirmed second stage T . b . gambiense infection ( ie , presence of parasite in the CSF upon microscopic examination or a WBC count in the CSF of >5 mm3 ) ; any active , clinically relevant medical conditions that in the investigator’s opinion might jeopardize patient safety or interfere with study participation; presented with a score of less than 9 on the Glasgow Coma Scale; were previously treated for HAT; or displayed other conditions that would compromise participation . Screening occurred within 7 weeks prior to dosing with pafuramidine or pentamidine ( within 6 weeks prior to the baseline evaluation ) using the card agglutination test for trypanosomes [11] [12] and microscopic examination ( thin and/or thick smear ) of blood and lymph node aspirate for trypanosomes either at the trypanosomiasis treatment centers or by mobile diagnostic units . All diagnostic tests performed by mobile teams were repeated in the treatment centers . Lumbar puncture was performed at the treatment centers in all trypanosome-positive cases detected by any method and the disease stage was determined by microscopic examination of CSF for trypanosomes and by WBC counts . If the result was negative , a blood sample was examined ( including hematocrit centrifugation [23] and miniature anion exchange centrifugation technique [m-AECT] [24] ) . All patients were tested for malaria , and filaria using thick and thin blood smears and for diarrhea . If necessary , malaria treatment was given before enrolment; filariasis therapy was administered after study treatment when necessary . Patients were admitted as in-patients to the clinical site for the full duration of the treatment/observation period ( 11 days for pafuramidine or 7 days for pentamidine ) . Other baseline documentation included demographics , medical history , signs and symptoms of HAT , and concomitant disease ( s ) and medication ( s ) . Clinical supplies of pafuramidine were provided to the sites in bottles ( 50 tablets of 100 mg ) labeled to indicate study drug , strength , expiration date , protocol number , and other information according to local regulations . Pentamidine was provided locally by the agency ( generally the national HAT control programs ) responsible for each center in the form of pentamidine isethionate for injection in single-dose vials at 200 mg/vial . For efficacy assessments , patients underwent microscopic examination of blood ( thin and/or thick smear ) , hematocrit centrifugation of blood [25] , microscopic examination of lymph node aspirate , and microscopic examination of blood after m-AECT concentration at the end of treatment and at 3 , 6 , 12 , and 18 months post-treatment [26] . Lumbar puncture was performed for microscopic examination of CSF fluid for WBCs and trypanosomes at baseline and 6 , 12 , and 18 months post-treatment , and at any other evaluation where relapse was suspected or trypanosomes were demonstrated in blood or lymph nodes . Additional assessments of clinical efficacy were performed at 24 months post-treatment . During the treatment and post-treatment period , safety evaluations included vital signs , physical examination , adverse event monitoring , laboratory tests , electrocardiogram ( ECG ) monitoring to the extent possible at each site , and documentation of concomitant medications . Signs or symptoms of HAT were queried and graded . Laboratory tests assessed clinical chemistry ( aspartate aminotransferase , alanine aminotransferase , total bilirubin , glucose , and creatinine ) and hemoglobin . Electrocardiograms were performed at baseline , 1 hour prior to dosing , and 1 hour after dosing for all patients . An additional ECG was obtained on Day 7 post-treatment for pafuramidine-treated patients . Clinical response definitions are listed in Table 1 . The primary efficacy endpoint was the combined rate of cure and probable cure at the 12-month follow-up in the per-protocol data set . The overall cure rate was defined as the proportion of treated patients with no clinical signs and symptoms of HAT , no evidence of trypanosomes in any body fluid examined , and no treatment with other trypanocidal agents for any reason; in addition , ≤5 WBCs/mm3 in CSF obtained from a lumbar puncture was required . Secondary efficacy endpoints were cure , clinical cure , probable relapse , relapse , and death rates at the end of treatment and all follow-up visits . Parasitological cure , probable relapse , relapse , and death rates were also assessed at the 12-month test of cure evaluation and at the 24-month evaluation; the clinical cure was considered equivalent to the parasitological cure at the 24-month evaluation . Study efficacy parameters and timing of post-treatment evaluations were based on WHO recommendations for clinical product development for HAT [27] . Although 18 months post-treatment is recommended to assess clinical cure in HAT control programs due to anticipated increased drop-out rates from follow-up after 6 to 12 months , the 12-month evaluation was chosen as the primary endpoint in this study in order to maintain a robust data set for the primary analysis . Safety was assessed through the end of treatment evaluation and included adverse events , laboratory results , vital sign measurements , physical examinations , and use of any concomitant medications . The term “adverse event” included any of the following events that developed or increased in severity during the study: 1 ) any signs or symptoms whether thought to be related or unrelated to HAT; 2 ) any clinically significant laboratory abnormality; or 3 ) any abnormality detected during physical examination . Adverse events were graded by the investigator ( 1 = mild , 2 = moderate , 3 = severe , 4 = intolerable ) . Adverse events were assessed at every study visit and were classified according to the terms found in the Medical Dictionary for Regulatory Activities ( MedDRA ) . A serious adverse event was defined as any event that suggested a significant hazard , contraindication , side effect , or precaution , it included any event that: 1 ) is fatal; 2 ) is life threatening; 3 ) is a persistent or significant disability/incapacity; 4 ) requires or prolongs in-patient hospitalization; 5 ) is a congenital anomaly/birth defect; or 6 ) is an important medical event , based upon appropriate medical judgment , that may jeopardize the patient or may require medical or surgical intervention to prevent one of the other outcomes defining serious . There were no changes to any of the outcomes . A total of 250 patients , 125 patients per treatment group , were originally expected to be treated in order to include 100 patients per treatment group in the per-protocol . This sample size provided more than 90% power to demonstrate non-inferiority of pafuramidine to pentamidine for the primary endpoint , when the study drugs have equivalent probable cure rates of 95% in the per-protocol analysis . Non-inferiority comparison was conducted with an alpha equal to 0 . 048 and non-inferiority margin ( ie , delta ) of 0 . 15 . The sponsor may have terminated this study prematurely , either in its entirety or at a particular site , for reasonable cause or safety concerns . The sponsor remained blinded and the data were provided to the data safety monitoring board ( DSMB ) for evaluation . Based on these data , the DSMB made recommendations to the sponsor regarding continuation of the study . The study could have been stopped if: 1 ) any new untoward safety issues were identified in the pafuramidine treatment group such that pafuramidine was significantly less safe than pentamidine; 2 ) the re-estimated sample size exceeded 500 patients to achieve 90% power for the primary efficacy endpoint; or 3 ) efficacy analysis indicated that pafuramidine was significantly more effective than pentamidine ( p<0 . 002 ) . An interim analysis was to be conducted when half of the enrolled patients reached the 12-month post-treatment endpoint , however , this was not done because the pafuramidine development program was discontinued due to delayed post-treatment toxicity ( details are provided in the Harms section ) . Patients were randomly assigned by the local investigators to receive either pafuramidine or pentamidine in the order in which they were enrolled . Randomization was carried out in blocks of variable size following a randomization schedule prepared by the sponsor; randomization of pregnant and lactating women was stratified . Each study site was provided with series of individual envelopes each containing a card with the treatment assignment for 1 patient and a control number . After a patient signed the informed consent and inclusion/exclusion criteria were confirmed , the investigator opened the next envelope in the randomization list to obtain treatment assignment for that patient and then transferred the control number to the patient’s case report form . The study was open-label , since pafuramidine is administered orally and pentamidine is administered intramuscularly . However , the sponsor was blinded to treatment assignments . The primary efficacy analysis , demonstrating the non-inferiority of pafuramidine to pentamidine for the combined rate of cure and probable cure , was conducted with an alpha equal to 0 . 048 and non-inferiority margin ( ie , delta ) of 0 . 15 . The comparison was made with a 1-sided 97 . 6% confidence interval ( CI ) for the treatment difference in parasitological cure rate . The normal approximation to the binomial distribution with continuity correction was used to construct the CI . The primary data set for efficacy analysis was the per-protocol data set . The efficacy analysis was carried out for the per-protocol data set ( primary analysis ) , the intention-to-treat ( ITT ) data set , and the modified ITT ( mITT ) data set ( supportive analyses ) . The per-protocol data set was defined as patients who received a minimum of 7 days of pafuramidine or five injections of pentamidine and who attended the test of cure assessment at Month 12 or reached an efficacy endpoint ( death , non-response , or relapse ) at an earlier time . Patients without lumbar puncture at Month 12 were included and their outcome was assessed based on clinical signs and symptoms and parasitological findings from any body fluid examined . The mITT data set consisted of all patients who received the minimum amount of randomized study drug ( 7 or 5 days ) and for whom an end-of-treatment assessment and at least one follow-up efficacy assessment were available . Patients who had received at least one dose of study drug were included in the ITT data set and patients who were lost to follow-up or discontinued from the study for any reason were considered treatment failures in the ITT analysis . The last observation carried backward was used to account for missing data at an earlier evaluation ( in case of cure at a later evaluation ) . For the mITT analysis , missing data were addressed according to the last observation carried forward principle . The secondary efficacy variables were summarized at all time points with point estimates and 1-sided 97 . 5% CIs for the difference between treatments . The safety data set consisted of all patients who received at least one dose of study drug and had at least one safety evaluation after dosing . Treatment group differences in the proportion of patients who reported treatment-emergent adverse events for Day 1 through Day 11 were assessed with Fisher’s exact test . The number and percentage of patients who reported treatment-emergent adverse events were summarized for each treatment group at the system organ class , high-level group terms , and preferred term level . Treatment group differences in the proportion of patients who reported each high-level group term were assessed with Fisher’s exact test . Any clinically significant physical examination changes from baseline were captured as an adverse event .
First stage HAT patients rarely present at a hospital or a treatment center . Therefore , intense screening activities were necessary . Between July 2005 and March 2007 , a total of 234 , 919 patients were screened to find 839 individuals affected with HAT ( Fig 1 ) . The exclusion rate was high ( 566 of 839 patients , 67 . 5% ) ; primary reasons were that patients had stage 2 HAT and did not meet inclusion criteria . A total of 273 patients were randomized: 136 patients received pafuramidine and 137 received pentamidine; all 273 completed the study . Most of the patients ( 91% ) were enrolled in the DRC ( 248 of 273 patients ) ; 5 . 5% were enrolled in Angola ( 15 of 273 patients ) , and 3 . 7% ( 10 of 273 patients ) were enrolled in South Sudan . As shown in Fig 1 , follow-up attendance at Month 24 was good; only 2 patients in the pafuramidine group and 8 patients in the pentamidine group were lost to follow-up . As seen in Table 2 , baseline demographic characteristics between the two treatment groups were similar . The median age of patients in both treatment groups was approximately 30 years , and the majority of patients in both groups were female ( 70% and 66% , respectively ) . As shown in Fig 1 , 133 of 136 patients ( 97 . 8% ) in the pafuramidine group and 129 of 137 patients ( 94 . 2% ) in the pentamidine group were included in the efficacy analysis . Three patients in the pafuramidine group and 8 patients in the pentamidine group were excluded because they were lost to follow-up . As shown in Table 3 at the test of cure evaluation ( 12 months post-treatment ) , the combined rate of cure and probable cure was 89% ( 118 of 133 patients ) in the pafuramidine group and 95% ( 123 of 129 patients ) in the pentamidine group in the per-protocol population . Pafuramidine was non-inferior to pentamidine as the upper bound of the 95% CI did not exceed 15% . This finding was supported by the 24-month follow-up data , where cure rates of 84% for the pafuramidine group and 89% for the pentamidine group were maintained ( Table 4 ) . Supportive analysis in the ITT and mITT populations generated similar results . Table 5 summarizes the secondary efficacy variables for each follow-up visit , including the cumulative number of cures , probable cures , probable relapses , relapses , and deaths for each treatment group . There were no deaths in either treatment group during the active study period , and all patients responded to treatment . Relapses in the pafuramidine treatment group appeared to be evenly distributed over the whole follow-up period , whereas a trend for late relapses was observed in the pentamidine treatment group . The numbers of adolescents and pregnant women ( 8 and 10 , respectively ) were too small to make any definitive conclusions about efficacy in these patients . For lactating women , the overall cure rates of pafuramidine and pentamidine at the test of cure evaluation were the same ( 23 of 26 patients [88 . 4%] in each group ) . However , the low number of lactating women also did not allow for definitive conclusions . No patients prematurely discontinued due to an adverse event during the study . As shown in Table 6 , the most commonly reported adverse events were injection site pain , pyrexia , hypoglycemia , and hypotension . These events occurred more frequently in the pentamidine group than the pafuramidine group , with the exception of pyrexia , which occurred more frequently in the pafuramidine group . The incidence of patients with at least one adverse event overall for Day 1 through Day 11 was statistically significantly less in the pafuramidine treatment group ( 82% , 111 of 136 ) than in the pentamidine group ( 99% , 135 of 137 ) ( p<0 . 05 ) . Among high-level group terms , there were statistically significant differences in favor of the pafuramidine treatment group versus the pentamidine group for hepatobiliary investigations ( 7% vs . 77% , respectively ) ; renal and urinary tract investigations and urinalyses ( 2% vs . 9% , respectively ) ; glucose metabolism disorders ( including diabetes mellitus ) ( 6% vs . 18% , respectively ) ; and decreased nonspecific blood pressure disorders and shock ( 44% vs . 62% , respectively ) ( p<0 . 05 for all ) . A statistically significantly greater percentage of pafuramidine patients than pentamidine patients experienced epidermal and dermal conditions ( 5% vs . 1% , respectively ) ( p<0 . 05 ) . The majority of the adverse events were mild or moderate in severity and typical for patients recovering from first stage HAT . The ECG results from this study were included in a separately published study on cardiac alterations in HAT [28] . In brief , the mean PQ and QTc intervals did not increase during treatment of first stage disease in either treatment group . The appearance and disappearance of repolarization changes at the end of treatment were comparable between groups . A total of 43 patients experienced serious adverse events during the study ( including the follow-up period ) : 19 of 136 patients ( 14 . 0% ) in the pafuramidine group and 24 of 137 patients ( 17 . 5% ) in the pentamidine group . Of these , only 3 patients had serious adverse events while on treatment: 1 in the pafuramidine group ( cellulitis considered probably not related to study drug ) and 2 in the pentamidine group ( hypersensitivity considered not related to study drug and subcutaneous abscess considered probably related to the study drug ) . All other serious adverse events occurred during the follow-up period . Of the 43 patients with serious adverse events , it was initially considered probable that only one was related to the study drug ( subcutaneous abscess in the pentamidine group ) . Re-evaluation of the relatedness of these events to the study drug was subsequently performed when serious renal and hepatic post-treatment toxicity was observed in 3 patients in a supportive Phase 1 study of pafuramidine , which was conducted in South Africa ( in December 2007 ) , during the follow-up period of the current Phase 3 study . The Phase 1 study included 175 male and female volunteers taking oral pafuramidine 100 mg BID for 14 days [19] . Retrospectively , the glomerulonephritis reported for the 2 pafuramidine patients in the current Phase 3 study was considered to be possibly related to the study drug . Thirteen patients ( 6 in the pafuramidine group and 7 in the pentamidine group ) died during the follow-up period . All deaths were considered not related or probably not related to the study drug . Two deaths in the pafuramidine group were considered to be related to HAT; one death was considered to be related to relapse of HAT and another was considered to be associated with treatment of a HAT relapse with melarsoprol . Safety data for adolescent as well as pregnant and lactating women were similar to the observations in the general population .
Although this study was conducted in rural conditions in Angola , South Sudan , and the DRC with local teams that had limited experience in clinical studies , this study was fully compliant with Good Clinical Practice and regulatory standards . In addition , this was the first Phase 3 study of a new drug intended for treatment of sleeping sickness conducted under a US FDA Investigational New Drug Application that followed contemporary International Conference on Harmonisation guidance . The results demonstrate the efficacy of pafuramidine in the treatment of first stage HAT , with an overall cure rate that was statistically non-inferior to that observed for pentamidine ( 89% vs . 95% , respectively ) at 12 months post-treatment . The results obtained in the per-protocol set were confirmed in the ITT and mITT analysis populations , which included missing follow-up visits and patients lost to follow-up . The Month 24 results in all populations corroborate the 12-month results and demonstrate the robustness of the primary efficacy analysis . Compared with patients who received pentamidine , pafuramidine-treated patients ( total population including the subpopulations of adolescents and pregnant and lactating women ) had lower rates of treatment-related adverse events ( 93% vs . 40% , respectively ) and lower rates of adverse events related to hepatic , renal , and metabolic toxicity . An ECG analysis revealed no cardiotoxicity for either drug [28] . These data are consistent with the good tolerability observed for pafuramidine in the previous Phase 2 studies [20] . This study was also designed to evaluate efficacy and safety of pafuramidine and pentamidine in subpopulations that are particularly vulnerable to the long-term socioeconomic burdens associated with HAT , mainly pregnant and lactating women . However , the number of participating pregnant and lactating women was too low for a thorough separate analysis . The low number may be a result of social pressures and fear of treatment , which could be a detriment to seeking HAT treatment and going to a hospital . Low fertility and amenorrhea , which are often associated with HAT , may also have contributed [29] . From the limited number of relevant patients , there was no evidence for reduced efficacy or additional safety issues relating to pafuramidine compared with those observed in the overall study population . Numeric cure rates were similar and no specific safety issues were identified . The initial safety profile observed for pafuramidine in this study was consistent with the results of preceding studies in the pafuramidine clinical development program [19] [20] . However , 3 patients in the pafuramidine group exhibited glomerulonephritis or nephropathy approximately 8 weeks post-treatment . On further examination , these events appeared to be similar to events that occurred in the previously mentioned supportive South African Phase 1 safety study . After re-evaluation , 2 of the 3 patients were considered to have events that were possibly related to pafuramidine by the principal investigator . It should be noted that the analysis of safety data , particularly of the serious adverse events that occurred in the HAT study reported here , did not reveal any apparent negative long-term effects of pafuramidine . The patients who experienced renal toxicity recovered without sequelae , and the additional safety data obtained during the follow-up revealed no differences in abnormal biochemistry values between pafuramidine and pentamidine groups . Eventually , clinical development of pafuramidine was discontinued in early 2008 , since the renal toxicity observed in the additional Phase 1 study was considered to be an unacceptable risk . Preliminary evidence of the involvement of the kidney injury molecule ( KIM-1 ) was only very recently provided through the use of a mouse diversity panel [30] . From the perspective of study design , it is noteworthy that the 12-month endpoint for efficacy effectively predicted the clinical outcomes determined at the 24-month evaluation . Thus , 12 months is a meaningful endpoint for a sleeping sickness study with adequately performed follow-up . The infrastructure and technical expertise developed during the Phase 2 development program for pafuramidine were effectively leveraged to guide the screening , enrolment , and oversight of the larger study population included in this Phase 3 registration study , and eventually led to a unique and comprehensive data set . The successful conduct of the study was evidenced by the retention of 97% ( 265 of 273 patients ) of the randomized patients at the 24-month ( end-of-study ) evaluation . Finally , the lessons learned from the Phase 2 development program were also helpful in ensuring that the Phase 3 study complied with Good Clinical Practice and regulatory standards required for a registration study [20] . The repeated success of clinical study conduct throughout the pafuramidine development program provides a model for future studies in rural Africa and will undoubtedly contribute to continued improvement of HAT control in sub-Saharan Africa .
|
Sleeping sickness , or human African trypanosomiasis ( HAT ) , is a neglected tropical disease . Because only 2 treatment options are available to treat persons with stage 1 disease , and both require parenteral administration , oral drugs would be of great benefit to the affected population . In this Phase 3 , multi-center , randomized , open-label , parallel-group study , we compared oral pafuramidine with intramuscular pentamidine in persons in sub-Sahara Africa with first stage HAT . At 12 months , the overall cure rates ( combined clinical and parasitological cure ) were similar: 89% in the pafuramidine group and 95% in the pentamidine group . At 24 months , the cure rates continued to be high: 84% and 89% , respectively . Pafuramidine’s safety profile was superior to the comparator drug , and it was consistent with the overall safety profile seen in previous Phase 2 studies . Upon further analysis , however , a renal safety issue was identified as being possibly related to pafuramidine and further clinical development was halted . Nevertheless , the clinical studies conducted in the pafuramidine development program provide a model for future studies in rural Africa .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"clinical",
"research",
"design",
"maternal",
"health",
"obstetrics",
"and",
"gynecology",
"african",
"trypanosomiasis",
"nervous",
"system",
"tropical",
"diseases",
"parasitic",
"diseases",
"parasitology",
"research",
"design",
"phase",
"iii",
"clinical",
"investigation",
"clinical",
"medicine",
"women's",
"health",
"pharmaceutics",
"pregnancy",
"neglected",
"tropical",
"diseases",
"pharmacology",
"research",
"and",
"analysis",
"methods",
"infectious",
"diseases",
"zoonoses",
"safety",
"studies",
"adverse",
"events",
"protozoan",
"infections",
"trypanosomiasis",
"drug",
"research",
"and",
"development",
"anatomy",
"clinical",
"trials",
"physiology",
"biology",
"and",
"life",
"sciences",
"drug",
"therapy",
"cerebrospinal",
"fluid"
] |
2016
|
Efficacy and Safety of Pafuramidine versus Pentamidine Maleate for Treatment of First Stage Sleeping Sickness in a Randomized, Comparator-Controlled, International Phase 3 Clinical Trial
|
Among mammals , only eutherians and marsupials are viviparous and have genomic imprinting that leads to parent-of-origin-specific differential gene expression . We used comparative analysis to investigate the origin of genomic imprinting in mammals . PEG10 ( paternally expressed 10 ) is a retrotransposon-derived imprinted gene that has an essential role for the formation of the placenta of the mouse . Here , we show that an orthologue of PEG10 exists in another therian mammal , the marsupial tammar wallaby ( Macropus eugenii ) , but not in a prototherian mammal , the egg-laying platypus ( Ornithorhynchus anatinus ) , suggesting its close relationship to the origin of placentation in therian mammals . We have discovered a hitherto missing link of the imprinting mechanism between eutherians and marsupials because tammar PEG10 is the first example of a differentially methylated region ( DMR ) associated with genomic imprinting in marsupials . Surprisingly , the marsupial DMR was strictly limited to the 5′ region of PEG10 , unlike the eutherian DMR , which covers the promoter regions of both PEG10 and the adjacent imprinted gene SGCE . These results not only demonstrate a common origin of the DMR-associated imprinting mechanism in therian mammals but provide the first demonstration that DMR-associated genomic imprinting in eutherians can originate from the repression of exogenous DNA sequences and/or retrotransposons by DNA methylation .
Genomic imprinting , or parent-of-origin-specific gene silencing , has been observed in both eutherian and marsupial , but not monotreme mammals . In eutherians , more than 80 imprinted genes have been found and differential DNA methylation plays a crucial role in regulating their imprinted expression patterns [1–5] . In marsupials , three genes—IGF2 , IGF2R , and PEG1/MEST—are imprinted , as they are in eutherians , but no differentially methylated regions ( DMRs ) have been found in marsupials [6–8] . Consequently , the regulatory mechanisms of genomic imprinting were thought to have evolved differently between marsupials and eutherians [6 , 8 , 9] . It has been hypothesized that genomic imprinting arose as a by-product of a DNA methylation mechanism that silences foreign DNAs [10] , such as retrotransposons [11 , 12] . Similarly , transgenes can also become methylated , depending on parent of origin , further supporting a link between genomic imprinting and silencing of foreign DNAs [13–15] . PEG10 is an imprinted gene sharing homology with the sushi-ichi retrotransposon , and in humans and mice it has a clear DMR in its promoter region . Interestingly , PEG10 is conserved in eutherian mammals but not in nonmammalian vertebrates , such as birds and fish [11 , 16–18] . Therefore , investigating the origin of the retrotransposon-derived PEG10 locus would clarify the relationship between retrotransposon ( or exogenous DNA sequence ) insertion and genomic imprinting . PEG10 is an essential placental gene in eutherians , since knock-out mice have severe placental defects with loss of spongiotrophoblast and labyrinth layers leading to early embryonic lethality [18] . The origin of PEG10 is therefore of interest in view of its possible contribution to the evolution of mammalian placentation . Sequence identified as PEG10 has recently been reported in the South American marsupial the grey short-tailed opossum ( Monodelphis domestica ) [16] but its precise location , genetic structure , and imprint status remains unknown . Here , we examined the PEG10 locus in two Australian mammals by isolating bacterial artificial chromosome ( BAC ) clones from a marsupial , the tammar wallaby , and from a monotreme , the platypus .
We used the tammar wallaby and the platypus as the representative species of marsupials and monotremes , respectively , to compare with representative eutherians , the human and the mouse . Several BAC clones were isolated from tammar and platypus containing SGCE ( sarcoglycan epsilon ) , the neighboring gene of PEG10 in eutherians . DNA sequencing of one tammar BAC clone demonstrated the existence of PEG10 and its conserved location adjacent to SGCE with transcription occurring in a head-to-head manner ( Figure 1A ) . The genetic structure of tammar PEG10 was also the same as that of eutherians , with two open reading frames ( ORFs ) related to the sushi-ichi retrotransposon GAG and POL proteins , respectively [11] . The CX2CX4HX4C RNA-binding motif of GAG and the DSG sequences of the proteinase activation site of POL were also conserved ( Figure S1 ) . Tammar PEG10 was localised close to the telomere of Chromosome 3q , consistent with its autosomal location on proximal mouse Chromosome 6 ( Figure 1B ) . However , in the platypus , there were no PEG10 homologous sequences between SGCE and PPP1R9A ( also called NEURABIN1 ) that flank PEG10 in other mammals ( Figure 1A ) . Using our tammar sequence and the published opossum genome sequence , we compared the entire region between SGCE and PPP1R9A with the equivalent region in several vertebrates from fish ( fugu ) to mammals . The size of this region in the platypus was similar to that of the chicken and was smaller than in other mammals . There were numerous long interspersed nuclear elements ( LINEs ) and short interspersed nuclear elements ( SINEs ) ( grey bars in Figure 1A ) in all mammalian groups . A previous report suggests that there are less long terminal repeat ( LTR ) -type retrotransposon-derived sequences in the opossum and wallaby genomes [19] , but a large number of these sequences were found in this region as well as in mouse and human ( blue bars in Figure 1A ) . Consistent with the previous report , these were absent in the platypus [19] , as was PEG10 . Most of the LTR-type retrotransposon-derived sequences observed in these regions are species specific . This suggests that most of these insertions occurred after species diversification . The presence of PEG10 sharing homology with one of the LTR-type retrotransposons , sushi-ichi [11] , in both marsupials and eutherians suggests that the original PEG10 sequence insertion in the common therian ancestor was an early event in the therian-specific expansion of LTR-type retrotransposons . These results indicate that the original PEG10 was inserted into the genome of the therian ancestor and evolved to its present function as an essential placental gene after divergence from the monotremes . The acquisition of a new function for an existing character during evolution , a process termed “exaptation” by Gould and colleagues , would be enhanced by the provision of novel genetic materials , such as retrotransposons [20 , 21] . Thus , the requirement for PEG10 in placental function is a clear example of “exaptation . ” The fossil record shows that there was extensive radiation of therian mammals after their split from the Prototheria . New LTR-type retrotranposon-derived sequences might therefore have contributed novel genetic resources to this radiation . In mice , there is a large imprinting cluster near Peg10 which includes the paternally expressed Sgce gene [22 , 23] Ppp1r9a , which is maternally biased in extraembryonic tissues [23] and Asb4 , which is completely maternally expressed in both embryos and the extraembryonic tissues [23 , 24] . Tammar PEG10 showed almost complete monoallelic expression in all individuals . Paternal expression was confirmed in two embryos and one yolk sac placenta sample ( Figure 2 ) . Unexpectedly , tammar SGCE showed predominantly biallelic expression with only a small paternal bias , despite a short 200-bp distance between the transcription start sites of PEG10 and SGCE ( Figure 2 ) . PPP1R9A and ASB4 showed biallelic expression without parental bias ( Figure 2 ) . These results clearly demonstrate that imprinting in this region is restricted to the PEG10 gene in the tammar . As described above , the eutherian PEG10 imprinted region includes several neighboring genes , suggesting that the imprinted region expanded in the eutherians while in marsupials imprinting was restricted to PEG10 . A CpG island is present in the promoter regions of SGCE and PEG10 in tammar as well as in mouse . To determine why SGCE did not show imprinted expression we examined its methylation status . Surprisingly , we found a DMR with a clear boundary of DNA methylation between the PEG10 side and the SGCE side of the CpG island in both embryos and yolk sac placentas ( Figure 3 ) . Furthermore , selective DNA methylation of the maternal allele was confirmed using a DNA polymorphism in this region as predicted by the paternal expression of PEG10 . The DNA methylation started about 60 bp downstream from the transcription start site of PEG10 , suggesting that maternal transcription is inhibited by methylation of downstream regulatory elements and not by the typical mechanism of promoter methylation ( Figure 3 ) . Both LTR [25] and non-LTR retrotransposons [26] are known to have internal transcriptional regulatory elements for their transcription . As PEG10 is a retrotransposon-derived gene , these elements may exist within the DMR not , as is usual , upstream of the transcription start site . A part of the CpG island was possibly derived from LTRs in the original PEG10 sequence and the methylation that originated from a host–defense mechanism may be restricted to the ancient repetitive-element homologous region . Alternatively , a boundary function for DNA methylation spreading and/or transcription regulation may be included in the marsupial CpG island . The retrotransposon LTR sequences are CpG-rich and have such a boundary function in Saccharomyces cerevisiae and Drosophila melanogaster [27 , 28] . In Drosophila , the boundary function has been attributed to the binding of the SU ( HW ) protein . A consensus SU ( HW ) binding site was not found around the methylation boundary in the tammar PEG10 . However , the CTCF protein , well known to have a similar insulator/boundary function in mammals , may bind to the possible boundary elements containing CT-rich sequences in this region . Even with the presence of a DMR , it is possible that the maternal copy of PEG10 in the tammar is silenced by another mechanism and is only secondarily methylated . We therefore examined whether the imprinted expression of tammar PEG10 was regulated by DNA methylation . A reduced level of DNA methylation was observed in three sites of the CpG island in cells cultured with 5-aza-2′-deoxycytidine , a DNA methylation inhibitor ( Figure 4A ) . Repetitive experiments performed for the most 3′ site using three independent cell lines established from fetal lung and endometrium also showed statistically significant reductions in DNA methylation levels ( Figure 4B ) , and increased PEG10 expression from normally repressed alleles was observed in each case ( Figure 4C , black and grey bars ) , although the expression levels were still much lower than active alleles ( Figure 4C , white bars ) . These results demonstrate the association between imprinted expression of PEG10 and DNA methylation in a marsupial , although it still remains unknown if the differential methylation originates in the germline as does a typical primary DMR in eutherians . The DNA methylation status of retrotransposons can differ between male and female germ cells . For example , IAP and LINE1 are more highly methylated in sperm than oocytes , while Alu is less methylated [12] . Mice and humans with paternal disomy that express PEG10 and SGCE biallelically have normal phenotypes , so monoallelic expression of these genes is not essential for development . Therefore , although PEG10 is essential for placental development in the mouse , the imprinting of this locus may be a functionally unimportant inheritance derived from the nature of the original retrotransposition of PEG10 . There were CpG islands in the putative promoter region of SGCE of the chicken , platypus , tammar , mouse , and human ( Figure 5A ) . We hypothesize that insertion of PEG10 after the divergence of therian from prototherian mammals expanded the CpG islands ( Figure 5 ) . In the tammar , DNA methylation is restricted to PEG10 , but in the mouse and human , the entire region is differentially methylated . These differences may be explained by the presence or absence of a boundary function of the CpG island in these groups as discussed above . However , in both cases , insertion of PEG10 , which must have occurred in the therian ancestor , is clearly sufficient to establish imprinting of this region ( Figure 5B ) . Our study confirms that silencing of exogenous DNA after retrotransposon insertion can drive the evolution of genomic imprinting in mammals .
Tammar wallabies of Kangaroo Island origin were maintained in our breeding colony in grassy , outdoor enclosures . Lucerne cubes , grass and water were provided ad libitum and supplemented with fresh vegetables . Fetuses and yolk sac placenta tissue were collected between days 22 and 25 of the 26 . 5-d gestation [29] . Experimental procedures conformed to Australian National Health and Medical Research Council ( 1990 ) guidelines and were approved by the Animal Experimentation Ethics Committees of the University of Melbourne . Each BAC clone in the tammar and platypus BAC libraries was stored separately and was spotted onto nylon membranes correspondently . These membranes were hybridized with the partial SGCE probes of each species , and the positive clones were identified according to the locus information of the signals on the membranes . The tammar PEG10 sequence was determined by the primer walking method from a partial fragment amplified using cross-species degenerate primers . The platypus sequence between SGCE and PPP1R9A was determined by the shotgun sequencing method , and it was completed using a published database of whole genome shotgun sequences and direct sequencing of PCR products . RepeatMasker ( http://www . repeatmasker . org ) was used for the detection of LINEs , SINEs , and LTR elements in the genomic region between SGCE and PPP1R9A . BAC DNA was labeled by nick translation with digoxygenin-11-dUTP . Hybridization was operated with the labeled BAC DNA and C0t-1 DNA at 37 °C overnight . Anti-digoxygenin-Cy3 and DAPI were used for the detection of the signals and for the counterstain , respectively . The details have been described in our previously published paper [8] . The 3′ UTRs of PEG10 and ASB4 including the polymorphisms were amplified by 30–35 cycles of RT-PCR using the following pairs of primers: PEG10-F1 , 5′-CAAATGCCATTGCCGTCT-3′ and PEG10-R1 , 5′-GTTAGACGGTCAGCTCCACG-3′; PEG10-F2 , 5′-CAACCAGGGGAGCTAGGATT-3′ and PEG10-R2 , 5′-GAACATCCATGCACCGTAGA-3′; PEG10-F3 , 5′-CTCTCTGGAGCGGTATCCAG-3′ and PEG10-R3 , 5′-TGTGAGATTTGGCAATCATACA-3′; and ASB4-F , 5′-AACACCCCGAGGTCTCTCAT-3′ and ASB4-R , 5′-GAGGACCATGGCATTTATTCA-3′ . The following correspondent SNuPE primers were used for the single nucleotide extension: PEG10-SN1 , 5′-ATTCATTTCCCTTCCCAACAT-3′; PEG10-SN2 , 5′-CCCTGGCTGCGAGACCA-3′; PEG10-SN3 , 5′-CCCGGGGAGCTCCGAGC-3′; and ASB4-SN , 5′-CAAGAAGCAAGTAGTTCTTCAAAGG-3′ The labeling of the primers for the last hot cycle was operated using [γ-32P]ATP and T4 DNA kinase . The final PCR products of SGCE and PPP1R9A were digested by Alul and Mspl , respectively . The 3′ UTRs of SGCE and PPP1R9A including the polymorphisms on the recognition sequences of these restriction enzymes were amplified by 30–35 cycles of PCR using the following pairs of primers . Asterisks indicate the labeled primers: SGCE-F , 5′-CAGTGATGGCGTTCTGTACG-3′; *SGCE-R , 5′-GTTGATGACCAGGTTGTGCC-3′; *PPP1R9A-F , 5′-CCAGGAGAAGATGGAGAAGC-3′; and PPP1R9A-R , 5′-GTTGGGGATGAAGGAGTGTG-3′ . Ten percent polyacrylamide gels were used for the gel electrophoresis of the digested samples . After the bisulphite treatment [30] for the genomic DNA of tammar , the region corresponding to the CpG islands over the promoter regions of SGCE and PEG10 was amplified by 35 cycles of PCR using the following pair of primers: CGI-F , 5′-GGAGTGATTGTGGAAATGGAGGTG-3′ and CGI-R , 5′-ATACAAAATCCCCCCCCTAAACCTC-3′ . The PCR products were cloned and the clones were analyzed by sequencing . Primary culture of tammar fetal lung cells from day 25 of gestation and adult endometrium cells were used in this study . Control cells were cultured in 50% AmnioMAX ( Invitrogen , http://www . invitrogen . com ) and 50% DMEM supplemented with 10% fetal calf serum and penicillin/streptomycin at 37 °C/5% CO2 . Cells for 5-aza-2′-deoxycytidine treatment were cultured in the same media but containing 10 μM of 5-aza-2′-deoxycytidine ( Sigma ) . Fresh media with 5-aza-2′-deoxycytidine were added every 24 h for 6 d . Three regions in the PEG10 DMR were amplified by 35 ( for the middle and right side in Figure 4A ) and 40 ( for the left side in Figure 4A ) cycles of PCR from the bisulphite treated genomic DNA of cells using the following pair of primers: LEFT-F , 5′-GTATTAGTTTTTTTGTAGTT-3′ and LEFT-R , 5′-CCTAAAAAACTACCCTACTCC-3′; MID-F , 5′- GAGATGGGGAGATTGATATTT-3′ and MID-R , 5′- CCCTATAACTAAACTACAATCTCTCC-3′; and RIGHT-F , 5′- CCTCCCATTAACTTTAAAATCACC-3′ and RIGHT-R , 5′- ATTGTAGTAATGGGGTAGGTTATG-3′ . PCR products were digested by RsaI ( for the left side ) or Aci I ( for the middle and right side ) for analyses . The 3′ UTR of PEG10 , including the polymorphism on the TaqI or BceAI recognition sequences , was amplified by 30 cycles of RT-PCR using the PEG10-F1R1 or PEG10-F3R3 primer pairs . PCR products were digested by TaqI or BceAI for analyses .
The National Center for Biotechnology Information GenBank ( http://www . ncbi . nlm . nih . gov/Genbank ) sequence accession numbers for tammar and platypus BACs are AB260975 and AB260976 , respectively .
|
Genomic imprinting is a gene regulatory mechanism controlling parent-of-origin-dependent expression of genes . In eutherians , imprinting is essential for fetal and placental development and defects in this mechanism are the cause of several genetic disorders . In eutherian mammals , genomic imprinting is controlled by differential methylation of the DNA . However , no such methylation-dependent mechanism had been previously identified in association with marsupial imprinting . By comparing the genome of all three extant classes of mammals ( eutherians , marsupials , and monotremes ) , we have investigated the evolution of PEG10 ( paternally expressed 10 ) , a retrotransposon-derived imprinted gene that is essential for the formation of the placenta in the mouse . PEG10 was present in a marsupial species , the tammar wallaby , but absent from an egg-laying monotreme species , the platypus . Therefore , PEG10 was inserted into the genome at the time when the placenta and viviparity were evolving in therian mammals . This study has shown that PEG10 is not only imprinted in a marsupial , but that its imprint is regulated by differential methylation , suggesting a common origin for methylation in the therian ancestor . These results provide direct evidence that retrotransposon insertion can drive the evolution of genomic imprinting in mammals .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] |
[
"evolutionary",
"biology",
"molecular",
"biology",
"developmental",
"biology",
"mammals"
] |
2007
|
Retrotransposon Silencing by DNA Methylation Can Drive Mammalian Genomic Imprinting
|
We often perform movements and actions on the basis of internal motivations and without any explicit instructions or cues . One common example of such behaviors is our ability to initiate movements solely on the basis of an internally generated sense of the passage of time . In order to isolate the neuronal signals responsible for such timed behaviors , we devised a task that requires nonhuman primates to move their eyes consistently at regular time intervals in the absence of any external stimulus events and without an immediate expectation of reward . Despite the lack of sensory information , we found that animals were remarkably precise and consistent in timed behaviors , with standard deviations on the order of 100 ms . To examine the potential neural basis of this precision , we recorded from single neurons in the lateral intraparietal area ( LIP ) , which has been implicated in the planning and execution of eye movements . In contrast to previous studies that observed a build-up of activity associated with the passage of time , we found that LIP activity decreased at a constant rate between timed movements . Moreover , the magnitude of activity was predictive of the timing of the impending movement . Interestingly , this relationship depended on eye movement direction: activity was negatively correlated with timing when the upcoming saccade was toward the neuron's response field and positively correlated when the upcoming saccade was directed away from the response field . This suggests that LIP activity encodes timed movements in a push-pull manner by signaling for both saccade initiation towards one target and prolonged fixation for the other target . Thus timed movements in this task appear to reflect the competition between local populations of task relevant neurons rather than a global timing signal .
In order to plan for upcoming movements and actions , the brain must be able to represent the passage of time . However , the nature of signals that encode time ( measurement ) and the way in which these signals are utilized in order to produce movement ( production ) are unclear [1]–[3] . In particular , signals associated with temporal measurement , the representation of the passage of time between external events , need not reflect temporal production , the execution of a behavior at a specific time [3]–[8] . Although the passage of time must be monitored during both temporal measurement and temporal production , in temporal measurement consistencies in the sequence and timing of external events can serve as a clock . For example , we can decide that it is time to go home after a workday by looking at the clock ( temporal measurement ) or by an internal judgment ( temporal production ) , irrespective of external cues , that it is getting late . The distinction is particularly important when looking for the physiological basis of temporal representations because consistent variations in activity over time in a stereotyped task may reflect task parameters , such as stimulus events or probabilities , that systematically vary over time , rather than representing time itself [9]–[21] . For example , if the animal is explicitly cued when to make a movement and that cue tends to happen at certain times , then activity may represent the time-dependent probability of cueing rather than temporal production per se [16] . Similarly , if movements are linked to sensory events , such as the approach of a moving target [22] , temporal variations in activity may reflect some combination of stimulus and task dynamics rather than the movement itself . Activity could also reflect external events such as reward that , by virtue of being tightly coupled to the upcoming movement , can be readily anticipated . In most timing studies , rewards are contingent on a single specific eye movement [4] , [9] , [13] , [16] , [18] , [20] that is cued at a particular time . Therefore , both the sensory cue instructing the movement and the reward that is linked to the production of that movement can be readily anticipated . This is of particular interest because neurons within the parietal cortex have been shown to modulate their activity during visual anticipation [23] and reward anticipation [24]–[29] as well as during movement planning [9] , . In contrast to temporal measurement , temporal production signals corresponding with the passage of time can be generated completely internally and need not have explicit environmental correlates . To investigate such a completely internal timing mechanism , we designed a task that requires animals to move consistently at regular time intervals without any external or environmental cues ( Figure 1A , B ) . Specifically , the task requires the animals to make rapid eye movements ( saccades ) back and forth between two fixed targets every second . Trials were immediately aborted if any intersaccadic interval , the time between subsequent saccades , differed by more than 200 ms from the 1 s standard . The lack of any external timing-related visual cues serves to control for sensory anticipation and temporal measurement . Trial length and reward amount were randomized on a trial-by-trial basis to minimize reward anticipation . We further dissociated reward from the saccadic movement by allowing trials to end at any time within an interval , not just immediately following the completion of a saccade . Finally , by utilizing saccades instead of other movements ( such as reaching ) , we minimized any variability in motor output since saccade metrics between two fixed locations are highly consistent . After animals were trained to consistently saccade at 1 s intervals , we recorded from individual neurons in the lateral intraparietal area ( LIP ) . Our recordings confirm previous suggestions of temporal representations within the area but suggest that the nature of these representations is far different from prior reports . First , we found that unlike the activity observed in previous studies , activity within LIP was characterized by a constant decrease in activity throughout the timed interval prior to movement initiation . Second , activity throughout intersaccadic intervals was significantly predictive of interval duration on an interval-by-interval basis . Lastly , the sign of the correlation between activity and behavior depended on saccade direction . Therefore , it appears that LIP activity contributes to both saccade initiation and fixation , and temporal production in our task reflects the balance between these two signals .
We trained two monkeys ( Macaca mulatta ) to perform a variant of the delayed-saccade task [36] , [37] called the self-timed rhythmic-saccade task ( Figure 1A , B ) . The task was designed to focus on temporal motor production and avoid any regular pattern of sensory stimulation or reward that might lead to temporal measurements within the task . In this paradigm , the monkey was required to rhythmically saccade back and forth between two static targets ( red and blue ) at a defined interval so that saccades occurred each second . Because there were no external cues regarding this interval , the monkey had to form and follow an internal and explicit temporal representation to successfully perform the task [23] . Saccades made toward the neuron's response field ( RF ) are referred to as “peripheral” saccades ( blue to red ) , while saccades made away from the RF ( or towards the central target ) are referred to as “central” saccades ( red to blue ) . To verify that the animals learned the trained interval of 1 s , we examined the produced intersaccadic intervals as a function of saccade direction and serial order during the rhythmic task ( Figure 2 ) . A total of 78 , 059 saccades ( average of 781 saccades/cell , with a standard deviation of 235 saccades/cell ) were analyzed over 19 , 177 trials . We found that both animals displayed highly consistent behavior , with standard deviations much smaller than the allowable behavior window of ±200 ms . Interval production depended neither on saccade direction nor serial order . Animal 1 produced an average intersaccadic intervals of 1 , 003 ms ( standard deviation of 111 ms ) , 1 , 022 ( 118 ) , and 985 ( 101 ) prior to all , peripheral , and central saccadic movements , respectively . Animal 2 produced those same average intersaccadic intervals at 973 ms ( 101 ms ) , 974 ( 103 ) , and 972 ( 100 ) , respectively . Average intersaccade times for the first five intervals serially ( excluding the first ) for Animal 1 are 991 ms ( 104 ms ) , 988 ( 102 ) , 1 , 028 ( 118 ) , 1 , 041 ( 129 ) , and 1 , 024 ( 126 ) . Animal 2 produced average intervals of 984 ms ( 94 ms ) , 971 ( 101 ) , 948 ( 99 ) , 976 ( 110 ) , and 978 ( 105 ) . Because the behavioral comparisons within each animal and between animals are very similar , all future analyses will combine data for all saccadic intervals ( except first intervals ) and both animals . We also compared successive intersaccadic intervals produced by the animals during the self-timed rhythmic-saccade task . Previous studies have noted that a negative correlation exists between subsequent repetitive behaviors , such as finger taps or saccades [38]–[40] . However , these studies focused on tempo reproduction rather than a self-timed behavior . An investigation of successive intersaccade times in our self-timed task revealed that the combined behavioral data displayed a small but significantly positive correlation ( r = 0 . 05 , p<0 . 0001 ) . The lack of r value consistency with previous studies is likely due to task differences . Unlike tempo reproduction , our task resets the required timed interval after each movement . Thus , there is no behavioral advantage to compensating for a short interval with a long one , as in a tempo reproduction task . In order to investigate the neural basis of this temporally precise behavior , we recorded individual neurons from a parietal area that has previously been implicated in timing , the lateral intraparietal area ( LIP; Figure 3A , B ) . We first examined the activity of individual neurons during a mapping task ( Figure 1C ) to ensure that the neurons we recorded displayed spatially selective activity typical of LIP neurons . Consistent with previous reports , we found spatially specific stimulus onset responses and presaccadic buildups in activity ( Figure 3C , D ) . Many neurons within LIP also exhibit stereotyped delay period activity during tasks in which the location of a transiently presented saccade target must be remembered , such as the memory-guided delayed saccade task [41]–[43] . As seen for both an example neuron ( Figure 3E ) and our population ( N = 100 , Figure 3F ) , the flashed target within the RF ( black bar ) elicited a transient increase in activity followed by sustained activity during the delay period ( the time between the RF target being extinguished and the fixation point being extinguished ) . This sustained activity remains above activity levels that were observed during mapping trials for non-RF locations ( dashed lines in 3C , D ) . Additionally , many neurons in our population displayed pre-saccade-related activity ( 59/100 showed this trend , while in 25/100 this increase was significant; t test , p<0 . 05 ) . This steady increase in activity is evident before saccade onset and can be seen in both the example cell and population activity . Such pre-saccadic activity has previously been attributed to saccade planning [34] , [41]–[43] . However , this increase could also be due to reward expectations or changes in attention closely linked to the required movement [24]–[29] . To quantify the level of delay period activity [34] , we calculated a memory index ( MI ) by dividing the average delay activity ( 320–720 ms following target onset ) by the average visual activity ( 0–320 ms following target onset ) [16] , [44] . Although on average our MI was larger than some reported previously ( mean = 0 . 78 , SD = 0 . 26 versus mean = 0 . 52 , SD = 0 . 63 [44] ) , this difference is not significant and to be expected given our exclusion of cells with no delay period activity . Also consistent with previous reports , we found a significant correlation between the MI and presaccadic activity ( r = 0 . 22 , p<0 . 05 versus r = 0 . 30 , p<0 . 05; [44] ) . After this memory-guided task , we recorded from the same neurons while the animals performed the self-timed rhythmic-saccade task ( Figure 1A , B ) . In the following analyses concerning the self-timed rhythmic task , the first interval was excluded because of the predictable onset of the subsequent target , which distinguishes the first interval from all subsequent intervals . As was done with the behavior , firing activity was segregated based on direction and the interval number within a trial ( Figure S1 ) . The firing rate for each direction and interval ( first intervals done separately ) was then analyzed in order to determine if the activity varied from interval to interval within a trial . No significant correlation was found between neuronal activity and interval number for either monkey ( p>0 . 05 ) . Therefore , as with the behavioral data , neural activity is combined across all intervals ( aside from first intervals ) in subsequent analyses . Because the firing activity between both monkeys was very similar , data from both animals will be combined in all future analyses as well . Figure 4A shows the neural activity of an example cell as the monkey performed the self-timed saccade task . The dashed vertical line represents saccade initiation . Blue to red traces are aligned to peripheral saccades ( saccades to the peripheral or RF target ) , while red to blue lines are aligned to central saccades ( saccades to the central target ) . The combined , average population activity for all 100 neurons is shown in Figure 4B . As expected , irrespective of the interval , response rates are consistently higher during the central fixation when a target is within the RF ( blue ) than during peripheral fixation ( red ) when there is no target within the RF . However , in contrast to expectations from memory-guided tasks , there is no significant pre-saccadic buildup in activity for the example cell or the population . As reflected in the population activity , the majority of the neurons ( 81/100 ) did not display significant increases in activity immediately prior to saccades . On the contrary , as evidenced by the population averages , activity decreases prior to saccade initiation for both directions of movement ( Figure 4B ) . To isolate the factors that could cause temporal modulations of activity prior to saccades , we compared the activity observed in the self-timed saccade task with activity observed in mapping and memory-guided saccade tasks for the same fixation location ( central fixation ) ( Figure 4C , D ) . The left portion of the plots are aligned following a central saccade ( blue ) or target onset ( black ) , while the right portion of the plots are aligned to peripheral saccade onset ( blue and black ) . In all of these tasks , the actual upcoming movement ( a peripheral saccade ) and the timing ( a 1-s interval ) are consistent . However , the stimulus events and rewards associated with this planned movement are different . For mapping trials , although the peripheral target remains lit for the entire interval , a visual cue tells the animal when to move ( central fixation point turns off ) and the animal only makes one saccade prior to receiving a reward . For memory trials , the peripheral target is only lit for the first 200 ms of the trial , the animal is cued when to move , and the animal only makes one saccade prior to reward delivery . In the mapping task , a transient increase in activity occurs following RF stimulus onset . A similar response is also evident during the self-timed saccade activity ( blue , left ) , although the cause is likely due to a central saccade bringing the RF to encompass the peripheral target instead of target onset since first saccades are not included in the rhythmic analysis ( Figure 4C ) . After the initial response , firing rates decrease in both the mapping and self-timed tasks . However , during this period the mean firing rate is consistently higher in the self-timed task , in which the animal cannot rely upon an external cue , than in the mapping task , in which such cue is available . This difference in firing rate is therefore consistent with a temporal production signal . Similar differences are observed when the responses prior to memory-guided and self-timed saccades are compared ( Figure 4D ) . Notably , the mapping response is virtually identical to the memory-guided response , indicating that the presence or absence of a stimulus within the RF prior to the saccade cue has a minimal effect on responses . This is also consistent with a temporal production signal , rather than any sensory effects , dominating LIP responses during the self-timed task . Immediately prior to the saccade , activity rises much more in both the mapping and memory guided tasks ( black ) when the animal knows that a reward is imminent than in the self-timed task ( blue ) . This difference suggests that the pre-saccadic rises in activity reported in previous studies may reflect specifics of the task ( such as reward or sensory anticipation ) , rather than generalized patterns underlying saccadic timing . Although the predominant feature of activity modulation during self-timed saccades is a near linear decline in firing rate over time , other modulations are clearly present . Around the time of saccade onset ( ±100 ms ) , the activity displays distinct modulations . Brief increases in activity just prior to saccade onset are followed by short intervals of decreased activity at the time of saccades . These peri-saccadic modulations in activity are similar between saccade directions and are consistent with previous studies as being signals of a global remapping of the RF [45]–[48] . The largest deviation from the overall decline in activity is the sudden increase in activity immediately following central saccades ( blue , 0 to 250 ms ) ( see Figure 4B ) . The increase in activity immediately following a central saccade is consistent with bottom-up sensory stimulation because , as an immediate consequence of the central saccade , the peripheral target is moved into the neuron's RF . A large increase in activity is also visible early within the interval prior to peripheral saccades ( red , −1 , 000 to −700 ) . Since saccades are performed back and forth between the two targets and since all trials outside of the first intervals are analyzed together , this increase may also represent sensory stimulation of the RF . The cyclical nature of the task makes it difficult to ascribe firing rate changes to particular saccades because , for example , firing rate changes following central saccades are also preceding peripheral saccades . To address this ambiguity , we took advantage of the variability in intersaccadic intervals . Specifically , we looked at whether activity locked to a particular saccade could completely explain the peri-saccadic activity aligned to the other saccade by generating firing rate predictions of each saccadic alignment on the basis of the other ( Figure 5A ) and behavioral variability . A good fit between the predicted rates ( green traces ) and the observed firing rates ( red and blue traces ) would indicate that activity locked to a particular saccade can largely explain the firing rate changes seen in the cyclical task . Conversely , a poor fit would suggest that activity aligned to a particular saccade cannot solely account for the observed neural activity . For example , if LIP activity were strongly modulated by both central and peripheral saccades , then a firing rate reconstruction based on only one of those saccades would poorly predict peri-saccadic activity for the other saccade . However , if the fit was good for one type of saccade and relatively poor for the other , it would indicate that firing rate modulations could be largely explained by only one of the two saccades . In general , the predicted rates were similar to the actual firing rates , consistent with an overall decrease in firing rate being the dominant activity modulation during intersaccade intervals ( Figure 5B , C ) . However , there are a few instances for which the predicted activity poorly fits the observed activity . The first of these is the time period just before and just after both directions of saccade initiation ( ±100 ms ) , which , as mentioned previously , is consistent with previous reports of RF remapping signals in LIP [46] . The second significant discrepancy is during the 250 ms immediately following central saccades ( Figure 5B ) . By contrast , peripheral saccade aligned activity ( Figure 5C ) during the corresponding period of time ( −1 , 000 to −700 ms ) is well fit . This suggests that the sudden increase in activity immediately following central saccades , which is likely explained by the saccade-related movement of the stationary peripheral target into the RF , is largely sufficient to explain the gradual increase seen approximately 1 s prior to peripheral saccades . Moreover , activity aligned to central saccades ( green , Figure 5C ) is better at predicting the firing rate variations seen in the rhythmic task than activity aligned to peripheral saccades ( green , Figure 5D ) . Thus , the activity modulations seen in our task can be largely explained by a gradual decrease in activity following the appearance of the peripheral target in the RF . The consistency and strength of responses immediately following central saccades suggests that it could serve as a “reset” signal for timing . The linear rate of activity decrease that follows this reset could then be used to accurately measure the passage of time . However , although suggestive , these modulations need not have any relationship to timed behavior . For example , although activity decays at a constant rate following the introduction of a stimulus into the RF , this decay might not have anything to do with how the animals actually timed their behaviors and may simply reflect some intrinsic decay constant . In such a situation , activity fluctuations in LIP that are due to noise or some uncontrolled variable such as attention would have no correspondence with fluctuations in the timed behavior . To examine this possibility , we studied whether LIP activity fluctuations during intersaccadic intervals were predictive of the animals' actual saccadic interval . Figure 6A , B show the firing activity as a function of interval length for each saccade direction . Each trace is an average rate of one-fifth of the trials , and interval lengths are sorted based on current intervals both prior and subsequent to the saccade displayed at time point 0 ( activity prior to the saccade is sorted based on the interval that ends at 0 and activity following the saccade is sorted based on the interval that begins at 0 ) . The red traces represent the shortest fifth of intervals produced by the animals , while the purple traces represent the longest fifth of intervals ( as indicated by the intersaccade distribution times shown at the beginning and end of each interval ) . The red and blue bars at the bottom of each figure indicate fixation location during each interval ( red , fixation of peripheral target; blue , fixation of central target ) . The consistent ordering of firing rates with respect to interval length suggests that activity is predictive of the behavioral interval . To further examine the relationship between activity and behavior , we computed the correlation between neural activity and the intersaccadic period . We investigated this relationship over the 800 ms before and after saccade initiation by looking at correlations over 100 ms bins ( Figure 6C , D ) . Correlations were calculated on an interval-by-interval basis across all trials of all cells . For each bin , the action potentials were summed and analyzed with respect to interval length . Correlations prior to saccades were analyzed using intervals that ended at time 0 , while correlations following saccades were analyzed using intervals that began at time 0 . In this way , all correlations are concerned with current intervals associated with upcoming saccades . This analysis allows us to examine whether time-related signals are consistently present throughout intersaccadic intervals or more prevalent immediately before or after saccades . Activity at the peripheral location was consistently predictive of interval duration irrespective of alignment ( red , Figure 6C , D , correlation analysis , * p<0 . 005 , # p<0 . 05 ) . These correlations are notable in two respects . First , they occurred during a period of time when there is no sensory stimulation in the RF and the RF is not a potential target . Second , and consistent with the orderly segregation of the traces in Figure 6A , B ( red shaded intervals ) , the correlations were positive , meaning that increases in activity were associated with increases in the intersaccadic interval . This is the opposite relationship to what would be expected by previously proposed activity-based threshold models of timing [4] , [22] and sensory integration [49] , in which increases in activity are associated with reaching a threshold earlier . Activity at the central location aligned to peripheral saccades also displayed significant correlations to interval length throughout nearly the entire intersaccade period ( blue , Figure 6D , * p<0 . 005 , # p<0 . 05 ) . However , in this case , the correlations are primarily negative . Surprisingly , the activity aligned to follow central saccades corresponding to the same fixation location does not consistently show significant correlations throughout the interval ( blue , Figure 6D , p>0 . 05 ) . This difference suggests that certain events , such as the sensory-driven response transient following central saccades which dominates response modulations in our task ( Figure 5 ) , can mask temporal production signals . Correlations between firing rate and saccade metrics ( saccade velocity ) were also calculated in order to determine if LIP activity was related to saccadic motor output . We found that the overall population activity was not significantly correlated to saccadic velocity ( p>0 . 05 ) . Therefore , LIP activity is likely related to motor planning rather than saccade metrics . Further support for motor planning is provided by the difference in the correlations between the self-timed task ( blue ) and the memory task ( black ) ( Figure 6D ) . Although these tasks are similar in that the same movement is required after the same delay , significant correlations only exist throughout the interval for the self-timed task . Since firing rates throughout the delay periods of these intervals are largely predictive of the interval length , this suggests that activity within LIP is a temporal production signal that can be utilized in order to time saccade initiation . Consistent with previous reports we found that a subset of individual cells were correlated to interval duration ( 10/100 prior to central , 8/100 following central , 11/100 prior to peripheral , 3/100 following peripheral , p<0 . 05 ) [22] . The sign of the cells' significant correlations typically had the same sign as population correlations . The low number of neurons displaying significant correlations between rate and interval length suggests that the timing signals of LIP in our task are most prominent at the population level but too weak to be observed in the activity of most individual neurons over the time period of our recording sessions . When the activity of each individual neuron was normalized ( z score ) prior to calculating the correlations , the correlation values throughout the interval generally maintained the same sign ( unpublished data ) . However , the r values were consistently dampened . This suggests that neurons with greater modulations in activity contribute more to behavioral timing . When we looked at neurons with higher degrees of rate changes , we found that activity displayed stronger correlations while the animals were fixated at the peripheral target ( Figure 6 , red bars ) . Overall correlations were significant for precentral and postperipheral saccade aligned activity and increased from 0 . 050 and 0 . 076 to 0 . 145 and 0 . 157 , respectively . Yet correlations were not significant for either saccade alignment as animals fixated on the central target in the high modulation cells . Cells with lower modulations in activity still displayed significantly correlated activity both prior to ( r = −0 . 042 ) and following ( r = 0 . 039 ) peripheral saccades , although the correlations following peripheral saccades were reduced compared to the combined population . These differences between cells with high and low modulations may support the idea that there are two populations of neurons within LIP [22] , each of which contributes to timing differently . However , since both populations of cells do contribute to the timing of saccades , we will continue to discuss temporal production for the entire combined population . Although we observed that LIP activity is significantly predictive of current interval length , other studies have shown that parietal activity can also represent past and future events [29] , [50] . To determine if LIP activity is also related to past and future intervals in our task , we performed a regression analysis where we examined the relationship between neural activity with past , current , and future interval lengths . Because fixation location and the presence of a target within the RF strongly modulates responses , this analysis was done separately for both locations ( central and peripheral ) . Firing rates for this analysis were obtained from the 800 ms adjacent to saccade onset for the intervals with the highest overall correlation values . We found the population activity to be significantly related to the current intervals at both locations ( linear regression coefficients: at center = −9 . 4 spikes/s/s , at RF = 7 . 9 spikes/s/s , p<0 . 005 ) , but not with past or future intervals ( p>0 . 05 ) . Very few individual cells were significantly related to any intervals . Only 0 , 1 , and 0 cells were significant for past , current , and future intervals , respectively , while the animals were fixated at the central target . Similarly , no cells were significant while fixated at the peripheral target ( regression analysis , p>0 . 05 ) . Therefore , it appears that neither future interval planning nor past interval production significantly contributed to LIP activity , and the relationship with current intervals is due to population activity . The change in the sign of the correlation between activity and timed saccades suggests that a push-pull mechanism may underlie temporal production in our task . Activity prior to a RF ( peripheral ) saccade “pushes” for saccade initiation in that more activity during this time leads to a faster onset of the behavior ( negative correlation ) . By contrast , activity prior to a central saccade “pulls” for maintaining the peripheral position . In our task , the hemisphere containing the response fields corresponding to the impending saccadic target would be “pushing” for a saccade , while the opposite hemisphere would be “pulling” or delaying a saccade . Because activity fluctuations are associated with the same change in timing no matter when they occur within the intersaccadic interval , the mechanism must involve the integration of activity with very little decay [49] . However , in contrast to previous proposals , LIP activity cannot be a direct proxy of a saccadic decision variable , since activity neither rises over time nor is it associated with a constant value immediately prior to saccade onset ( Figure 4 ) . The simplest realistic model is therefore one in which a difference between the temporally integrated activity of the two LIP locations representing the targets underlies the decision to saccade . Because we have data for saccades both towards and away from the RF , we can use the average population data of Figure 4 to construct such a model . Specifically , we integrate the activity over the cells representing the impending target and cells representing the “anti-target . ” When such integrals are differenced , the result is a signal that increases in a near linear fashion throughout the interval prior to saccade initiation ( Figure 7B , C , black trace ) . A threshold is then applied ( Figure 7B , C , dashed horizontal line ) so that when the “push” of the RF location outweighs the “pull” from an opposite location by a set amount ( threshold ) , a saccade is signaled . On average , this threshold is reached at the beginning of the presumptive peri-saccadic remapping signal , consistent with typical latencies between LIP activity and saccades ( ∼100 ms ) [51] . In the absence of any activity fluctuations from the responses shown in Figure 7A , this model would produce completely regular intersaccadic intervals . However , in the presence of activity fluctuations in one hemisphere , either due to noise or changes in an uncontrolled variable such as attention , this regularity changes , allowing for a variety of intersaccadic durations . Because of the differencing operation , fluctuations at the different locations have opposite effects ( Figure 7B , C ) . Because of the integration , brief activity fluctuations anytime during the interval have an equivalent effect on intersaccadic duration . Thus , a momentary increase ( or decrease ) in activity will have an identical effect of the timing of the impending saccade no matter when it occurs . To test how this model compares with experimental observations , we plotted how differences in the integrated activity would increase or decrease the intersaccadic interval with the same constant threshold ( Figure 7D , black line ) and compared these predictions with the activity differences observed in our interval sorted response plots ( Figure 6A , B ) . The model ( slope = 239 ms/spike , black ) accurately predicts the average responses prior to central ( Figure 6A , squares in Figure 7D ) and peripheral ( Figure 6B , triangles in Figure 7D ) saccades for timed intervals of different durations .
In order to investigate neural activity related to temporal production , we devised a self-timed rhythmic-saccade task that controlled for temporal measurements while minimizing sensory and reward anticipation . Animals were required to make saccades back and forth between two fixed targets at a fixed interval so that saccades occurred each second ( Figure 1A , B ) . We found a systematic decrease , rather than an increase , in activity within LIP prior to saccades . The systematic decrease in activity was significantly predictive of intersaccadic interval length ( correlation analysis , Figure 6 ) . The relation to interval length was found to only be significant for current ( not past or future ) intervals ( regression analysis ) , but the sign of the correlation between activity and timed intervals depended on the direction of the impending saccade . The animals in our study displayed the ability to precisely and consistently produce a rhythmic behavior very near the trained interval ( Figure 2 ) . This was true regardless of the direction of the saccade ( peripheral or central ) or the saccade number ( second , third , etc . ) . However , aspects of our results are not consistent with previous studies that investigated motor timing in repetitive behaviors . One timing model used to describe rhythmic saccades [40] was developed by Wing and Kristofferson from studies utilizing finger tapping [38] , [39] . The model stipulates a negative correlation between subsequent timed repetitive movements . For example , if a saccadic interval is longer than the trained interval , then the following saccadic interval is likely to be shorter . This negative correlation helps ensure accuracy in tempo replications , since short intervals can be compensated by longer intervals ( and vice versa ) in order to stay on beat . This negative relationship has been attributed to the variability of motor output delay and the idea that chance variations about the mean delay will tend to produce a negative correlation between adjacent intervals [52] . In our design , we failed to find a negative correlation between subsequent timed intervals . Instead we found a small ( r value = 0 . 05 ) , but significantly positive correlation ( p<0 . 0001 ) between subsequent saccadic intervals . The fact that our results do not display the negative correlation described by the Wing and Kristofferson model likely reflect task differences . Unlike our task , these other studies did not require their subjects to precisely execute a trained interval . Instead , the subjects first followed along with a cued motor sequence and then continued the motor task in a self-paced manner after cue extinction . There were no repercussions for imprecise timing as in our task ( trial ends , no reward ) . Additionally , our task resets the trained interval following each saccade . This means that the better the animal is at precisely producing the trained interval on each saccade , the better chance it has of receiving a reward . In these other studies , the subjects attempted to replicate a tempo that is not reset with their behavior . Therefore , the Wing and Kristofferson model may only be useful in describing tempo replication and may not reflect general mechanisms governing temporal production . The near independence between adjacent timed intervals suggests that the underlying temporal production signal is being reset by each saccade . The notion of saccades effectively resetting time keeping in our rhythmic task is consistent with our physiological observations in several respects . First , as evidenced by saccade aligned firing rates , a similar up-down peri-saccadic modulation in firing is present irrespective of saccade direction . These modulations in activity occur approximately 100 ms prior to saccade initiation , similar to saccadic latency times within LIP [51] . Therefore , remapping may serve as a reset signal . Second , as would be expected by a reset , activity in LIP was only correlated with the current behavioral interval within a sequence and not with past or future temporal production . Third , the relationship between activity and behavior flips after a peripheral saccade from negative to positive . Neurons within our sample exhibited response properties largely consistent with previous reports from LIP . They had spatially specific response fields depending on the direction of the saccade target ( Figure 3C , D ) , responded to target onsets within those response fields , maintained responses when delays were imposed between target extinction and saccade initiation ( Figure 3E , F ) , and showed peri-saccadic modulations ( Figure 4 ) consistent with remapping [46] . Our population of neurons also displayed task-dependent anticipatory behavior . For instance , increases in activity are observed just prior to saccade onset for tasks in which the movement is cued and only a single saccade is required ( Figure 3C–F ) . However , this same level of presaccadic increase in activity is not apparent in the population activity during the self-timed task ( Figure 4A , B ) . This suggests that climbing activity may be associated with sensory and/or reward anticipation rather than motor anticipation . Response rates were consistently higher prior to saccades in the self-timed task than saccades in the memory-guided task . If there were saturation issues , this higher response level might preclude our ability to observe presaccadic buildups . We consider this unlikely for two reasons . First , the mere presence of an RF stimulus does not preclude our ability to see presaccadic buildup , since a buildup is evident during the mapping task ( Figure 3C , D ) . Second , the light-sensitive response early within the interval ( Figure 4A , B ) shows that the population is capable of higher rates of firing . In our self-timed task , although activity consistently declined over time , the mean levels of activity were significantly different for the two fixation locations ( Figure 4 ) . Because the “real-world” or head-centered positions of our targets remained constant throughout the trials ( Figure 1A , B ) , this response difference could be due to the presence or absence of a target within a retinotopic RF . However , as can be seen in the comparison of mapping to memory-guided responses , the presence or absence of a stimulus within the RF during a 1 s interval had very little effect on responses ( Figure 3D , F ) . In both cases , a low-latency stimulus-evoked onset response is followed by a gradual decrease in response during the delay period and then an increase in activity prior to the saccade , and the response rates during all of these phases are virtually identical . Another potential complication is that eye position varies between the two intersaccadic intervals . Approximately 50% of LIP neurons are modulated by eye position [41] . In many cases , this modulation can be described as a gain effect , in that visual and delay responses are consistently modulated by a single factor according to eye position . For example , in one study , 30% variations in response magnitudes were reported over eye-position ranges of 45 by 45 deg [41] . Although this is approximately the modulation seen between our central and peripheral locations ( Figure 4B ) , it is unlikely that an effect that is only observed in half of LIP neurons would give this degree of modulation over the eye-position shifts ( typically around 15 deg ) in the self-timed task . Furthermore , because the direction of the eye position effects is not consistent between neurons , any effect seen in individual neurons would average out to zero over a relatively large sample such as ours . Finally , as a part of our eye calibration procedure , the position of the central fixation point on the screen varied from trial to trial along the corners of a 4 deg square , which would further reduce any eye position effects . Thus , we feel that the most likely explanation for the difference between central and peripheral fixation is the well-established sensitivity of LIP responses to the location of the impending saccade . However , even if other factors contribute to this response difference , the interpretation of our data with regard to timing remains unaltered . Because during a single intersaccadic interval neither the stimulus nor the eye position is changing , response changes during this interval cannot reflect these factors . Since all visual stimuli are stationary and the reward cannot be anticipated , the only factor that consistently varies over the intersaccadic intervals is the passage of time . Consistent with a role in the representation of time , we found significant correlations on an interval-by-interval basis between neuronal activity in LIP and the duration of this interval . Because these correlations are present for both locations , they are robust to changes in eye position , direction of the impending saccade , or whether there is a stimulus within the RF . Since an internal sense of time is the only cue available to the monkey with which to initiate saccades , we have interpreted this correlation , which , to our knowledge , has never been previously reported , as reflective of a temporal production signal . The specificity of these correlations , which are absent in a task not requiring timing ( Figure 6D , black ) and reverse sign according to the specific saccade which is impending ( Figure 6C , D , blue and red ) , rule out that they are the result of a generalized vigilance or task-related factor . However , it is also possible that LIP , instead of representing information relevant to a decision to saccade , namely the passage of time , instead represents a motor plan whose execution after a decision has been made can be delayed . We consider this explanation unlikely for several reasons . First , there is no evidence to suggest that increases in activity in LIP would be associated with delays in the execution of a motor plan . On the contrary , many experiments have demonstrated that LIP activity appears to be associated with predecision information , whether that information be stimulus related [23]–[24] , [49] , [53]–[63] or time related [4] , [16] , [22] , [64] . Second , we found no evidence that actual saccade metrics ( e . g . , velocity ) depended on LIP activity . Third , because LIP activity was correlated with timing even when the RF location was not a potential target ( e . g . , when the monkey was fixating at the peripheral location ) , our results are not consistent with changes in a motor plan strictly associated with a particular retinotopic location . Fourth , if LIP activity solely reflected a motor plan , then activity fluctuations near the time of the saccade should have particularly strong correlations with behavior . By contrast , we find that behavioral correlations are relatively constant throughout the entire intersaccadic interval , even when the saccade is going to occur 800 ms in the future . Finally , since the motor plan for memory-guided and self-timed saccades are identical , one would expect little difference in firing rates or behavioral correlations , in contrast to our observations ( Figures 3 and 6 ) . Neuronal representations of time within LIP have previously been described by climbing activity , a steady increase in neuronal activity over time to a threshold level , at which time an action ensues [4] , [16] , [22] . A higher rate of activity ( or a faster rise to threshold ) produces a shorter interval and therefore a negative correlation between rate and time . Although brief periods of increased activity can be seen in our population activity ( Figure 4B ) , these increases can be explained by RF remapping [46] and sensory responses to the peripheral target being moved in and out of the RF as the animal produces saccades . These brief periods of increases in activity do not fit the parameters of climbing activity as a timing signal [65] , [66] . Instead , the prominent pattern of activity is a steady decrease in firing rate over the delay period . One possibility for why we observed falling , as opposed to climbing , activity prior to saccades is the differences between our task and those employed previously to study timing . Because of the close associations of sensory cues and reward that occur near the time of the behavior in previous studies , which are absent by design in our task , it is possible that climbing activity is more related to sensory and/or reward anticipation ( time measurement ) than motor planning . This notion is consistent with our observations of neuronal activity during a task that is much more analogous to previous studies . The same neurons that displayed falling activity during the self-timed rhythmic-saccade task displayed very different activity during the memory-guided delayed-saccade task . In the memory task , the basic behavior , namely waiting 1 s prior to making a saccade , is similar to the self-timed task . However , in terms of temporal measurement , the tasks are quite different . Specifically , in the memory task ( and unlike the self-timed task ) , the timing of the cue to make a saccade and reward can be readily anticipated . Consistent with previous observations , a presaccadic rise in activity was observed in the memory-guided task . However , this rise is largely absent in the self-timed task , suggesting that climbing activity may reflect reward anticipation rather than a motor plan . Given these task differences , it is also possible that distinct timing systems are responsible for tasks that require temporal measurement and those that do not . Lewis and Miall [67] have proposed that there are two distinct timing systems: an automatic system responsible for predictable intervals defined by movements and a cognitively controlled system involved in temporal measurements that direct attention . Since LIP is involved in both motor production and attentional allocation [23] , [68]–[73] , it may be that this area is a part of both timing systems and that the task determines which timing system is utilized . For instance , when the animal is performing an interval duration comparison task or a task for which movement is cued or immediately rewarded [4] , [16] , the cognitive timing system would be engaged since these tasks require the timing of discrete epochs and do not control for the attentional effects that sensory and reward anticipation can have [74] , [75] . The cognitive system would employ climbing activity as its timing signal in order to time the temporal measurement-related events of the task . However , when those forms of anticipation are minimized ( as in our self-timed delayed-saccade task ) , the automatic timing system may be engaged since the task requires saccades be made at regular intervals . This would then allow falling activity , the signal responsible for the production of the timed interval , to emerge as the temporal production signal . If the primary role of the cognitive timing system is to direct attention , it may be particularly unlikely to play a role in our task given that the spatial positions and direction of the impending saccades are never ambiguous or subject to cognitive choice . Although spatial attention may not be required , this does not mean that attention is not allocated to the targets at some point prior to saccade initiation . However , the activity we observe during the self-timed saccade task displays a decrease in rate prior to saccade initiation , not an increase as might be associated with the increasing priority of making a saccade as time elapses [76] . Another difference between our task and previous single-saccade tasks is the potential for sequence planning in our task . Psychophysical evidence suggests that , when confronted with an array of saccade targets , subjects naturally plan entire saccade sequences [77] . The planning of entire sequences has been shown to take place in a number of brain regions and for a number of tasks [50] , [78]–[83] . Additionally , a study by Seo et al . [29] showed that LIP activity contains information about past events . However , we found that neither future nor previous temporal production significantly contributed to the activity of our population during this task ( regression analysis , p>0 . 05 ) . Also , our observation of near independence between adjacent intervals ( correlation value of 0 . 05 ) is not consistent with sequence planning . These data suggest that , presumably because we reset the behavioral requirement after each saccade , both the animals and our neural population are concerned solely with timing single intervals within the rhythmic sequence . The observation that activity is only correlated with the present interval implies that the correlations observed prior to central saccades to timing do not reflect past or future planning of peripheral saccades . This correlation , which gave rise to our “push-pull” model , is inconsistent with the belief that LIP RFs are exclusively tuned for contralateral visuomovement space [84] . In a purely retinotopic framework , activity prior to a central saccade should be minimal and irrelevant given that there is not a stimulus in the RF nor is that location a potential saccadic target ( Figure 1 ) . However , a study by Dickinson et al . [85] found that neurons in LIP can be activated by the instruction to perform a saccade , in the absence of any spatial information . Similarly , Bennur and Gold [63] found information on perceptual decisions within neurons whose RFs did not correspond with the upcoming movement . Our results also have implications regarding the role of LIP activity changes in directing eye movements . A common theory regarding LIP is that its activity can represent the accumulation of saccade-relevant evidence . In this notion , a saccade is initiated once activity reaches a threshold . For example , during two choice motion discrimination tasks , stimulus-dependent climbing activity has been observed as animals monitored a weak motion stimulus whose direction indicates the saccade target . While the rate at which the activity accumulated was dependent on motion strength , consistent levels of activity were observed prior to saccades irrespective of the stimulus [49] , [54] , [56] , [86] , [87] . In subsequent models , such results were explained by LIP neurons integrating the motion evidence arising from MT inputs [49] , [56] . Evidence consistent with a threshold mechanism has also been observed in an LIP timing study [22] . Our results differ in two fundamental respects: activity does not accumulate over time and there is not a common activity level prior to saccades ( Figure 6 ) . Thus , increases in delay period responses prior to saccades are not a universal characteristic of LIP neurons [44] . However , our model is of the same basic form as those based on the motion-based decisions in that relevant evidence is integrated and thresholded to arrive at a decision . The critical distinction is that , unlike the motion-based decision models in which the evidence for the direction of motion is present in MT neurons and evaluated by LIP neurons , in our model , evidence for the passage of time is present in LIP neurons and integrated elsewhere to arrive at a decision . In any case , the observation that firing rates in LIP are dependent on task design is evidenced by the difference in our population between mapping , memory-guided , and self-timed saccades ( Figure 4C , D ) . This demonstrates that LIP activity can only be interpreted with knowledge of the behavioral context [44] , [63] , [88] . For example , LIP activity cannot be strictly interpreted as reflecting an evidence signal whose magnitude is associated with the increasing likelihood of reaching a decision to saccade , since in our experiments activity decreases with the passage of time , which is the sole evidence the animal can use to make a saccade . Similarly , our data are not consistent with LIP solely representing an attention signal , since there are no stimulus cues present or relevant for saccades , and the observation of positive correlations between activity and interval means more activity can actually delay a saccade . Our results also constrain the spatial distribution of timing signals within the brain . Two traditional theories concerning where timing signals originate are the central and distributed timing mechanisms [1] , [89] . In the central timing model , a specific brain region produces a timing signal that is utilized for all timing-related events for all modalities . The distributed timing model suggests that there is no dedicated timing system but that the ability to represent time is an intrinsic property of distributed cell populations that are required for a given task . If LIP activity strictly reflected a broad timing system ( like those described by centralized timing models ) , its activity would have a consistent relationship with time irrespective of saccade direction . Because activity patterns and behavioral correlations depend in a number of respects on the particular planned saccade , our results support the notion that local neuronal populations are responsible for temporal production . First , the activity immediately preceding central and peripheral saccades is different when sorted by intersaccadic interval . Prior to central saccades , there is no evidence for a response threshold because different rates are seen at saccade onset ( Figure 6A ) . By contrast , a common activity level is observed at peripheral saccade onset . Second , although activity was consistently predictive of saccadic interval duration , the exact relationship was significantly different for peripheral and central saccades . Activity prior to saccades made to the peripheral target was negatively correlated with interval production , while activity prior to saccades to the central target had a positive correlation . Our results also provide insight concerning the neural mechanisms underlying timing . Multiple mechanisms have been proposed to underlie behavioral timing . Three mechanisms include the clock ( pacemaker/accumulator ) model , labeled lines , and population clocks [1] . In the clock model , a neural pacemaker produces rhythmic pulses . These pulses are then counted ( or accumulated ) in order to time an event . Clock models are generally classified as centralized systems , as this one clock is used in all timed events [1] . Because the relationship between LIP activity and behavior varies depending on the impending saccadic target , it is not consistent with a single universal representation of time . Moreover , because activity was observed to decrease rather than increase over time , accumulation as a single timing mechanism is ruled out . In the labeled line model , different neurons within a population respond at different interval lengths . For example , one neuron may respond at 100 ms while a second neuron responds at 200 ms . Labeled line models could work in a distributed timing system . For example , the labeled line population could be used to determine time according to the subset of neurons that are active . However , our data did not show strong evidence of individual neurons being significantly correlated to specific intervals . The population clock model encodes time through the population activity of a network of neurons . There is no specific time at which neurons are active , instead dynamic interactions or time-dependent changes between neurons within the network provide information about lapsed time ( e . g . , short-term synaptic plasticity , inhibitory feedback , etc . ) . Such a model could account , in part , for the small correlation values we observed between neuronal activity and interval length if the neuronal populations underlying timed behavior were much larger than our sample . In this model , individual neurons will not contain large amounts of temporal information , consistent with the low number of individual cells that display significance in the regression and correlation analyses . From this model , we would predict that as we sampled more neurons from this population , the population correlation values would increase . The fact that activity fluctuations are correlated to timed interval production for both saccades toward the RF and away from the RF , but those correlations are opposite in sign , suggests that activity differences between the two hemispheres may drive temporal production in our task via a push-pull mechanism . While “push-pull” models have been invoked in the past in the context of sequential saccades [90] , our model relies on the specific and consistent dynamics of the push and pull signals suggested by the linear changes in firing rates over the intersaccadic intervals we have observed . The success of a simple version of such a model , a differencing of the integrated activity between hemispheres , in explaining the quantitative relationship between activity and interval supports such a proposal . Although this model does utilize climbing activity and a threshold , the location at which the push and pull signals are integrated and compared in order to signal a saccade is downstream of LIP . Therefore , LIP activity does not represent the evolving intent to make a saccade but rather provides a measure of the time elapsed since the last saccade to a particular location . Although activity from other areas , such as the frontal eye fields , could contribute to opponent push-pull fixation-saccade signals such as we propose , the mere presence of opponent signals is not sufficient to explain the behavioral timing we observed . As demonstrated by our model , the signals must vary over time in a very specific and consistent manner in order to produce precise timing . Combined with the behavioral and physiological evidence of a reset upon every interval , our results support the notion of LIP activity providing clock-like inputs to a subsequent decision mechanism . The decaying activity that we observed is ideally suited for the accurate representation of time . First , the activity is accurately “reset” by the vigorous sensory response evoked by the saccade-driven appearance of a target within the response field ( Figure 5 ) . Second , activity decays at a constant linear rate , so that for any increment in time a common decrement in response is observed ( Figure 4 ) . Finally , perturbations in activity , which might be seen as a clock skipping a beat , have an equal effect on the eventual behavior no matter when in the interval they occur ( Figure 6 ) . The success of this simple model demonstrates that decaying activity within neuronal populations is sufficiently accurate to explain timed behaviors on the scale of seconds and suggests that temporal production may generally reflect competition between localized and precise timing representations .
All surgeries were done under aseptic conditions and full anesthesia in accordance with the animal care guidelines of the University of Minnesota and the National Institutes of Health . Two male monkeys ( Macaca mulatta ) ( 8 . 3 and 9 . 3 Kg ) were seated in a darkened room in front of a computer monitor . Animal training began by learning to maintain fixation ( within 3 deg ) of a single target in order to receive a juice reward . The length of the fixation requirement was gradually increased to 1 , 000 ms . Once the animal could fixate for a full second , a second target appeared and the initial target was extinguished . Animals then had to fixate on one of the two targets as each was displayed for 1 , 000 ms . Each target was identical in size ( 0 . 15 deg ) and color ( white ) . The number of fixations ( and therefore the number of timed-delayed saccades ) required was gradually increased and randomized ( 2–10 saccades ) . To this point , we have somewhat mimicked previous designs , in that temporal measurement ( the regularity of the fixation target disappearance and cue appearance ) and temporal production ( the regularity of saccades ) were both likely to be occurring . To encourage the animal to rely solely upon temporal production signals , the luminance of the nonfixated target was gradually increased until both targets remained on constantly and with equal luminance throughout the trial . Upon isolating a neuron , we recorded responses while the monkey performed a delayed-saccade task to a variety of locations . In this “mapping” task , targets were randomly sampled from eight different locations fixed radially about a mapping center ( typically 10–14 deg eccentricity ) at a distance of 4 deg . The animal performed single-saccade trials in which he was required to saccade to the peripheral target after waiting 1 s ( Figure 1C ) . The location with the maximum visually evoked response was defined as the neuron's response field ( RF ) and used as the peripheral position in all subsequent tests . Memory trials were saved for 39 out of the 100 cells that displayed LIP activity . Animals then performed a memory-guided delayed-saccade task [36] , [37] to this RF location . The memory-guided task was used to determine which cells displayed stereotyped LIP firing activity . Trials began by requiring the monkey to fixate on a target near the center of the monitor ( Figure 1D ) . Following fixation , a target within the RF was flashed for a brief period of time ( 200 ms ) and then extinguished . The monkey was required to remember the location of the flashed target , while maintaining fixation at the center target . The monkey was then required to make a single saccade to the remembered location following extinction of the central target 1 , 000 ms after the trial began in order to receive a juice reward . To be selected for further analysis , neurons had to display a light-sensitive response to the target flashed in the RF [34] , [41]–[43] ( N = 100 ) . The animals then performed a self-timed rhythmic-saccade task . Trials began with the monkey fixating the first of two targets to appear on the monitor ( Figure 1A ) . One of the targets was positioned near the center of the screen while the other target was positioned peripherally within the response field ( RF ) [33]–[35] , [43] , [91]–[93] of the neuron being recorded ( dashed box in Figure 1A ) . Immediately following fixation of the first target ( “Initial Target” ) , the second target appeared on the monitor ( “Subsequent Target” ) . Once fixation of the initial target occurred , the monkey was required to perform saccades back and forth between the two targets so that a saccade occurred each second ( “Self-Timed Saccades” ) ( ±200 ms ) ( 0 . 5 Hz ) . After both targets had appeared , no further changes in visual stimuli occurred . The monkey was required to continue making saccades ( 2–10 ) between the two targets at the 1 s interval for a randomized trial length before receiving a juice reward ( Figure 1A , B ) . Trials randomly alternated between having the initial target appear at the central and peripheral locations to ensure that we observed both saccadic directions for each interval within the sequence . Saccade targets were constantly displayed following the first interval of the trial in order to minimize sensory anticipation [23] . To help minimize reward expectations , the hazard function , which represents the instantaneous probability of the trial ending given that it has not yet ended , was flat throughout each trial [16] , [94] . Therefore , the instantaneous probability that the animal received a reward at any given instant was kept constant throughout each trial . Trial times were randomly chosen from an exponential distribution ( decay constant = 1 , 000 ms ) . Trials could end at any point , including the middle of an interval , in order to further dissociate saccadic movements from reward . Average trial length for both animals equaled 3 . 96 s ( SD = 1 . 14 s ) . A minimum of 50 saccades was required from each cell while the animal performed the self-timed task . Eye movements made to the peripheral target ( in the direction of the RF ) are termed “peripheral” saccades , while eye movements made to the central target ( or away from the RF ) are termed “central” saccades ( Figure 1A ) . For clarity , during this task , when the animal was fixating on the central target , the response field was located at the peripheral target . The animal's next saccade would then be made to the peripheral target , toward the RF . However , once the animal fixated on the peripheral target , the RF was no longer located at either target and the next move ( central saccade ) would be made away from the direction of the RF . Although the delay period of the rhythmic and memory tasks remained the same ( 1 , 000 ms ) , the self-timed rhythmic-saccade task differed from the memory-guided delayed-saccade task in a number of ways ( Figure 1A , D ) . First , in the memory task , the peripheral target was extinguished after its initial appearance within the neuron's response field instead of remaining throughout the entire trial . Thus , the animal was required to remember the target location following the disappearance of the peripheral target . Second , the memory task always required just a single saccadic movement and rewards were always delivered immediately following a correct saccade . Lastly , the animal was cued when to make the movement by the extinction of the central fixation point during the memory task . Because the animal could rely solely on this external cue to initiate a saccade to that location , no temporal production signals were required for successful completion of the memory task . Prior to training , animals were chronically implanted with titanium head posts in order to stabilize head position . Animals were also implanted with scleral eye coils in order to monitor eye position ( sampling rate of 200 Hz ) , although an infrared eye tracking system ( iView X Hi-Speed Primate camera system , SensoMotoric Instruments ) was used most often to track eye position . Following training , animals were implanted with chronic stainless steel or customized PEEK ( polyether ether ketone ) recording cylinders . Cylinders were placed , stereotactically , in a manner that allowed electrode penetration of the lateral bank of the intraparietal sulcus ( area LIP ) . Area LIP was identified anatomically using MRI prior to recording cylinder implantation . Co-registered MRI and CT images taken after chamber placement were used to confirm electrode placement within LIP ( Figure 3A , B ) . All surgeries were done in accordance with animal care guidelines of the University of Minnesota and the National Institutes of Health . Surgeries were performed under aseptic conditions with full anesthesia . Single-cell recordings were done from 175 well-isolated neurons using standard extracellular recording techniques ( Mini-Matrix , Thomas Recording ) . Action potentials were isolated on the basis of waveform ( APC , FHC ) , sampled ( 1 , 000 Hz ) , and digitized for on and off-line analysis . One hundred of the 175 neurons sampled displayed both the light-sensitive and the non-zero memory activity during the memory task and they are analyzed here . Recordings were typically taken from the right hemisphere ( 26/50 cells for animal 1 and 50/50 cells for animal 2 ) . Visual stimulation , behavioral control , and data acquisition were controlled using customized computer software ( http://www . ghoselab . cmrr . umn . edu/software . html ) . Online analyses of average firing rate were used to determine RF locations . Offline analyses of firing rates , correlations ( corrcoef ) , and regressions ( regress ) relative to the events of the saccade tasks were done using Matlab ( MathWorks ) . Average firing rates were smoothed by convolving with a Gaussian kernel ( SD = 35 ms ) . However , no smoothing was done for any correlation analysis . Saccade onset was defined according to eye velocity ( >85 deg_/s ) in conjunction with the computer software's recognition of the animal's gaze arriving within the fixation window . Intersaccade times are defined as the interval between successive saccades . The first saccadic intervals of a trial were analyzed separately from all subsequent intervals since , for initial saccades to the central position , the appearance of the subsequent target within a neuron's RF elicited a response . Significant increases in presaccadic activity were calculated by comparing firing activity within 150 ms of saccade initiation ( −150 to 0 ) with the activity 250 ms prior to that interval ( −400 to −150 ) ( t test , p<0 . 05 ) . In order to determine what factors are associated with firing rate changes , we generated a prediction of neural activity by convolving observed neural activity aligned with one saccade direction with the intersaccade distribution times aligned with the other saccade direction . For example , if we convolve the intersaccade distribution times aligned to central saccades with the actual firing rate aligned to peripheral saccades , we get a prediction for central saccade aligned activity ( see Figure 5A for example ) [convolution: ( f*d ) ( t ) = ∫ f ( τ ) d ( t−τ ) dτ] . The difference between this prediction and the observed firing rate for activity aligned to central saccades indicates how well activity associated with peripheral saccades can completely explain task-related modulations in activity . The same analysis is then repeated using activity aligned to central saccades . Fit is given by: %Fit = ( 1−NMSE ) *100 , where NMSE is the normalized mean squared error . The NMSE is calculated by dividing the mean squared error by the explainable variance of the actual unsmoothed firing rate ( variance of firing rate over the interval − average variance of all time points of the rate ) .
|
One advantage of studying sensory systems is that external stimuli can be readily controlled and quantified . Our internal perception of time , however , is not as easily approachable . To address this challenge , we developed a task that leverages our knowledge of the neuronal circuits underlying eye movements . Monkeys were trained to move their eyes consistently at regular time intervals without any external cueing or immediate expectation of reward . The animals were remarkably precise and consistent in their timing . Recordings from individual lateral intraparietal area ( LIP ) neurons , an area associated with eye movement planning , showed a linear decrease in activity between eye movements . This contrasts with previous studies that have reported increases in LIP activity immediately prior to eye movements and suggests that expectation of reward may have influenced such results . We also found that variations in this decreasing activity were predictive of the animals’ timing . Finally , we demonstrate through a simple model that LIP activity was sufficiently precise to completely explain the timing of the animals' actions . The model demonstrates that a precise internal sense of time can arise by comparing activity across different neuronal populations .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"social",
"and",
"behavioral",
"sciences",
"decision",
"making",
"neuropsychology",
"neuroscience",
"motor",
"systems",
"cognitive",
"neuroscience",
"cognition",
"computational",
"neuroscience",
"coding",
"mechanisms",
"circuit",
"models",
"biology",
"psychology",
"central",
"nervous",
"system",
"computational",
"biology",
"animal",
"cognition",
"neurophysiology"
] |
2012
|
Temporal Production Signals in Parietal Cortex
|
Host-directed therapies ( HDTs ) constitute promising alternatives to traditional therapy that directly targets the pathogen but is often hampered by pathogen resistance . HDT could represent a new treatment strategy for leishmaniasis , a neglected tropical disease caused by the obligate intracellular parasite Leishmania . This protozoan develops exclusively within phagocytic cells , where infection relies on a complex molecular interplay potentially exploitable for drug targets . We previously identified naloxonazine , a compound specifically active against intracellular but not axenic Leishmania donovani . We evaluated here whether this compound could present a host cell-dependent mechanism of action . Microarray profiling of THP-1 macrophages treated with naloxonazine showed upregulation of vATPases , which was further linked to an increased volume of intracellular acidic vacuoles . Treatment of Leishmania-infected macrophages with the vATPase inhibitor concanamycin A abolished naloxonazine effects , functionally demonstrating that naloxonazine affects Leishmania amastigotes indirectly , through host cell vacuolar remodeling . These results validate amastigote-specific screening approaches as a powerful way to identify alternative host-encoded targets . Although the therapeutic value of naloxonazine itself is unproven , our results further demonstrate the importance of intracellular acidic compartments for host defense against Leishmania , highlighting the possibility of targeting this host cell compartment for anti-leishmanial therapy .
Protozoan parasites of the genus Leishmania are the causative agents of a wide variety of diseases ranging from self-healing or severe mucocutaneous lesions to a visceral disease which is lethal in the absence of treatment . Leishmaniasis is one of the most significant neglected tropical diseases , with an estimated 12 million people infected . Leishmania parasites have a digenetic life cycle; switching from an insect vector in which parasites dwell as extracellular promastigotes , to a mammalian host , where parasites reside exclusively intracellulary ( intramacrophage amastigote stage ) . Pentavalent antimonials ( SbV ) like sodium stibogluconate ( SSG ) have been the first-line treatment against leishmaniasis for several decades but their clinical value has become compromised by increasing treatment failure and the emergence of resistant parasites . This concern is particularly important in the Indian subcontinent where visceral leishmaniasis ( VL ) caused by Leishmania donovani is endemic and where most VL cases occur [1] . Current treatment alternatives consist of amphotericin B , miltefosine or paromomycin ( in mono- or combination therapy ) but these compounds also have drawbacks including cost , toxicity or decreased efficacy after a few years of use [2] . Although the mechanism of action of these compounds is not fully understood , they are all known to target Leishmania components , therefore directly interfering with parasite growth: amphotericin B forms a complex with ergosterol , the main sterol of Leishmania cellular membrane , leading to formation of aqueous pores and increased membrane permeability [3]; miltefosine has been shown to inhibit the parasite cytochrome c oxidase and to cause apoptosis-like processes in L . donovani [4]; and paromomycin is an aminoglycoside antibiotic that inhibits protein synthesis in Leishmania with low host cell toxicity [5] . SbV on the other hand , has been shown to target both the parasite and the host cell: SbV is reduced to trivalent antimony ( SbIII ) , which directly alters the parasite redox metabolism and antioxidant defense system , but SbV itself also indirectly affects parasite survival by increasing host cell production of toxic oxygen and nitrogen intermediates , thereby creating additional oxidative and nitrosative stress upon SbIII-sensitized parasites [6] . Antimonial anti-leishmanial activity is thus partly indirect , targeting host cell pathway ( s ) that consequently affect Leishmania intracellular development . Targeting host cell pathways to interfere with the intracellular development of pathogens is a strategy increasingly investigated for antimicrobial therapy that might bring novel therapeutic approaches in a context of increased treatment failure and poor alternatives [7 , 8] . Following this line , a recent high-throughput screening campaign against kinetoplastids at GlaxoSmithKline identified several compounds associated with human proteins with no known homologs in kinetoplastids , highlighting the possibility of targeting host-pathogen interactions[9] . Here we report the host-dependent anti-leishmanial activity of naloxonazine , a mu-opioid receptor ( MOR ) antagonist . This compound was first identified in a high-throughput screen against Leishmania donovani intracellular amastigotes [10] . We now show that it affects host cell intracellular compartments thereby inhibiting Leishmania establishment in the phagolysosomal vacuole .
Parasite strains used in this study included L . donovani 1S2D ( MHOM/SD/62/1S-cl2D ) , L . donovani 1S2D expressing the enhanced green fluorescent protein ( eGFP ) and two L . donovani clones of clinical isolates from the Terai endemic region in Nepal ( MHOM/NP/02/BPK282/0cl4 and MHOM/NP/03/BPK275/0cl18 respectively susceptible and resistant to SSG and further designated SSG-S BPK282 and SSG-R BPK275 ) . Promastigotes were maintained at 26°C in hemoflagellate modified Eagles’s medium ( HOMEM ) supplemented with 20% Foetal Bovine Serum ( FBS ) . Differentiation of promastigotes into axenic amastigotes was achieved as described previously [11] . THP-1 cells ( human acute monocytic leukemia cell line–ATCC TIB202 ) were grown in RPMI supplemented with 10% FBS and 50 μM 2-mercaptoethanol at 37°C in 5% CO2 . For Leishmania infections , THP-1 cells were treated with 0 . 1 μM phorbol myristate acetate ( PMA , Sigma ) at 37°C for 48 h to achieve differentiation into adherent , non-dividing macrophages . Cells were washed and incubated with complete RPMI medium containing stationary phase L . donovani promastigotes at a macrophage/promastigote ratio of 1/10 . After 4 h incubation at 37°C , non-internalized promastigotes were removed by 3 successive washes with PBS and incubated with naloxonazine , naloxone , β-funaltrexamine , CTOP , endomorphine , DAMGO , sinomenine , concanamycin A ( all purchased from Sigma ) or imatinib ( Cell Signaling Technology ) for 24 to 72 h . Half maximal inhibitory concentrations ( GI50 ) were determined using a high-content imaging assay as described previously [10] . Briefly , compounds were serially diluted 3-fold in DMSO , with final assay concentrations ranging from 50 μM to 0 . 02 μM ( 1% final concentration of DMSO ) , 2 μM amphotericin B and 1% DMSO were used as positive and negative controls respectively . For confocal microscopy , infected cells were washed with PBS , fixed for 30 minutes with 4% formaldehyde , rinsed again with PBS and stained with 4’ , 6’-diamidino-2-phenylindole ( DAPI 300 nM ) . Images were acquired with an LSM 700 Zeiss confocal microscope . 20 μM naloxonazine was incubated for 50 h in RPMI 10% FBS- 50 μM 2-mercaptoethanol with or without THP-1 cells at a concentration of 106 cells/ml . 100 μl of culture media were collected at different time points ( T0; 0 , 5; 1; 5; 10; 20; 30; 45 and 50 h ) and kept frozen . 20 μl of these samples were then mixed with 40 μl cold acetonitrile containing either 2 μM naloxone or 2 μg/mL K777 ( N-methylpiperazine-PhehomoPhe-vinylsulfone-phenyl ) , centrifuged , and 3 μL per sample injected into an API4000 ( AB Sciex ) LC-MS/MS system and analysed with positive-ion-mode electrospray ionization . A binary mobile phase ( A , 15% methanol:water; B , 100% methanol:water; both containing 0 . 1% formic acid , 0 . 1% ACN and 160 mg/L NH4OAc ) was pumped at 0 . 5 mL/min through a 4 . 6 x 50 mm , 5 μm , 100 Å pore Kinetex C18 column ( Phenomenex ) . The gradient used was: 0–0 . 5 min , 0% B; 0 . 5–3 . 0 min , linear ramp to 100% B; 3 . 0–4 . 0 min , 100% B; 4 . 0–4 . 5 min , linear ramp to 0% B; 4 . 5–7 . 0 min , 0% B . MS settings were as follows: common settings were temperature = 600°C , GS1 ( ion source nebulizer gas ) = GS2 ( ion source heater gas ) = 50 lbf in-2; CUR ( curtain gas ) = 35 lbf in-2; CAD ( collision gas ) = 12 lbf in-2; IS ( ion spray voltage ) = 5500 V; analyte-specific settings for naloxonazine , naloxone and K777 , repectively , were DP ( declustering potential ) = 101 V , 76 V and 56 V; EP ( entrance potential ) = 13 . 2 V , 10 V and 10 V; CE ( collision energy ) = 47 eV , 37 eV and 57 eV; CXP ( collision cell exit potential ) = 18 V , 14 V and 18 V . The MS/MS transitions used were naloxonazine , m/z 651 . 5 → 325 . 3; naloxone , 328 . 3 → 253 . 1; K77 , 575 . 5 → 101 . 3 , and retentions were 3 . 16 , 3 . 03 and 4 . 43 min , respectively . MOR-specific siRNA was purchased from Qiagen ( Hs_OPRM1_7 FlexiTube siRNA ) . THP-1 cells were transfected with 1 μM of siRNA using the Amaxa nucleofector kit V following the manufacturer’s instructions . Control cells were mock transfected in parallel ( “mock” control ) . After nucleofection , THP-1 cells were resuspended in complete RPMI medium ( 5 . 105 cells/ml ) , treated with PMA for 24 h and infected with stationary phase L . donovani 1S2D promastigotes as described above . Parasite infectivity was assessed 48 h after infection . Down-regulation of MOR mRNA level was analysed 24 h after nucleofection and 48 h after infection by qRT PCR as described below using the following primer sets: MOR fwd: 5’ GGTACTGGGAAAACCTGCTGAAGATCT , rev: 5’ GGTCTCTAGTGTTCTGACGAATTCGAGTGG and 18S rRNA: Fwd 5' ACCGATTGGATGGTTTAGTGAG , Rev 5' CCTACGGAAACCTTGTTACGAC . The relative expression level of MOR was determined based on the Ct value normalized to the Ct value of the reference 18S rRNA . siRNA treated cells were compared to the mock transfected cells . Non-infected , PMA-activated THP-1 cells ( 5 . 105 cells/ml , 10 ml ) were treated with 10 μM of naloxonazine or 10 μM of naloxone . After 4 h of treatment , compounds were removed by 2 washes with PBS and cells were further incubated 20 h in compound-free RPMI medium . This time-point was chosen to maximize the chances of detecting naloxonazine-induced transcriptional changes while limiting the observation of downstream effects . Total RNA was extracted using TRIzol ( Invitrogen ) and amplified with the Amino Allyl MessageAmp™ II aRNA Amplification Kit ( Ambion ) following the manufacturer’s protocol . The monofunctional NHS-ester Cy3 and Cy5 dyes ( GE Healthcare Life Sciences ) were coupled with 10 μg amplified RNA . The two aRNA pools to be compared were mixed and applied to the Human Exonic Evidence Based Oligonucleotide ( HEEBO ) array ( Stanford Functional Genomics Facility ) . HEEBO oligonucleotide set consists of 44 , 544 70mer probes that were designed using a transcriptome-based annotation of exonic structure for genomic loci . Four samples ( two from naloxonazine-treated and two from naloxone-treated cells ) were competitively hybridized on two individual chips ( further called “array1” and “array2” ) . The hybridization was performed at 63°C for 16 h in a humidified slide chamber containing the labeled probe , 3X SSC , and 0 . 2% SDS . After hybridization , the hybridization chamber was removed from the 63°C water bath , washed with 0 . 6X SSC , 0 . 03% SDS , and then 0 . 06X SSC . Microarrays were scanned using a GenePix Pro Axon 4000B scanner , data were analysed with the Acuity software ( Molecular Devices ) . Fluorescent data were background adjusted and the ratios of naloxonazine-treated to naloxone-treated data were calculated for each probe set . Sets of genes showing a ratio > 2 were functionally clustered using DAVID [12 , 13] . RNA was extracted as described above , from non-infected THP-1 cells treated with 10 μM of naloxonazine for 4 h and further incubated 20 h in compound-free RPMI medium . cDNA synthesis was done with Transcriptor Reverse Transcriptase ( Roche ) and a 15-mer oligo ( dT ) primer from 1 μg of total RNA . qPCRs were run with the SensiMix SYBR no-ROX kit ( Bioline ) on a LightCycler 480 ( Roche ) . The following primer sets were used: vATPase subunit c ( ATP6V0C ) : Fwd 5' ATGTCCGAGTCCAAGAGC , Rev 5' CTACTTTGTGGAGAGGATGAG; vATPase subunit a ( TCIRG1 ) : Fwd 5’ ATCTGGCAGACTTTCTTCAG , Rev 5’ AAGATGCTGGTGGCGCGACT; B-Actin ( ACTB ) : Fwd 5' TCCCTGGAGAAGAGCTACGA , Rev 5' AGCACTGTGTTGGCGTACAG; 18S rRNA: Fwd 5' ACCGATTGGATGGTTTAGTGAG , Rev 5' CCTACGGAAACCTTGTTACGAC . The relative expression levels of vATPase and actin were determined based on the Ct value of each gene normalized to the Ct value of the reference 18S rRNA . Naloxonazine treated cells were compared to untreated cells . Total protein extracts of THP-1 cells infected with L . d . 1S2D , treated or not with 10 μM of naloxonazine for 24 or 48 h , were prepared in Laemmli sample buffer ( Bio Rad ) and analysed by SDS-PAGE and western blotting . The equivalent of 105 cells were loaded per well . Membranes were first incubated with an anti-vATPase subunit a3 ( rabbit polyclonal anti-TCIRG1 , abcam , 1:1000 ) , and an anti-rabbit HRP ( 1:5000 ) , then stripped with Restore western blot stripping buffer ( Thermofisher ) and further incubated with an anti α-tubulin ( mouse monoclonal , abcam , 1:1000 ) and an anti-mouse HRP ( 1:5000 ) . Proteins were detected by chemoluminescence following the manufacturer’s instructions ( PierceECL western blotting substrate , Thermofisher ) . Quantitative densitometry was performed using Image J . THP-1 cells infected with eGFP-expressing L . donovani were treated or not with 10 μM of naloxonazine , 10 μM of naloxone or 80 nM of concanamycin A for 24 h , or co-treated with either naloxonazine ( 10 μM ) and concanamycin A ( 80 nM ) or naloxone ( 10 μM ) and concanamycin A ( 80 nM ) for 24 h , then stained with 180 nM of the Lysotracker red DND-99 ( Life Technologies ) for 1 h at 37°C . For microscopy , cells were further stained with 500 nM of the nucleic acid stain Hoechst 33342 ( Life Technologies ) and images were acquired with an LSM 700 Zeiss confocal microscope . For flow cytometry , Lysotracker red DND-99 stained cells were first trypsinised with TrypLE Select ( Invitrogen ) , washed with PBS and analysed with a BD FACSVerse flow cytometer and the BD FACSSuite software . GraphPad Prism 5 software was used to determine the statistical significance ( Two-way ANOVA or t-test as specified in the figure legends ) . Clinical samples were from an already existing collection ( B . P . Koirala Institute of Health Sciences in Dharan ) . All samples were anonymized and their use was approved by the review boards of the Nepal Health Research Council , Kathmandu , the Institute of Tropical Medicine , Antwerp and the Antwerp University .
The activity of naloxonazine was tested in vitro against three stages of Leishmania donovani: insect-stage promastigotes , intracellular amastigotes ( within the macrophage host cell ) and host cell-free axenic amastigotes ( an amastigote-like stage obtained from differentiation of promastigotes in vitro in the absence of a host cell ) . Naloxonazine was shown to be active against the intracellular amastigote stage with a half maximal inhibitory concentration ( GI50 ) of 3 , 45 μM . It exhibited a reasonable selectivity , with a GI50 of 34 μM against the THP-1 host cell . Remarkably , the compound was inactive against L . donovani promastigotes or axenic amastigotes , indicating the importance of the host cell microenvironment for compound activity ( Fig 1A ) . Naloxonazine was also tested against two L . donovani clinical isolates , one showing susceptibility , the other showing resistance to antimonials ( SSG-S and SSG-R strains ) . The activity of naloxonazine against these isolates was comparable , suggesting that the mechanism of resistance developed against SSG does not affect naloxonazine activity ( Fig 1B ) . The necessity of the host cell presence for naloxonazine’s anti-leishmanial activity might be hypothesized to be linked to the metabolic properties of macrophages , i . e . naloxonazine could be a prodrug dependent on host cell metabolism to gain anti-leishmanial activity . In order to define the exact chemical moiety endowed with anti-leishmanial activity and to evaluate whether the macrophage would metabolize naloxonazine into an active compound , naloxonazine stability during incubation in THP-1 cell medium was evaluated by LC-MS/MS in the presence or absence of THP-1 host cells . Naloxonazine had a half-life of 15 h and was shown to be degraded into naloxone , another MOR antagonist , regardless of the presence of THP-1 macrophages ( Fig 1C ) . Interestingly , naloxone was shown to be inactive against all stages of L . donovani , including the intracellular amastigotes ( Fig 1D ) . Naloxonazine is thus not a prodrug activated by the macrophage host cell but its activity seems inherent , associated with its unperturbed chemical identity . The kinetics of naloxonazine activity showed that parasite growth was already inhibited by 70% after 24 h of compound incubation; 95% of growth inhibition was achieved after 72 h incubation ( Fig 1E ) . Remarkably , exposure of infected macrophages to naloxonazine for 4 h , followed by a washing step to remove the compound from the cells and an additional incubation of 70 h , led to the same level of parasite growth inhibition as a 72 h-incubation with the compound ( Fig 1F ) . This observation is in accordance with the degradation time of naloxonazine ( Fig 1C ) . Moreover , delaying addition of naloxonazine to 24 or 48 h after infection reduced its anti-leishmanial effect by 75% , suggesting that naloxonazine is most active at early stages of infection . We hypothesized that naloxonazine anti-leishmanial activity is dependent on its antagonistic effect towards MOR of macrophages . siRNA-mediated knock-down of MOR was therefore carried out in THP-1 cells to evaluate the importance of these receptors for parasite intracellular growth . Fifty percent down-regulation of MOR mRNA was obtained , but this was not accompanied by changes in infection levels ( Fig 2A and 2B ) . The amount or function of the MOR protein was not assessed in the siRNA-treated cells; however , phenocopy of naloxonazine’s effect on L . donovani growth could not be observed with a set of other antagonists or agonists of opioid receptors ( Fig 2C ) , supporting the conclusion that the activity of naloxonazine on Leishmania intracellular growth is independent of opioid receptors . Microarray profiling of THP-1 cells treated with naloxonazine or naloxone was performed to pinpoint host cell pathways differentially affected by the drugs and identify pathways that could be important for Leishmania intracellular growth . A 4 h-compound incubation followed by an additional compound-free incubation of 20 h was chosen , to maximize the chances of detecting naloxonazine-induced transcriptional changes while limiting the observation of downstream effects . Two percent of the probes showed at least two-fold differential gene expression in naloxonazine versus naloxone treated THP-1 cells . These upregulated genes were functionally clustered with the Database for Annotation , Visualization and Integrated Discovery ( DAVID [12] ) . Both the vacuolar H+ ATPase gene family and actin and actin-related genes clusters were perturbed , pointing to a possible effect of naloxonazine on phagolysosome formation or maturation ( Table 1 ) . The expression level of two vATPase subunits and actin was also analysed by qRT PCR in naloxonazine-treated compared to untreated cells at the same time point as the one chosen for the microarray experiment . Upregulation of these genes after naloxonazine treatment was confirmed ( Fig 3A ) . Upregulation of vATPase subunit a3 was also established at protein level after 24 and 48 h of treatment ( Fig 3B ) . To test whether naloxonazine could affect the phagolysosome , L . donovani-infected THP-1 cells treated or not with naloxonazine were stained with Lysotracker , a fluorescent acidotropic probe that accumulates in cellular compartments with low internal pH . Stained cells were analysed by confocal microscopy and flow cytometry . The intensity of the Lysotracker signal was increased after naloxonazine treatment , indicating an increased combined volume of acidic vacuoles ( Fig 3C , 3D and 3E ) . These results suggested that naloxonazine influenced the intracellular acidic compartments of the host cell . In order to determine if naloxonazine-induced changes in intracellular compartments were responsible for the effect on L . donovani intracellular growth , L . donovani-infected THP-1 cells were treated with both naloxonazine and the vATPase inhibitor concanamycin A . Remarkably , concanamycin A was sufficient to restore normal infection levels in naloxonazine-treated cells , confirming the importance of host cell-acidic compartments for controlling L . donovani intracellular growth ( Fig 4A and 4B ) . To further establish the importance of acidic compartments for Leishmania intracellular growth inhibition , we evaluated the activity of imatinib , an inhibitor of Abelson tyrosine kinase previously shown to trigger intracellular acidification in monocyte-derived macrophages [14 , 23] . In agreement with our previous observations , imatinib exhibited anti-leishmanial activity in the low micromolar range ( GI50 of 4 μM; Fig 4C ) .
We showed that naloxonazine does not directly target L . donovani but rather interferes with intracellular acidic compartments of the host cell . Upon infection , Leishmania parasites are recognized by phagocytic cells and internalized by phagocytosis in a phagosome/parasitophorus vacuole . During the process of phagocytosis , the phagosome matures through fusion with endosomes and lysosomes , ultimately leading to a highly microbicidal environment . One component of this microbicidal response is the acidification of the phagosome due to the recruitment of the vATPase proton pump to the mature phagosomal membrane [15] . Intracellular pathogens can survive these extreme conditions by arresting phagosomal maturation at an early non-microbicidal stage , developing resistance to the microbicidal arsenal of the phagolysosome , or escaping from the phagosome into the cytosol . It is well established that Leishmania amastigotes are adapted to the acidic pH found in the parasitophorus vacuole and are able to proliferate under these conditions [16 , 17] . In contrast , it has been proposed that promastigotes , the parasite stage of the insect vector , delay phagosome maturation to avoid destruction before differentiation into amastigotes [18 , 19] . Although this hypothesis has been challenged by the observation that promastigote-containing parasitophorus vacuoles do fuse with lysosomes [20] , the importance of acidic pH for controlling intracellular Leishmania growth is well recognized . Infection studies in Stat1 deficient mice for instance showed an increased Leishmania intracellular growth that was associated with an increase in phagosomal pH [21] . Our study demonstrated a naloxonazine-induced increased expression of the vATPase transporter as well as an increase of Lysotracker-positive intracellular compartments which was associated to an enhanced capacity of the host cell to control infection . Whether naloxonazine influences the pH of the parasitophorus vacuole or the amount of acidic vacuoles is unclear at this stage . Time-course analysis supported a naloxonazine anti-leishmanial effect at early stages of infection , in accordance with previous observations showing the importance of acidic compartments in the early steps of infection [18 , 19] . The microarray analysis performed in this study also showed upregulation of actin and some actin-related genes in naloxonazine-treated cells . However , further investigation is required to assess the importance of actin for the anti-leishmanial activity of naloxonazine . The molecular pathways affected by naloxonazine that lead to modification of the phagosome are yet to be determined . Naloxonazine is a MOR antagonist [22]; however , albeit such receptors are expressed by macrophages and more specifically by the THP-1 cell line , they could not be linked to Leishmania growth inhibition . Naloxone , β-funaltrexamine or CTOP , other MOR antagonists , were inactive against L . donovani and knock-down of MOR in THP-1 cells did not affect L . donovani intracellular growth . Moreover , naloxonazine is a very potent MOR antagonist ( Kd < 2 nM ) while its activity against L . donovani is in the micromolar range . These data suggest that the cellular target of naloxonazine in this case is independent of MOR . Targeting phagolysosome acidification to fight against intracellular pathogens is a strategy that has previously been validated for Mycobacterium tuberculosis , a pathogen that infects macrophages through delayed phagosome maturation . Imatinib , an Abelson tyrosine kinase inhibitor used to treat early chronic myeloid leukemia , was shown to decrease the pH of intracellular compartments which in turn reduced M . tuberculosis intracellular growth in vitro and in vivo ( 14 , 23 ) . Whether Abelson tyrosine kinases are involved in the naloxonazine-induced pH decrease of intracellular compartments deserves further investigation . Host-directed therapies ( HDT ) are considered an innovative strategy for infectious diseases , given the concern over parasites evolving resistance to current treatments , combined with the recognition of the importance of host determinants for progression of infections . HDTs are receiving increasing attention in treatment of tuberculosis for instance , and are expected to improve treatment outcomes against drug-susceptible as well as multi-drug-resistant strains [24] . Examples of HDTs under evaluation for tuberculosis treatment include anti-inflammatory compounds , statins and imatinib [25] . In antiviral therapy , HDTs have also been raised as interesting alternatives and possible solutions for limiting the emergence of drug resistance [26] . Treatment against leishmaniasis is also jeopardized by increasing treatment failure and drug resistance . Of importance , treatment failure does not necessarily correlate with parasite drug resistance ( at least against SSG or miltefosine ) , highlighting the importance of the host background for treatment outcome in this case [27] . This observation would therefore also argue in favor of HDTs against leishmaniasis . In this context , immunotherapy has received considerable interest in recent years , with the idea of modulating the immune system to achieve a protective response and parasite elimination [28 , 29] . Combination of immuno- and chemo-therapy is believed to be synergistic , allowing infection control while reducing the threat of drug resistance . Whether HDTs would be less likely to induce resistance remains an interesting and important question , and for Leishmania at least , the example of SbV raises concern . Indeed , although SbV targets host cell defense pathways , SbV-R parasites have been isolated in the Indian subcontinent , and notably these parasites showed increased infectivity [30] . This ought to raise concern of possibly generating more virulent strains should parasites become resistant to immuno-therapy . Targeting host-encoded functions unrelated to the immune response but important for parasite invasion or intracellular development provides additional options for HDTs for leishmaniasis and intracellular pathogens in general . This is exemplified by naloxonazine or imatinib and their interference with endocytic components of the host cell . Although the potential of naloxonazine itself as a therapeutic anti-leishmanial drug is unproven , targeting pathways linked to phagosome acidification remains of great interest . In addition , naloxonazine was as potent against both SSG-R and SSG-S clinical isolates , highlighting the possible benefits of such a drug target for parasites resistant to classic chemotherapy .
|
Leishmaniasis is a poverty-related disease threatening 350 million people throughout the world . It is caused by the protozoan parasite Leishmania , a digenetic organism that switches from an extracellular stage in the sand fly vector to an intracellular stage in phagocytes of the vertebrate host . Drugs currently available to treat leishmaniasis are toxic to the patient and drug-resistant parasites are emerging , urging for new therapeutics . A novel strategy to tackle intracellular pathogens entails targeting the host cell , in order to indirectly interfere with pathogens growth . Here we analysed the mechanism of action of naloxonazine , a compound previously shown to specifically affect the intracellular amastigote stage of Leishmania . We show that this compound affects acidic compartments of macrophages and that these naloxonazine-induced modifications are responsible for Leishmania intracellular growth inhibition . Even though the therapeutic potential of naloxonazine itself is not proven , our results reveal the possibility of targeting host cell intracellular acidic compartments for anti-leishmanial therapy .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"intracellular",
"pathogens",
"viral",
"transmission",
"and",
"infection",
"pathogens",
"immunology",
"microbiology",
"parasitic",
"diseases",
"protozoan",
"life",
"cycles",
"parasitic",
"protozoans",
"developmental",
"biology",
"protozoans",
"leishmania",
"white",
"blood",
"cells",
"animal",
"cells",
"life",
"cycles",
"leishmania",
"donovani",
"pathogenesis",
"amastigotes",
"host",
"cells",
"cell",
"biology",
"host-pathogen",
"interactions",
"virology",
"biology",
"and",
"life",
"sciences",
"protozoology",
"cellular",
"types",
"macrophages",
"organisms"
] |
2016
|
Naloxonazine, an Amastigote-Specific Compound, Affects Leishmania Parasites through Modulation of Host-Encoded Functions
|
PhyloGibbs , our recent Gibbs-sampling motif-finder , takes phylogeny into account in detecting binding sites for transcription factors in DNA and assigns posterior probabilities to its predictions obtained by sampling the entire configuration space . Here , in an extension called PhyloGibbs-MP , we widen the scope of the program , addressing two major problems in computational regulatory genomics . First , PhyloGibbs-MP can localise predictions to small , undetermined regions of a large input sequence , thus effectively predicting cis-regulatory modules ( CRMs ) ab initio while simultaneously predicting binding sites in those modules—tasks that are usually done by two separate programs . PhyloGibbs-MP's performance at such ab initio CRM prediction is comparable with or superior to dedicated module-prediction software that use prior knowledge of previously characterised transcription factors . Second , PhyloGibbs-MP can predict motifs that differentiate between two ( or more ) different groups of regulatory regions , that is , motifs that occur preferentially in one group over the others . While other “discriminative motif-finders” have been published in the literature , PhyloGibbs-MP's implementation has some unique features and flexibility . Benchmarks on synthetic and actual genomic data show that this algorithm is successful at enhancing predictions of differentiating sites and suppressing predictions of common sites and compares with or outperforms other discriminative motif-finders on actual genomic data . Additional enhancements include significant performance and speed improvements , the ability to use “informative priors” on known transcription factors , and the ability to output annotations in a format that can be visualised with the Generic Genome Browser . In stand-alone motif-finding , PhyloGibbs-MP remains competitive , outperforming PhyloGibbs-1 . 0 and other programs on benchmark data .
Complex , carefully orchestrated cascades of gene regulatory events control various biological phenomena , from the cell cycle to stress response to the development of an organism and differentiation of its tissues . Gene regulation can be pre-transcriptional ( such as by epigenetic silencing of genes ) , transcriptional ( controlling the recruitment of the RNA polymerase ) , or post-transcriptional ( by degrading messenger RNA before it is translated ) . Transcriptional regulation is mediated , in prokaryotes and eukaryotes , by specialised proteins called “transcription factors” ( TFs ) that bind to the DNA near a gene and recruit the RNA polymerase ( or inhibit its recruitment ) . This regulatory control is often combinatorial , with many TFs controlling a gene , and highly complex . A gene that encodes a TF , when turned on , may cause many more genes to be turned on . To understand gene regulation , therefore , it is important to identify potentially regulatory DNA and to understand how and where individual TFs may bind there . Typically TFs recognise short patterns or “motifs” in DNA that they bind to . For this reason , “motif-finding” , or detection of short patterns that are over-represented in a generic “background” , is an important computational problem in studying gene regulation . These motifs are generally not exact strings , but indicate weaker site-specific nucleotide preferences . Though several highly efficient substring-finding algorithms exist in computer science , they are of limited utility here . Instead , a common approach is to assume that the background is either random , or contains short-ranged correlations that can be described by a Markov model , while binding sites for transcription factors can be represented as samples from “position weight matrices” ( PWMs ) [1] . For a motif of length ℓ , a PWM is a 4×ℓ matrix giving the probability , at each position , of seeing each of the four bases ( A , C , G , T ) at that position . Two standard ways of detecting conserved regulatory sites amidst “background sequence” are Gibbs sampling , first described in this context by Lawrence et al . [2] , and expectation maximisation over mixture models , implemented in the MEME algorithm of Bailey and Elkan [3] . Recently we presented a new implementation of the Gibbs sampler , PhyloGibbs [4] . The primary goal was to deal systematically with the case of orthologous sequence from closely-related species , where naive scoring of overrepresentation will fail because much sequence has not diverged sufficiently . This is handled using a user-specified phylogenetic tree and a modified scoring scheme for phylogenetically related sequence . Additionally , PhyloGibbs evaluates its own site predictions via statistical sampling of the entire state space , so that it can report the posterior probability , given all prior assumptions , that a given site is indeed a binding site . We showed that this self-assessment is an improvement on previous programs , which either do not assess their predictions , or use other statistical significance measures that do not evaluate to posterior probabilities . A related point is that , in addition to individual site predictions , PhyloGibbs outputs weight matrices that are constructed from the predicted binding sites weighted by their significance , and are not merely counts of nucleotides . A problem with experimentally determined weight matrices is that they are often constructed from a small number of annotated binding sites , all of which are weighted equally , even though they may not all be of equal affinity in practice . Indeed , not all experimentally-determined sites may be known with equal confidence: binding assays often localise a much larger region of DNA , within which the putative binding sites are found bioinformatically . PhyloGibbs can be used to construct weight matrices from such data , weighted by confidence , as discussed below ( Materials and Methods , “CRM Prediction” ) , a point that was not fully explored in the previous paper . Here we present PhyloGibbs-MP , an extension of PhyloGibbs in several directions that go well beyond standard motif-finding . MP stands at the moment for “module prediction” ( and possibly also “multiprocessor”: it has preliminary support for shared-memory multiprocessor systems , using OpenMP , and future support for distributed-memory clusters , via MPI , is planned ) . In the Results section , we benchmark PhyloGibbs-MP in motif-finding , module prediction and discriminative motif-finding . In the Methods section , we describe the implementation of these features . In addition , several smaller changes in the algorithm have been made , discussion of which occurs towards the end of the Methods section . Also , many command-line options are no longer compatible with the earlier program . To avoid confusion , we have renamed the program “PhyloGibbs-MP” . In this paper , “PhyloGibbs-1 . 0” and “PhyloGibbs-MP” are used for statements specific to those versions of the program , and “PhyloGibbs” is used for remarks common to both programs .
Before discussing new features , we first test the straightforward motif-finding capability of PhyloGibbs-MP , on test datasets of known binding sites in yeast ( Saccharomyces cerevisiae ) and fruitfly ( Drosophila melanogaster ) . This is essentially a repeat of tests reported for PhyloGibbs-1 . 0 [4] , using the SCPD database [11] of experimentally-determined transcription-factor binding sites in S . cerevisiae . A filtered list of these binding sites , rejecting very large and very small sites , was used; this contains 466 binding sites upstream of 200 genes . The advantage here is that every site in this database is experimentally validated; it thus provides a very good measure of real-world performance of various algorithms . The disadvantage is that there may be many sites that are not known . We previously argued [4] that we expect roughly one in three sites to be known ( and present in this database ) , and showed that PhyloGibbs' self-assessment of its predictions is consistent with this expectation . For each of these 200 genes , we select up to 1000 bp upstream sequence ( not overlapping coding sequence ) from S . cerevisiae , orthologous sequence from S . paradoxus [12] , S . mikatae , S . kudriavzveii , S . bayanus [13] , and run various motif-finders on them . The orthologous sequences were determined ab initio using BLAST and synteny as criteria . The motif-finders tested were AlignACE [14] , [15] , MEME [3] , PhyME [16] , EMnEM [17] , and the Gibbs sampler from the Wadsworth Institute [2] . Other than AlignACE , the other programs had been previously tested against PhyloGibbs-1 . 0 [4] . Here , however , we use an updated Wadsworth Gibbs sampler , which has recently acquired [18] the ability to do “phylogenetic” sampling . We tested this program both in non-phylogenetic mode and in the phylogenetic mode . The results are shown in Figure 1 , in the form of specificity ( fraction of binding sites predicted that are known ) as a function of sensitivity ( fraction of known binding sites that are predicted ) . The sensitivity and specificity are varied by choosing different “cutoff scores” for the significance scores assigned by various programs; only sites with a significance above the “cutoff score” are considered . The higher the cutoff , typically , the lower the sensitivity but the higher the specificity . In the case of AlignAce , which does not assign significance scores to individual site predictions , we used the different predicted motifs as cutoffs . ( That is , the different points on the sensitivity/specificity curve correspond to the sensitivity/specificity calculated from all sites predicted in the first n motifs , from n = 1 onwards . ) PhyloGibbs-1 . 0 , PhyloGibbs-MP and EMnEM used multi-fasta sequences aligned with Sigma [19] version 1 . 1 . 3 , PhyME used sequences aligned with a bundled version of Lagan [20] , and the phylogenetic Gibbs sampler used sequences aligned with ClustalW [21] . All other programs used unaligned sequences . PhyloGibbs-1 . 0 , MEME , PhyME and EMnEM perform similarly to previously reported . AlignACE performs rather poorly on this dataset . This is probably a result of the lack of site-specific significance information in its output . The Wadsworth Gibbs sampler , run in the normal ( non-phylogenetic , non-centroid ) mode shows a much improved performance from the version we previously reported . However , when run in phylogenetic mode , the Wadsworth Gibbs sampler makes very few predictions indeed . The Wadsworth Gibbs sampler in phylogenetic mode was run with a commandline suggested by W . Thompson ( personal communication ) . Commandlines in other cases were chosen based on available documentation . Details are in Materials and Methods . PhyloGibbs-MP is run in two modes: searching for a maximum of 3 or a maximum of 8 simultaneous motifs; and in the latter case , with or without “importance sampling” . All choices show good performance but performance is clearly superior when searching for 8 motifs , with or without importance sampling . This is in contrast to most other programs ( not shown ) , including PhyloGibbs-1 . 0 , where searching for too many simultaneous motifs hurts performance . Importance sampling ( the default ) gives a speed increase of a factor of about 10 , and these data show that the effect on the quality of predictions is minor . This is further discussed in “Materials and Methods” ( subsection “Importance sampling” ) . PhyloGibbs-MP performs well when searching for multiple simultaneous motifs ( “colours” ) because it allows each colour to contain only as many windows as actually belong ( that is , it enables toleration of overestimates of the number of binding sites ) . For example , suppose one assumes that there are 3 regulatory motifs , and 1% of a 1000 bp sequence is functional: that would yield 10 sites overall , or 3 to 4 sites per motif . In fact , however , there may be only two sites per motif . Providing more allowed colours lets the “good” motifs be grouped together , and irrelevant motifs are placed into other colours ( and , since they are not selected often , do not accumulate high tracking scores ) . To some extent this applied to PhyloGibbs-1 . 0 too; but PhyloGibbs-1 . 0 insisted , for technical reasons , on having at least one selected site for every colour , which hurt performance when the number of colours was large . We used the REDfly 2 . 0 [22] transcription factor binding site database . Since many of these reported binding sequences are much longer than the expected length of an individual binding site , we chose a subset for which we could bioinformatically determine the likely binding site with reasonable confidence using independent data , as described in Materials and Methods . Moreover , we wanted to include PhyloCon [23] , a motif-finder that makes somewhat different assumptions about the nature of input sequences ( in particular , it expects multiple sets of orthologous sequences , with at least one binding site in each set ) . Therefore , we chose only factors for which , after our processing , multiple binding sequences were available . Other than the addition of PhyloCon , all programs used in the yeast benchmark were run on this data , except PhyME , which we could not successfully run on this data ( it crashed ) , and Phylogenetic Gibbs , which showed poor performance on the yeast data even after its commandline parameters were extensively adjusted on the advice of one of its authors . Also , in the PhyloGibbs family , only PhyloGibbs-MP with eight input sites was run , since this showed the best results in the yeast benchmark . In addition to D . melanogaster sequence , orthologous sequence was used from the recently sequenced genomes of D . yakuba , D . erecta , and D . simulans [24] . ( Though many papers have used D . pseudoobscura , the second fly genome to be sequenced , we rejected it because its distance from melanogaster suggests that gene evolution may have evolved significantly . ) The results are shown in Figure 2 . They are largely similar to the yeast results , but overall the sensitivity is poorer than in the yeast data: that is , all programs perform with poor specificity for a sensitivity greater than about 0 . 12 , and their performance is probably no better than random at this level . If one focuses on high-quality predictions ( which means poor sensitivity ) , PhyloGibbs-MP sharply outperforms all other programs . PhyloCon performs particularly poorly: a detailed look at the output shows that it predicts sites only for two factors , ftz and bcd , and most of these predictions don't correspond with the annotated sites . It should be noted that binding motifs in fly ( and other higher eukaryotes ) are much fuzzier and less specific than in yeast , and ( especially with the homeotic factors ) tend to be rich in A's and T's; similar motifs may well occur by chance , to lead the motif-finders astray . Moreover , all the input sequences probably contained several binding sites for several factors other than the ones annotated by REDfly . Many of these unknown binding sites may have been successfully detected but not measured in the benchmark . Nonetheless , PhyloGibbs-MP performs clearly better at the high-confidence end , suggesting that it is indeed better at distinguishing conserved binding sites from background . Indeed , the top few sites predicted by PhyloGibbs-MP are all known sites ( specificity 1 . 0 ) and account for about 2% of the total number of sites in this benchmark . Discriminative motif-finding has been discussed in the literature previously . We choose three previously published programs , ALSE [25] , Dips [26] , and DEME [27] , which are available for downloading and running locally , to benchmark against PhyloGibbs-MP . Other published programs , not available for download , include LearnPSSM [28] , DME [29] , and dPattern [30] . Unlike previous programs , PhyloGibbs-MP handles multiple input data sets , and treats them symmetrically: rather than requiring a “positive” and a “negative” set , it seeks sites in any one “discriminative group” that are over-represented in that group but under-represented in the other groups . ( A set of genes A may be up-regulated relative to another set B , not only because genes in A are being activated by a common factor , but because genes in B are being repressed by a common factor . Moreover , one is often interested in the regulation of multiple sets of genes , that are differently expressed over a set of conditions , even if no one set of genes is preferentially expressed overall . ) Moreover , the degree of differential discrimination may be controlled by a command-line parameter ( -d ) . If this parameter is provided but is zero , sites are found in a single group without regard to how much they may be represented in other groups . Very high values of the parameter strongly repress prediction of sites that have counterparts in other groups ( this is similar to how other discriminative motif-finders work ) . In addition , PhyloGibbs-MP takes systematic account of phylogenetic relationship between species . We performed four tests: on synthetic data; on yeast data using the same SCPD database as in the previous section; on fruitfly data using REDfly binding site data , again as in the previous section; and on groups of putative co-regulated yeast genes obtained from genome-wide binding data from Harbison et al . [31] . Results from these tests suggest that making the discrimination too aggressive is counterproductive , which may account for why discriminative motif-finders have not achieved as much popularity as conventional motif-finders . On actual genomic data , the performance of PhyloGibbs-MP compares well with , or exceeds , the performance of DEME and ALSE . For synthetic data , we generated two sets of phylogenetically-related sequence as follows: first , three random motifs A , B , C of 10 bp each were selected , drawn position weight matrices where the consensus base had 85% weight , with the remainder uniformly distributed . Two “ancestral” sequences were generated , each containing five copies of motif A; one of these also contained five copies of motif B , while the other had five copies of motif C . These ancestral sequences were then evolved , according to our evolutionary model with an expected mutation rate of 0 . 5 per nucleotide , to five descendants; mutated background bases are re-sampled from the background model , and mutated bases in binding sites are re-sampled from the PWM . Thus we have two sets of five phylogenetically related sequences , each containing 5 copies of a “common” motif and 5 copies of a “discriminative” motif . 200 such pairs of sequence sets were generated , and PhyloGibbs-MP , ALSE , DEME and Dips were benchmarked on them , for sensitivity and specificity in detecting the common motifs and the discriminative motifs . The results , with different choices of the discriminative parameter for PhyloGibbs-MP , are shown in Figures 3 and 4 . When we are not seeking discriminatively occurring motifs , PhyloGibbs-MP finds the common motifs ( which are abundant ) with high sensitivity and high specificity , while the discriminative motifs are found with poor sensitivity and specificity . As we turn up the discriminative parameter , the significance of common motifs is reduced , and discriminative motifs are found more significantly . At “ -d 0 . 4” , discriminative motifs are found with moderate sensitivity but very high specificity; common motifs are significantly suppressed . Further increasing the parameter yields smaller gains . DEME picks up hardly any common motifs , and ALSE also picks up very few . Dips picks up common motifs , but with a specificity of about 0 . 1 ( independent of sensitivity ) that indicates random performance . ( Each 1000 bp sequence contains 5 embedded common motifs each 10 bp long , but overlaps of up to 5 bp in predictions would be considered “hits”; therefore about 10% of each sequence would be “hit” randomly . ) On the other hand , ALSE and DEME both perform very well in picking up differential motifs; PhyloGibbs-MP , with differential parameters -d 0 . 4 and -d 0 . 99 performs reasonably well , outperforming Dips . With yeast data , we selected pairs of genes for which differing regulatory factors are listed in SCPD , and picked 1000 bp upstream sequence ( excluding overlapping coding sequence ) with orthologous sequence from other sensu stricto species , as for the motif-finding benchmark . Having generated 571 such pairs , we ran PhyloGibbs-MP ( with discriminative setting 0 . 4 ) , ALSE , DEME and Dips on each pair and measured their performances . ( PhyloGibbs-MP treats the members of a pair symmetrically; the other programs were run twice on each pair , alternately choosing one member as the “positive” set and the other as the “negative” set . ) The results are in Figure 5 . DEME is the best performing program on this set , except for high-significance ( low sensitivity ) predictions , where PhyloGibbs-MP is competitive with it . In interpreting this data ( and the data in the following fruitfly benchmark ) , one should note that , first , there could well be unknown common factors regulating many of these pairs of genes , and second , different factors may nevertheless bind to somewhat similar binding sites ( since proteins have a limited number of DNA-binding domains ) . The second point is even more important in the fruitfly case . We used the same sequences and binding-site data as in the fruitfly motif-finding benchmark , but also included factors for which only one sequence was available . Similarly to the discriminative yeast benchmark , we chose pairs of sequence sets that contained binding sites for different known factors . It should be emphasised that it is very likely—even more so than in the yeast case—that these pairs of sequence sets contained unknown common binding sites , and also that many different factors in this case contain rather similar motifs . The results are in Figure 6 . In this case , the gap between PhyloGibbs-MP and ALSE is quite low; DEME , surprisingly , performs significantly worse . Finally , as a test on realistic data of the sort where a discriminative motif-finder would be useful , we considered DNA-binding data from genome-wide location microarray experiments ( “ChIP-chip” ) reported by Harbison et al . [31] in S . cerevisiae . We chose 15 factors with “known” motifs ( as reported by them ) that bind between 4 and 9 probes , under at least two conditions including rich medium , with a p-value below 0 . 001; reported binding sequences ( and orthologues ) were in the “positive” set and an equal number of randomly chosen non-binding sequences were in the “negative” set . The factors thus chosen were AFT2 ( YPL202C ) , BAS1 ( YKR099W ) , CBF1 ( YJR060W ) , DAL81 ( YIR023W ) , GAL4 ( YPL248C ) , HAP2 ( YGL237C ) , MET4 ( YNL103W ) , MSN4 ( YKL062W ) , PUT3 ( YKL015W ) , RCS1 ( YGL071W ) , RDS1 ( YCR106W ) , ROX1 ( YPR065W ) , RTG3 ( YBL103C ) , STP1 ( YDR463W ) , YAP1 ( YML007W ) . Details are in Materials and Methods . Figure 7 summarises the results . Unlike other benchmarks , the comparison here is qualitative , and at the motif-level not the site-prediction level . On the whole , most predicted motifs bear little resemblance to the “known” motifs . The various programs perform as follows: This is a qualitative comparison and many of the above comparisons are rather nebulous . In particular , if one omits poly-A or poly-T predictions ( which are plentiful in the yeast genome ) , the number of “predictions” for DEME and Dips falls sharply . Though DEME makes only eight predictions that match ( or somewhat match ) the known motifs in these data , its authors report benchmarks that predict 13 of 15 motifs in sequences drawn from the same ChIP-chip data [27] . However , both their selection of sequences ( which is not described in detail ) and their methodology differ from ours . In particular , where we assume no prior knowledge of the motif width and use a width of 10 in all cases , they use the actual width of the motif as prior information . In realistic situations , one is unlikely to know the width of an uncharacterised transcription factor's binding site . When one takes the four benchmarks together—one on synthetic data and three on actual genomic data of the type likely to arise in realistic research situations—it is clear that , in this particular problem , synthetic data captures very poorly the complexities of an actual situation . There are various issues at work in actual genomic sequence: motifs for different transcription factors may resemble each other strongly , especially if they come from the same family; weak , non-specific motifs may have close matches by “chance”; there could be many relevant factors regulating a gene or set of genes , only some of which are discriminative; and , in data arising from high-throughput experiments , there could be “noise” in that not all sequences reported to bind a protein may actually do so . Over-aggressiveness in discriminative motif-finding leads to excellent synthetic-data performance but poor sensitivity and/or specificity with real genomic data . All the programs tested suffer from this issue , but PhyloGibbs-MP mitigates the problem with the tunability of its discriminative parameter . The benchmarks do indicate that , over a somewhat broad range of data , DEME and ALSE are excellent discriminative motif-finders , performing far better than PhyloGibbs-MP on synthetic data and very well on real data . However , with real data , PhyloGibbs-MP is competitive with or superior to both those programs ( in particular , it appears markedly superior on the ChIP-chip benchmark ) , while including the flexibility of a general-purpose motif-finder . An example of PhyloGibbs-MP's ability to predict cis-regulatory modules is shown in Figure 8 , which depicts the region upstream of the eve gene . Without prior information , PhyloGibbs-MP successfully predicts all four annotated upstream CRMs from the REDfly database: the proximal promoter , the stripe 2 enhancer , the stripes 3+7 enhancer , and the mas enhancer . ( In the case of mas , the predictions are not exactly overlapping the annotated enhancer , but are nearby and over a broader region . In the case of stripe 2 , the annotated 500 bp enhancer in REDfly is somewhat shorter than other reports ( e . g . , Ludwig et al . [32] ) that suggest an enhancer close to 800 bp in length; most of the predicted sites fall in this larger region . With prior information in the form of nine weight matrices corresponding to gap factors , PhyloGibbs-MP fails to pick up the proximal promoter , but finds the remaining CRMs with greater confidence than before . For a more thorough and quantitative benchmark , we used the REDfly database [22] , [33] of cis-regulatory modules in D . melanogaster . We filtered the REDfly CRM list for CRMs of suitable length ( <3000 bp ) , fused nearby CRMs , and selected sufficient neighbouring sequence ( details are in Materials and Methods ) that we were left with 234 stretches of DNA that were at least 10000 bp long and contained at least one annotated CRM . PhyloGibbs-MP , and four other downloadable programs—Stubb [8] , [9] , Cluster-Buster [7] , EMCMODULE [34] , and CisModule [35]—were run on these segments . ( Details and exact command-line parameters are in Materials and Methods . ) Cis-Analyst [36] , [37] , and other programs that are not downloadable or can only be run via a web-server , were not tested . For priors we used a set of 73 weight matrices , that we created from DNAse I footprints in the FlyReg database [38] and orthologous sequence in five other species ( details of how these matrices were generated are in Materials and Methods , and the matrices themselves are available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ . As in the motif-finding case , sensitivity was varied by varying the significance cut-off of individual site predictions ( or , in the case of Stubb , of individual free-energy “windows” of 100 bp each ) . The sensitivity of the output was measured , for predictions above various cutoff thresholds , by what fraction of CRMs ( weighted by length ) , of the total known , were successfully predicted by the programs . A predicted site that lay within the boundaries of a module was counted as a “prediction” of that module . The specificity was measured by what fraction of site predictions , for each program , occurred within known CRMs . To test the effect of prior information and orthologous sequence availablity on the performance of PhyloGibbs-MP , it was run with and without priors , and with one ( D . melanogaster only ) , two D . melanogaster and D . yakuba ) or four ( D . melanogaster , D . yakuba , D . simulans and D . erecta ) aligned species . The results are in Figure 9 . The best performance came when prior information was supplied , but ( somewhat surprisingly ) when only two input species , not four , were used . This suggests that spurious conservation across multiple species may lead PhyloGibbs-MP astray . Next , PhyloGibbs-MP ( best-performing parameters ) was compared with the output of the other programs . The results are in Figure 10 . In terms of speed , PhyloGibbs-MP is much slower than the other programs: being a Gibbs-sampling ab initio motif-finder , it runs roughly quadratically in sequence length ( given typical parameters ) , whereas a search and clustering for motifs can be done in linear time . Its utility becomes apparent when used as a combined CRM predictor and ab initio motif-finder , and it performs competitively with dedicated programs at both these tasks . In recent work , we have used a subset of the above FlyReg-derived matrices , and a few additional literature-derived weight matrices , to make detailed studies of a myoblast-related enhancer that exhibits a complex modular expression pattern ( K . G . Guruharsha et al . , in preparation ) .
PhyloGibbs-MP is available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/ . It is open source , licensed under the GNU General Public License . Supporting data is available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ .
The PhyloGibbs algorithm was described in detail earlier [4] , so a brief summary will suffice here . ( Some changes in PhyloGibbs-MP from PhyloGibbs-1 . 0 are described in Materials and Methods , “Changes to Algorithm” . ) PhyloGibbs models “generic” non-coding DNA sequence by a Markov model of order k ( where typically k is 1 , 2 , or 3 ) whose parameters are estimated ( preferably ) from an auxiliary file of background sequence , or ( less reliably ) from the input sequence itself . It assumes that some locations in the input sequence are binding sites for transcription factors , and are not described by the background model but by “position weight matrices” ( PWMs ) : matrices of order 4×ℓ that give the probabilities of seeing each of the four nucleotides at positions 1 through ℓ in the site . All binding sites belonging to a common transcription factor are given by the same ( often unknown ) PWM . A “parse” of the sequence consists of a selection of particular sites as putative binding sites . For each such parse , C the likelihood of seeing the given sequence P ( S|C ) can be calculated ( as described in [4] ) , and then the posterior probability of C follows by Bayes' theorem: ( 1 ) Here , P ( S ) ( loosely , the “prior probability” of the sequence S ) is a constant , equal to ΣCP ( S|C′ ) P ( C′ ) , while P ( C ) is the prior probability of the parse , which is chosen to incorporate as much prior information as possible . For example , it can be chosen to be constant for a given number of colours and a given number of sites , and zero otherwise . One can also use a “chemical potential” to allow flexibility in the number of allowed sites . PhyloGibbs-MP fixes a maximum number of colours , and an expected number of binding sites ( in contrast , PhyloGibbs-1 . 0 fixes the number of windows , that could encompass many sites ) . PhyloGibbs-MP also adjusts P ( C ) when dealing with informative priors , module prediction , and discriminative motif-finding . PhyloGibbs uses a moveset that preserves detailed balance ( some caveats apply to PhyloGibbs-MP: see the appendix ) . It samples the space of parses , first finding the parse with maximum a posteriori probability P ( C|S ) , then evaluating the significance of each predicted site by further sampling . It can simultaneously detect binding sites for multiple different TFs by labelling each with a different “colour”; the posterior probability is a product over all colours . In addition , PhyloGibbs can deal with phylogenetically related sequence that has been pre-processed by a multiple sequence alignment program . It does this by treating sites in an aligned block not as independent , but as descendants of a common ancestor , and modifying the scoring appropriately ( whether as “binding sites” or as “background” ) . The scoring is governed by the transition probabilitywhere α is the ( unknown ) ancestral base , a is the descendant base , q is the rate of conservation ( “proximity” ) between the ancestor and the descendant , and wa is the probability of seeing a at that position in the descendant under the assumption that this is a binding site for a weight matrix ( or a background site , as the case may be ) . In other words , the ancestral site is conserved with probability q and mutated with probability 1−q; if mutated , it has undergone fixation that preserves the functionality , or the background statistics . This transition probability is transitive and has the correct limits at extreme q . The above applies directly to “star phylogenies” ( where all species are independently descended from a common ancestor ) , but arbitrary phylogenetic trees are handled by converting them into sums of products of “subtrees” that individually have “star phylogenies” . ( This is exact , but approximations are made in dealing with the subtrees . ) Internally , PhyloGibbs represents such related sites by “windows” , aligned blocks of sequence that are either all functional , or all non-functional . Details , again , are in the earlier PhyloGibbs paper [4] and are unchanged in PhyloGibbs-MP . PhyloGibbs-MP takes as optional parameters the maximum length l of a cis-regulatory module , and the average spacing d between two CRMs . Then , on each input sequence , it requires that not more than max ( 1 , ( L+d ) / ( l+d ) modules exist , where L is the length of the sequence . Each site , or multispecies “window” , that is selected now must satisfy existing module constraints on all sequences that it is a part of . In other words , only windows that can satisfy the current constraints are sampled for . For example , if two modules are allowed on a sequence , but all windows that currently occupy that sequence fit within the width of one module ( any two windows are within a distance l of each other ) , a new window can be sampled anywhere on that sequence ( provided it satisfies constraints on other sequences that it is a part of ) , and will define a new module . However , if the currently selected windows must be spread across two modules , newly selected windows must fit within the constraints of those two modules—that is , they must be no more than l nucleotides away from any window in their module . Thus , the allowable window placements , and the modules defined by the selected windows , are dynamically updated and need not stay localised as sampling proceeds . At the end of the run , the tracking scores , visualisable via the GBrowse annotation ( see Figure 8 ) for an example ) , reveal the positions of the predicted modules , which may be sharply localised or may be spread all over the sequence . When given sets of regulatory regions for genes that are believed to be regulated differently , it is of interest to find motifs that occur preferentially in one set rather than in the other . When run in discriminative mode , PhyloGibbs-MP accomplishes this by , for each “colour” , selecting sites only in one regulatory set . However , it also selects “mirror” sites in other groups . The total number of mirror sites , across all groups and all types of motifs ( “colours” ) , is the same as the total number of actual binding sites expected . The mirror windows are sampled for in the same way as the “real” windows , by scoring them together with the real windows . Thus , if a colour has n real windows and m mirror windows selected , a new window is sampled with the posterior probability Pn , m of being drawn from the same weight matrix as these n+m sites . It is possible for a colour to contain no mirror windows . If a colour gets emptied of real windows , the mirror windows are also emptied and re-sampled into other colours . These “posterior probabilities” are calculated as described in our earlier paper [4] . However , it is convenient below to use the language of thermodynamics . By analogy with the Boltzmann probability of finding a thermodynamic system in a state of free energy F , which is equal to exp ( −βF ) where β is the inverse temperature , we define a “free energy” corresponding to a posterior probability P . In our case the temperature is fictitious: β = 1 during the initial equilibriation and tracking phases , and is slowly increased during the simulated anneal . In the usual ( non-discriminative ) Gibbs sampler , we start from a state with n+1 windows , remove a window ( to leave n windows selected ) , and sample replacements according to the “free energies” Fn+1 , which correspond to the posterior probabilities that the n selected windows plus the one new window are all sampled from the same position weight matrix . ( More accurately , the new window may be placed into any existing “colour” , not necessarily the one from which a window was removed . Below we write a “colour index” explicitly for clarity . ) For the discriminative motif-finder , in any given configuration , each colour has “real” windows as well as “mirror” windows selected . A move starts with removing a real window , and a mirror move by removing a mirror window; these are then resampled . Let's say , after the removal , one is left with n real windows and m mirror windows . The “mirror” sites are sampled as above , but treating real and mirror sites as the “same” . That is , they are sampled using the free energies Fn+m+1 , c , which correspond to the posterior probability of all n real windows , all m mirror windows , and the newly selected mirror window all being sampled from the same PWM ( that corresponds to colour c ) . The goal here is to maintain a mirror list that is as faithful as possible to the real list . The “real” sites are instead sampled according to ( 2 ) where c stands for the colour ( motif type ) into which the window is being sampled Fn+1 has the same meaning as in the non-discriminative case , Fm is the free energy for the case that the m mirror windows are sampled from a common PWM ( not necessarily the same as the “real” windows ) , and Fn+m+1 is the free energy for all n selected windows , all m mirror windows and the one newly selected window are all sampled from the same PWM . A window may be sampled into any colour , and each colour has a different set of mirror windows associated with it ( thus n and m both depend on c ) ; so this free energy is calculated for each candidate window and for each colour into wich it may be sampled . The bracketed term has the effect of penalising cases where the mirror windows strongly resemble the real windows . Here , α is a parameter , greater than 0 , that determines how strongly to penalise motifs that are well represented outside the current discriminative group . For very small α , sites for a given colour are selected only from one discriminative group but little consideration is made to whether similar sites occur elsewhere . For larger α , similar sites in other discriminative groups will penalise the score more severely . Typically we find that α = 0 . 4 suffices to predict genuinely discriminative motifs , while excessively large values of α will cause chance occurrences of the motif in other groups to excessively penalise the “good” motifs—which is also noticed in the benchmarks for discriminative motif-finders ( cf . section “Results” , subsection “Discriminative motif-finding” ) . Optionally , PhyloGibbs-MP can take a file containing prior position weight matrices corresponding to possibly relevant transcription factors , and bias its search to sites corresponding to those prior PWMs . This is done as follows: Input sequence is pre-parsed in PhyloGibbs-MP ( as in PhyloGibbs-1 . 0 ) , into “windows” of specified length . When prior PWMs are given , each prior PWM is associated with a unique “colour” . Then each window is given a prior probability of being a binding site for each of the given prior PWMs , as well as of being background . This probability is calculated as follows: Let Pbg be the probability that the window is background ( estimated from a background model ) , and Pδ ( W ) be the probability that the window contains a binding site for PWM W with offset δ from the start of the window . Then the probability that the window is a binding site for W is given by ( 3 ) where only offsets where at least 50% of the PWM is in the window , or at least 50% of the window is covered by the PWM , are considered in the sums over δ . ( PWMs that are too large or too small to satisfy these criteria are filtered out . ) Then , during the window-shift moves , when a new window and colour are being sampled , the posterior probability calculated for the new configuration is multiplied by the prior probability of that window being in that colour . Optionally , PhyloGibbs-MP can output annotation files for its predictions that are readable by the Generic Genome Browser [39] as well as by a visualisation tool that we are developing and have used to generate Figure 8 in this paper ( S Acharya and RS , in preparation ) . Annotations are for one species only , the “anchor species” that must be the first specified in the phylogenetic tree . The headers for sequences from that species must be formatted appropriately; details are in the PhyloGibbs-MP manual , distributed with the software . While PhyloGibbs-MP can be used as a conventional motif-finder , for the most part in the same way that PhyloGibbs-1 . 0 can , several changes have been made to the details of the algorithm , with an aim at improving performance . PhyloGibbs-1 . 0 strictly maintained detailed balance in all movesets; PhyloGibbs-MP is not quite so rigid . We discuss the deviations below . While we would prefer to maintain detailed balance strictly , we note that detailed balance , combined with ergodicity , ensures a sampling of state space with the appropriate posterior probability distribution only in the infinite time limit . But a useful program must run in limited time , and therefore good convergence of the moveset is equally important in practice . ( For example , in PhyloGibbs-1 . 0 , the “colour-change move” [4] is by itself ergodic and satisfies detailed balance , and therefore should suffice in the infinite time limit . But in any realistic running time it does nothing useful by itself . Similarly , in the infinite-time limit the “global shift moves” would not be necessary . ) We argue that our breaking of detailed balance is sufficiently rare to be harmless , and sufficiently useful to be justifiable . Rare detailed-balance-breaking moves correspond to having several Markov chains through state space , where links within a chain satisfy detailed balance , but the moves connecting different chains do not . If the number of links within a chain is much greater than the number of chains , ( in particular , if each chain is allowed to grow infinitely long while the number of different chains remains finite ) , the desired posterior probability distribution will be reached ( since it is reached separately by each of the chains ) . In PhyloGibbs-1 . 0 , these moves maintained a constant number of windows ( which may contain multiple orthologous sites ) , by replacing one window with another . In PhyloGibbs-MP , the window-shift move maintains a constant number of sites . This is done as follows: First , an initial input parameter is p ( by default 0 . 01 ) , which is the expected “density” of sites in the input sequence . For example , if the input sequences have lengths Li , the total expected number of sites in these sequences is NE = Σip ( Li−w+1 ) where w is the width of a window . Each window-shift move removes an existing coloured window ( randomly chosen , of any colour ) , and replaces it with a new coloured window . However , the weight ( posterior probability ) of the move is multiplied by the heuristic exp ( −β ( ( Ns−NE ) /Nm ) 8 ) . Here , Ns is the number of sites that would be selected if that window were picked , and Nm is the maximum number of sequences in any window in the set , and serves as a “margin” for the amount that Ns can deviate from NE . The inverse temperature β is unity except during the simulated anneal , where it is increased gradually . We use the eighth power to allow deviation with little penalty within the margin Nm , but rapidly growing penalties for any larger deviation . This satisfies detailed balance . However , we make two exceptions: if , before the move , Ns ( the number of selected sites ) is smaller than NE−Nm , we do not remove any window , but directly pick a new window . And if , after removing a window , Ns is larger than NE+Nm , we do not pick a new window . These exceptions occur sufficiently rarely that the breaking of detailed balance is not serious . The ability to specify a “density of sites” , independent of the input sequence length or the degree of homology , is a large advantage , which contributes to PhyloGibbs-MP's superior performance over PhyloGibbs-1 . 0 . Another breakage in detailed balance occurs when module prediction is enabled: adding or removing windows may change module boundaries , with the result that the set of available states is different before and after the move . Again , this is relatively rare and in a useful cause . In this move , an attempt is made to shift all windows of a given colour by a fixed distance , left on one strand or right on the other . This is to move out of “local minima” where the sampler has found a non-optimal solution that is offset by a fixed amount , and moving one window at a time would take a long time to happen . To maintain detailed balance , PhyloGibbs-1 . 0 sampled all possible shifts , up until such distance as no shift was possible ( for instance , because it was blocked by other windows , or because it would run off the edge of the sequence ) . While this does maintain detailed balance , it fails to sample some legitimate shifts: in particular , if a window was at the edge of an aligned block , PhyloGibbs-1 . 0 could not shift it beyond , because the number of sequences in the window would change , and therefore would not shift any window in that colour . Instead , PhyloGibbs-MP samples shifts of only one space left or right , and allows shifts beyond window boundaries ( such windows may either “gain” sequences from other species , or be broken into smaller windows with fewer sequences ) . In this case , general balance is significantly broken for a given global shift move , because the set of available states is not the same before and after the move . ( A window at position i could be shifted , with other windows of that colour , to i+1 or i−1 or could stay at i; if it shifts to i+1 , the available states are now i , i+1 , i+2 . Also , in case of blocked windows that are “thrown away” after the shift , new windows are resampled , and again the set of possible states is not the same . ) Again , the rareness of the moves compared to the window-shift moves , and their utility in practice , justifies the breakage . The colour-change move in PhyloGibbs-1 . 0 has been removed . Instead , a maximum number Nc of colours is specified , but ( unlike in PhyloGibbs-1 . 0 ) fewer than Nc colours may actually be selected at any time . PhyloGibbs-1 . 0 had an optional “maskbit flip” move , where certain columns are optionally not scored; these columns are sampled for using Metropolis moves . In practice , however , motifs tend to be strong at the centre and weak at the edge , except for symmetric dimer motifs ( mainly in bacteria ) , where they can be weak in the middle . Therefore PhyloGibbs-MP allows such unscored columns only at the edge of the window ( and , in the case of symmetric motifs , in a contiguous symmetric block at the centre ) . The advantage of this is that , when a length w is specified , smaller motifs ( for example , of length w−1 and w−2 ) may also be found . PhyloGibbs-1 . 0 used an annealing schedule where β ( the inverse temperature ) was increased linearly from 1 . 0 to a final lower value , followed by a short “deep quench” with β = 20 . Instead , PhyloGibbs-MP uses the “free energy” E ( the logarithm of the posterior probability ) , averaged over a cycle , to determine the start and stop of the simulated anneal . The starting temperature is chosen to be a value where the fluctuations in energy , ( averaged over one cycle of moves ) , are at least 0 . 3 times the average energy Eavg . Then β is increased exponentially , by a factor of 1 . 2 at each step . At least two cycles are run at each β , and β is increased only when the difference in average energy in the last two cycles at that β is less than the fluctuation ΔE in the last cycle . The anneal is stopped when the relative difference in average energy at the last two values of β is less than 0 . 005 . There is no “deep quench” . The number of moves used in the tracking phase is , by default , the same as the number ( excluding the equilibriation moves to find the initial temperature ) in the simulated anneal . This can be overruled . A significant improvement in running time ( typically a factor of 10 or so ) is obtained by using a form of “importance sampling” on top of the Gibbs sampling scheme implemented in the “window shift moves” . When sampling a replacement window , PhyloGibbs-1 . 0 would consider every available window , with every possible new colour for that window; this requires NwNc calculations , where Nw is the number of windows and Nc is the number of colours available . Typically Nw is large , several thousands or tens of thousands; but only a small fraction of available windows tends to get selected . This is not an unusual situation in sampling problems . When one has an estimate of the bias , one often uses an “importance function” F ( C ) to indicate which configurations C are more frequently visited . This is chosen ( often heuristically ) to be large where P ( C ) ( the posterior probability of C ) is likely to be large large , and small where P ( C ) is small . Then states are sampled not according to P ( C ) but according to P ( C ) F ( C ) ; but their contribution to running averages of the form 〈E ( C ) 〉C are taken to be not E ( C ) but E ( C ) /F ( C ) . This makes the average come out right , while causing the sampler to spend most of its time in “important” parts of configuration space . In our case , we have a related situation: some windows tend to be selected much more often than others , and therefore , during a particular Gibbs move , although we need to select from Nw windows , only a few of these are actually likely to be selected . Therefore we maintain an “importance” for each window , a number between 0 and 1 , which is the fraction of time that that window has actually been selected ( in any colour and either orientation ) up until that time . During the setting of the initial temperature , all windows are treated as important ( their importance counter is incremented whether or not they are selected ) . Thus , at the start of the anneal , all windows have importance 1 , but as the running time proceeds , the importance of many windows decreases rapidly . Let the importance of a window at any time be I ( w ) . Normally , if Nw windows may possibly be selected , we need to calculate the posterior probability P ( wi , c ) for each available window wi into each available colour c . The time-consuming step is the calculation of P ( wi , c ) which is a wasted calculation for most available windows , which are in fact never selected . Therefore , we pre-select a subset of the Nw windows: Each available window wi is selected to be sampled with a probability proportional to I ( wi ) . ( I is normalised in such a way that the importance of the most important available window is 1 , i . e . it will always be pre-selected , so the pre-selected subset is never null . ) However , it will then be selected with a probability proportional , not to P ( wi , c ) , but to P ( wi , c ) /I ( wi ) . In other words , if a subset {W} of windows is preselected by importance , windows in that subset are sampled according to ( the importance cancels because the P's have been preselected with probability i , but then been multiplied by 1/I ) . When the reverse transition is considered , in individual cases a different subset of windows {W′} will be used in the denominator , and therefore the sum in the denominator will not be the same . However , on average ( for times not too far apart ) it will be the same: there is no systematic bias and detailed balance should apply . In the long time limit , I ( wi ) will tend to ( 4 ) This is a non-rigorous argument , for which the justification is as earlier: statements about detailed balance are rigorous only in the infinite-time limit , while running-time efficiency is important in real life . For those who are unconvinced , importance sampling may be turned off by a command-line parameter . Results do not seem greatly affected by this . An example is in Figure 1 , where PhyloGibbs-MP with 8 possible motifs is run with ( black line ) and without ( blue line ) importance sampling . The effect of importance sampling seems , as one would expect , to somewhat inflate the reported significance of high-confidence predictions and somewhat deflate the significance of lower-confidence predictions . For all benchmarks , detailed commandlines for individual programs and all input and output files are available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ . Several scripts used to process the data are also included . For the yeast benchmark , the SCPD database [11] was used , but edited to remove very long and very short sites . The edited database contained 466 binding sites upstream of 200 genes . For each gene , sequence upstream of that gene in S . cerevisiae up until the next coding sequence was used , to a maximum of 1000 bp . Orthologous sequence was found in other sensu stricto species ( S . bayanus , S . mikatae , S . kudriavzveii , S . paradoxus ) using BLAST . For PhyloGibbs-1 . 0 , PhyloGibbs-MP , and EMnEM , these sequences were aligned with Sigma [19] , version 1 . 1 . 3 . For the phylogenetic Gibbs sampler [18] , the sequences were aligned with ClustalW [21] , and for PhyME [16] , they were aligned with the bundled version of Lagan [20] . Command-line options were chosen to be roughly comparable for all programs; for the phylogenetic Gibbs sampler , they were suggested by one of the authors ( W . Thompson , private correspondence ) . For the fly benchmark , we used the REDfly 2 . 0 [22] database of transcription factor binding sites . The “sites” in this database are often many times longer than the expected length of an individual binding site , so the most probable binding site needed to be found bioinformatically . Therefore , we selected a subset of these binding sequences corresponding to factors for which high-quality position weight matrices were available from the TRANSFAC 7 . 0 public database [40] , and used these matrices to find the most probable ( highest log-odds ) binding sites within the binding sequences in the database . ( As noted in “Module prediction” , we have independently constructed position weight matrices for a much larger set of factors from the Flyreg [38] database , which formed the basis for REDfly's binding site database , using PhyloGibbs-MP . But for the purposes of this benchmark , we prefer to use matrices of a “neutral” origin . ) The specific factors chosen from the TRANSFAC database were Abd-B , Adf1 , Cf2 , Dfd , Eip74EF , Stat92E , Su ( H ) , Ubx , bcd , dl , ftz , hb , ovo , pan , sna , z . Binding sites located less than 200 bp apart were clubbed together; then surrounding sequence was selected , such that the total length of the sequence was 250 bp per binding site , to a maximum of 2000 bp . In addition to the programs benchmarked on the yeast data , we wanted to include PhyloCon [23] , a somewhat different category of motif-finder that requires multiple input sequences , each with its own orthologous sequences , and expects one site per input sequence . We therefore chose only a subset of the above factors for which , after the above process , multiple binding sequences were available for each factor . These were Adf1 , Cf2 , Stat92E , Su ( H ) , Ubx , bcd , dl , ftz , hb , ovo , and sna . Orthologous sequence was identified in D . yakuba , D . erecta and D . simulans using multiple sequence alignments from VN Iyer , DA Pollard and MB Eisen ( personal communication ) , but were re-aligned using Sigma . For the synthetic-sequence benchmark , 200 sets of input files were generated . Each set consisted of two files; each file contained five sequences . These sequences were generated from a single ancestor , containing five copies each of two embedded motifs , and evolved according to the evolutionary model assumed by PhyloGibbs , with q = 0 . 5 and no indels . ( As in the evolutionary model , embedded sites , when mutated , were re-sampled from the weight matrix that described them . ) Three motifs were used for each set , where one was common to both files in the set and one each was unique to each file in the set . The motifs were random weight matrices with a “polarisation” ( maximum element in each column ) of 0 . 85 ( with the remaining weights 0 . 05 each ) . The performance of PhyloGibbs-MP ( with different settings of the “discriminative parameter” -d ) , ALSE , DEME and Dips in detecting the common , and the discriminative , motifs was measured . For the yeast motif-finding benchmarks , 571 pairs of genes were chosen such that the pairs had no documented binding sites in SCPD from a common transcription factor . As in the motif-finding case , orthologous sequence from the four other sensu stricto species was used . PhyloGibbs-MP was run in discriminative mode ( -d 0 . 4 ) , and performance in detecting known sites was compared with ALSE , DEME and Dips . For the fly benchmarks , we used the REDfly data used in the motif-finding example , but removed the PhyloCon-imposed requirement of multiple sequence sets per transcription factor ( allowing both single and multiple sequences per set ) . We ended with 1404 pairs of sequence sets in which each member of a pair was associated with a different factor . As in the motif-finding benchmark , orthologous sequence from D . yakuba , D . erecta , and D . simulans was used . Similarly to the yeast case , the four discriminative programs were run . For the yeast ChIP-chip benchmarks , we used the spreadsheet Harbison_Gordon_yeast_v9 . 11 . csv from the supplementary data of Harbison et al . [31] to extract transcription factors that bind to between 4 and 9 regulatory sequences , both in rich medium and in at least one other environmental condition , with a p-value better than 0 . 001; and for each factor we retrieved the sequences it bound to . We used intergenic sequence upstream of the regulated gene , to a maximum of 1000 bp , as in the SCPD benchmarks . 17 such factors were found , of which 15 included motifs reported in their supplementary data file Final_InTableS2_v24 . motifs . Figure 7 lists these factors , and sequence logos constructed from the motifs listed in that file . For each factor , also , if it was reported to bind to n sequences ( 4≤n≤9 ) , we selected n sequences at random to which it was not reported to bind , to a p-value of 0 . 01 or less . These were used as the “negative” set . Orthologous sequence from sensu stricto species were included , aligned for PhyloGibbs-MP with Sigma version 1 . 1 . 3 ( as in other benchmarks ) . In the case of PhyloGibbs-MP , predictions in the negative set were discarded ( PhyloGibbs-MP does not distinguish between positive and negative sets ) , and only predictions from the positive set that arose from multiple tracked windows , at least one of which had a tracking score better than 0 . 2 , were considered . The other programs reported at most one motif each for the positive set , and each such prediction was considered . All input and output files , and detailed commandlines , are available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ . CRMs documented in the REDfly [33] database were chosen , that were under 3000 bp long . Surrounding sequence was included to bring the total length of the sequence to 10000 bp . If two CRMs lay within 15000 bp of each other , the associated sequence was fused into a single sequence . In this way , 234 CRM-containing sequences were identified in D . melanogaster . Orthologous sequence was selected from D . pseudoobscura , D . yakuba , and D . simulans . Orthology identification was made using multiple sequence alignments from VN Iyer , DA Pollard and MB Eisen ( personal communication ) . The sequences were re-aligned with Sigma 1 . 1 . 3 . PhyloGibbs-MP was run without priors and with priors constructed from the FlyReg database [38] ( see below ) . Stubb ( version 2 . 1 ) and Cluster-Buster were run using the FlyReg priors . Cis-Module requires no priors . EMC-Module requires a prior set of known binding sites , so the FlyReg priors were used to predict these sites ( sites with a log-odds larger than 7 were included , and the specified width of sites was 8 ) . PhyloGibbs-MP was run with 1 , 2 ( melanogaster and yakuba ) or all 4 input sequences , aligned with Sigma 1 . 1 . 3; Stubb was run with 2 input sequences , melanogaster and yakuba . Other programs were run on melanogaster alone . All input and output files , and detailed commandlines , are available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ . The FlyReg [38] database of DNAse I footprints in D . melanogaster was used . Only those TFs were considered for which two or more footprints were available . For each footprint , orthologous sequence was extracted for D . pseudoobscura , D . yakuba , D . simulans , D . erecta , and D . ananassae using multiple sequence alignments from VN Iyer , DA Pollard and MB Eisen ( personal communication ) . Thus , where N footprints may have been available in FlyReg , up to 6N sequences were used including orthologous sequences . The command-line for PhyloGibbs-MP was chosen via a heuristic depending on the number and lengths of footprints . The detailed commandlines and output files , and the generated weight matrices , are available at http://www . imsc . res . in/∼rsidd/phylogibbs-mp/supporting-data/ .
|
Proteins in a living cell are not expressed all the time: instead , genes are turned on or off on demand . Indeed , though nearly every cell in a multicellular organism has a complete copy of the genome , each cell expresses only a fraction of the encoded proteins . Regulation of gene expression occurs in various ways . One of the most important ( especially in simpler organisms ) is “transcriptional regulation , ” where specialised DNA-binding proteins , “transcription factors , ” attach to the DNA to recruit the gene-transcriptional machinery . Detecting binding sites in DNA for these factors has long been a problem of interest in computational biology . Here , a program , PhyloGibbs-MP , is presented that extends our previously published motif-finder PhyloGibbs to handle some important related problems , in particular , detecting “discriminative” sites that distinguish differently regulated groups of genes and finding “cis-regulatory modules , ” regions of DNA that contain large clusters of regulatory-protein-binding sites . PhyloGibbs-MP compares well on benchmarks with the best specialised programs for all these tasks , while being the first to integrate them in one consistent formalism . Regulatory regions in higher eukaryotes can be highly complex , and PhyloGibbs-MP is expected to be a very useful tool in identifying and analysing regulatory DNA .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"computational",
"biology/transcriptional",
"regulation"
] |
2008
|
PhyloGibbs-MP: Module Prediction and Discriminative Motif-Finding by Gibbs Sampling
|
Pathogens have evolved diverse strategies to maximize their transmission fitness . Here we investigate these strategies for directly transmitted pathogens using mathematical models of disease pathogenesis and transmission , modeling fitness as a function of within- and between-host pathogen dynamics . The within-host model includes realistic constraints on pathogen replication via resource depletion and cross-immunity between pathogen strains . We find three distinct types of infection emerge as maxima in the fitness landscape , each characterized by particular within-host dynamics , host population contact network structure , and transmission mode . These three infection types are associated with distinct non-overlapping ranges of levels of antigenic diversity , and well-defined patterns of within-host dynamics and between-host transmissibility . Fitness , quantified by the basic reproduction number , also falls within distinct ranges for each infection type . Every type is optimal for certain contact structures over a range of contact rates . Sexually transmitted infections and childhood diseases are identified as exemplar types for low and high contact rates , respectively . This work generates a plausible mechanistic hypothesis for the observed tradeoff between pathogen transmissibility and antigenic diversity , and shows how different classes of pathogens arise evolutionarily as fitness optima for different contact network structures and host contact rates .
There are two major principles by which pathogens avoid their elimination: escaping the host immune response via antigenic variation or immune evasion , or transmission to a new immunologically naive host . Directly transmitted pathogens which cause chronic diseases , such as many sexually transmitted infections ( STIs ) , tend to rely more on the former , while many acute infections , for instance measles , rely more on high transmissibility . Indeed pathogens such as measles show very little antigenic diversity , with immune responses being strongly cross-reactive between strains . There are then those pathogens which have intermediate levels of both immune escape and transmissibility — such as influenza , rhinovirus and RSV ( here referred to as FLIs — flu-like infections ) . The evolutionary success of directly transmitted pathogens can also be seen to depend on the nature , frequency and structure of contacts between hosts . Infections transmitted to a small number of hosts ( per time unit and infected individual ) via intense contact ( e . g . , via fluids ) are usually caused by pathogens of high antigenic diversity and long duration of infection , while those transmitted via casual contact ( e . g . , via aerosol ) with a large number of hosts may typically have lower diversity and much shorter durations of infection . While many of the evolutionary constraints are different [1] , [2] , vector-borne infections typically fall in the former of these two classes [3] , [4] . The relationship between so-called infection and transmission modes with respect to substitution rates of RNA viruses has been investigated in [5] . It is straightforward to explain the long duration of infection and consequent antigenic diversity of sexually transmitted or blood-borne infections: the frequency of relevant contacts between hosts is low , meaning infection needs to be extended to ensure the reproduction number ( the number of secondary cases per primary case [6] ) exceeds one . However , many childhood diseases ( ChDs ) — at least those caused by RNA viruses — would also seem to have the genetic potential to prolong their survival within one host via by generating antigenic variants . The fact this is not observed is much harder to explain . At its root are the tradeoffs between maximizing between-host transmissibility and within-host duration of infection , and these are what we focus on exploring in this paper . The molecular genetic basis of transmissibility is still poorly understood for most pathogens . However , all other things being equal , the level of pathogen shedding by a host ( whatever route is relevant ) must be positively correlated with infectiousness . A first-pass analysis might therefore postulate that overall transmissibility ( as quantified by the basic reproduction number , ) might be proportional to the total number of pathogen copies produced during an infection — the cumulative pathogen load . Past work using a simple model of the interaction between a replicating pathogens and adaptive host immune responses examine what rate of antigenic diversification within the host would maximize cumulative pathogen load [7] . This showed that the combination of resource-induced ( whether nutrients or target cells ) limits on peak pathogen replication rates and an ever more competent immune response mean that the optimal strategy is not to diversify as rapidly as possible , but instead to adopt an intermediate rate of diversification . In addition , there are further tradeoffs associated with high mutation rates — the ultimate being the error catastrophe associated with error rates in genome replication which exceed those seen in RNA viruses [8]–[11] . However , the assumption that transmission fitness ( as quantified by ) is linearly proportion to total pathogen load is clearly naïve . The instantaneous hazard of infection for a susceptible host in contact with an infected host at a point in time may indeed be linearly related to pathogen load at that time , but going from this assumption to a calculation of the overall reproduction number is far more complex than simply calculating the area under the pathogen load curve . Integrating a hazard over the finite time of contact gives an exponential dependence between the probability of infection and pathogen load , i . e . , . Such an expression fits experimental data [12] on the relationship between HIV viral load and transmission rates well ( cf . Fig . 1 ) . This means the parameter represents a pathogen load threshold below which the probability of infection declines rapidly , and above which it rapidly saturates to some maximal value . Hence can be thought of as the characteristic pathogen load required for transmission — though it is not a true minimum infectious dose — there is a finite probability of infection for , but that probability decays exponentially fast with reducing . A key insight ( and assumption ) of the work presented here is that while we might expect pathogens to be able to evolve to reduce ( or increase ) , there are fundamental physical constraints imposed by transmission routes on the minimum value of attainable . An STI might have a minimum value of approaching a single pathogen particle ( e . g . virion ) but , for respiratory infections , the much lower proportion of all pathogen particles emitted from a host , which have any chance of contacting epithelial tissues of a susceptible host ( even conditioning on a susceptible host being in the near vicinity of the infected individual ) , necessarily means that must be orders of magnitude larger for such pathogens . We will show that there is a critical value of above and below which two different sets of pathogen types are evolutionarily favored ( in terms of having maximal ) . Within each set , the particular type which has maximal will be seen to depend on the local structure of the contact network between hosts . Our approach is to construct a model of within-host pathogen dynamics which incorporates adaptive host immunity and antigenic diversification . The key output from this model is how pathogen load varies through time during an infection . We then calculate the basic reproduction number , , for that infection assuming a particular local contact network structure and frequency of contacts . The within-host model developed here is an extension of a model studied earlier by one of us [7] . Our work builds on a range of past work examining the tradeoffs between within-host replication and persistence , antigenic variation and between-host transmission success , initiated by [13] , and followed by [14] , [15] , which first include immune response and explore cross-immunity . More recent studies , to mention a few , investigate pathogen evolution under limited resources [16] , include virulence [17] , consider the immunological response in more detail [18] , examine the impact of between-host contact structure on pathogen evolution [19] , [20] , and explore host-pathogen co-evolution [21] , [22] . We use as our fitness measure for determining evolutionarily optimal phenotypic strategies . We do not explicitly model competition between pathogen strains with different phenotypes co-circulating in a host population , since for infinite populations , has been shown to be the fitness measure which determines the outcome of such competition [23] . This holds even when comparing strains with different rates of antigenic diversification — if the strain with lower induces no long-lived immunity in the host ( giving SIS dynamics ) and the higher strain induces life-long immunity , ( giving SIR dynamics ) the higher strain will still always ( eventually ) outcompete the lower strain . There are limitations to the use of as a fitness measure ( further considered in the Discussion ) — for instance , in situations where strains interact asymmetrically via cross-immunity , or when populations are small and stochastic extinction is significant . In addition , while we take account of local ( egocentric ) network structure in defining in our analysis , large-scale network structure might also affect the determinants of evolutionary fitness . However , we feel these limitations are outweighed for an initial analysis by the analytical and computational tractability afforded by use of a relatively simple transmission measure , and the consequent ability not to rely on unintuitive large-scale simulations . We do not explicitly consider how a pathogen could evolve its biological characteristics to maximize transmission fitness ( i . e . the evolutionary trajectory a pathogen would take through parameter space ) . There are undoubtedly many constraints on the possible paths which pathogens can take [24] , however , and exploring how these affect , for instance , pathogen adaptation to a new host species , will be an important topic for future work .
The multi-strain model used extends past work [7] by adding cross-immunity between strains ( see Methods for details ) . The infection within one host starts with a single strain , with further strains arising through random mutation . All strains compete for resources ( e . g . target cells ) to replicate . Immune responses to strains are assumed to be predominantly strain-specific , albeit with a degree of cross-immunity , the strength of which decays with the genetic distance between strains . Pathogen replication depletes resource , and independently from immunity , limits to pathogen growth are set by the replenishment rate of resource . This quantity only determines the short-term dynamics of the model whereas immunity is also responsible for the long-term behavior . The dynamics of the model is characterized by an initial period of exponential growth of the pathogen load , which eventually slows due to immune responses and resource limitations . One observes a latency period and an initial peak . Pathogen load then declines exponentially . If the trough load of a pathogen strain drops below a threshold level we assume the pathogen is eliminated from the host ( to avoid persistence at unrealistically low , fractional , loads ) . However if a novel strain emerges before the seed strain goes extinct , pathogen load can recover , so long as there is sufficient resource available and cross-immunity is not too strong — leading to a second , albeit lower peak in pathogen load . Further peaks in pathogen load can occur via the same mechanism . The rate at which new strains arise is the most important determinant of the number of pathogen load peaks seen and thus the overall duration of infection . Less intuitively , this rate also determines the size of the initial peak ( discussed below ) . Since mutation is modeled stochastically , we average over multiple realizations ( e . g . Fig . 2A , B ) of the model to calculate an average pathogen load distribution over time ( Fig . 2C ) . The average distribution consists of a first latency period , a large initial peak , a second latency period and possibly an irregular oscillating part of low pathogen load . The point at which the viral load vanishes determines the duration of infection . We systematically calculate average pathogen load curves from the within-host model for wide ranges of two biological parameters: the antigenic mutation rate ( i . e . , the rate of mutations which lead to antigenically novel strains ) and the pathogen replication rate . These two parameters span what we call pathogen parameter space , in which evolutionarily favored pathogens are represented by points that are associated with maximal fitness values . From the discussion in the introduction , we can immediately identify the cumulative pathogen load and duration of infection as epidemiologically relevant quantities . Fig . 3A , B show these as a function of the parameters and . In addition , Fig . 3C shows a quantity — interpolating between the two former — evaluated only for the initial period of the infection ( utilizing the expression relevant for transmission , i . e . , , quantified at the initial peak of the pathogen load ) . We will see below that all the surfaces shown in Fig . 3A–C crudely represent fitness surfaces associated with three distinct pathogen types . The plots in Fig . 2 show the corresponding within-host dynamics for the different pathogen types . The within-host dynamics generate a tradeoff between initial peak pathogen load and antigenic diversity: high initial peak load corresponds to low diversity and vice-versa ( see Methods for more details ) . This tradeoff has implications for transmission , giving an enhanced spread of pathogens of low antigenic diversity during the initial peak of pathogen load . This effect explains the emergence of ( ChD-like ) infections with short durations of infection within our model framework ( Fig . 3C vs . 3F ) . Long durations of infections ( Fig . 3B ) are also obtained , as expected , for pathogens with greater antigenic variation . To calculate the reproduction number ( i . e . , the pathogen fitness ) , we model a dynamic contact network in the neighborhood of one initially infected host . The profiles of pathogen load over time obtained from the within-host model then determine the infectiousness of the infected host to its neighbors . ( We utilize the mean-load profiles averaged over individual hosts . ) Epidemiological dynamics are determined by 4 parameters . Two of these relate to properties of the transmission route: the infectiousness parameter and the contact rate between hosts . Together these define a two-dimensional parameter space we term transmission space . The other two define properties of the contact network between hosts: the replacement rate of neighbors and the cliquishness/clustering of the network ( i . e . , the proportion of pairs of contacts of a host who are also contacts of each other ) . These two parameters define what we term contact space . We build a model ( cf . Methods ) incorporating these 4 parameters ( plus implicitly the within-host pathogen space parameters ) to calculate the number of first generation infections from an infected individual in an entirely susceptible population . Varying the 4 parameters of transmission and contact space , we obtain three different classes of fitness landscapes over pathogen space — as represented by Fig . 3D–F . The maxima of each landscape differ with respect to their antigenic mutation rate ( and hence the resulting level of antigenic diversity ) and within-host pathogen replication rate . By changing the contact rate and keeping the other transmission as well as the contact space parameters fixed , one can shift between these classes . In general ( as shown further below ) , low , intermediate , and high contact rates induce moderate , high , and low antigenic diversity , respectively , as evolutionarily favored outcomes ( represented by the locations of the fitness maximum in Fig . 3D–F ) . There are clear similarities between the three classes of fitness landscapes ( Fig . 3D–F ) and the different within-host infection characteristics plotted in Fig . 3A–C . Low contact rates induce landscapes that resemble the cumulative pathogen load , intermediate contact rates give landscapes resembling the the duration of infection surface , and high contact rates map onto the surface of Fig . 3C which characterizes the relative importance of the initial peak in the pathogen load profile . We classify the optima of these 3 classes of fitness landscape infection types , labeling them A , B , and C , respectively . Varying the infectiousness parameter can also move the fitness landscape between these types — as ( the STI limit; i . e . , , ) , the fitness landscape becomes more similar to the duration of infection surface ( Fig 3B ) , while for ( the FLIs limit; i . e . , , ) , it becomes more similar to the cumulative pathogen load surface ( Fig . 3A ) ; cf . ( 7 ) and ( 6 ) . It is important to note that both of these limits involve substantial antigenic diversity — where transmission fitness is dominated by cumulative pathogen load ( infection type A ) , while moderate antigenic diversity is seen , and when infection duration dominates fitness ( infection type B ) , high antigenic diversity is selected for . Neither maps on to the special case of infection type C ( Fig . 3F ) in which optimal transmission fitness is achieved by a set of parameters giving very low antigenic diversity ( in essence a single strain ) . For low antigenic diversity to be optimal , it is necessary for fitness to be dominated by the peak pathogen load achieved during primary infection ( i . e . , the first peak of pathogen load ) . Varying the transmission and contact space parameters more systematically , one can map out the regions of parameter space for which particular infection types are optimal ( Fig . 4 ) . This shows how the emergence of pathogens of different types depends on the properties of the between-host contact network . Pathogens with low antigenic diversity ( and thus short infectious periods ) are favored by high network cliquishness ( i . e . , when an individual's contacts are contacts of each other — as is the case for household and school contacts ) , and the rate of turnover of network neighbors is low ( again the case for household and school contacts ) . So far we have assumed only the pathogen space parameters ( and ) can change during pathogen evolution . Now we examine making the infectiousness threshold a parameter which can evolve under selection — albeit with constraints on its lower bound set by the transmission route of the pathogen concerned . Fig . 5 shows the results as a function of contact rate for two different choices of contact space parameters and lower bounds on the infectiousness threshold parameter , suitable for a respiratory pathogen and an STI respectively . Reproduction numbers ( Fig . 5B ) lie in the expected range , and the three regimes of antigenic diversity corresponding to the types A/B/C ) can be found in the evolutionarily optimal values of ( Fig . 5A , C ) . Note that only type A and type C diversity is seen for the respiratory pathogen parameter choices , while only type B is seen for the STI parameter set . Indeed for the STI parameter set , the evolutionary stable state is independent of the contact rate , and is determined by evolving to its minimum value . As expected , the evolutionary optimal value of the infectiousness parameter ( Fig . 5B ) is always close to the minimal attainable value , except in the type C pathogen regime ( where cliquishness is necessary; cf . Fig . 4 ) . The reason for the deviation from the minimum value lies in a reduced local network saturation , which is characteristic for type C: concentrating infectiousness over the shortest possible time period ( and consequently lengthening the latent period ) shortens the overlap between generations of infections , and this reduces the chance that the secondary cases of an index case infect remaining susceptible contacts of the index ( before the index can infect them ) . The effect ( which yields an enlarged susceptible number in ( 6 ) ) is minor , however — the difference in between the optimal value of and the minimum bound set for a pathogen type is typically very small . The evolutionarily optimal replication rate is always low for STI-like contact parameters ( giving type B pathogens ) , reflecting the need for long-lived infections , but shows greater variability for respiratory pathogen parameter regimes ( Fig . 5D ) — being high in the type A regime , but low for type C . The latter result reflects a tradeoff between height of the initial peak in pathogen load and length of the latent period — longer latency , as explained above , can increase the number of direct infections caused by an index case by reducing the overlap between generations of infection . Only higher ( minimal ) infectiousness values — realistic for ChDs utilizing the respiratory transmission route — increase the optimal replication rate for type C infections ( cf . Text S1 , Sect . B . 2 ) . Note that these results are consistent with a recently formulated hypothesis on tradeoffs between reproductive rate and antigenic mutability [25] , proposing a reciprocal relationship between these two ( pathogen space ) parameters in real-world infections . Re-examining Fig . 4 , it is clear that type A infections ( green areas ) only exist when the infectiousness parameter exceeds some minimum value ( indicated on the graphs in Fig . 4 with an arrow ) . In the absence of constraints , selection for maximal transmissibility will clearly cause to evolve towards 0 . Hence the effect of constraints on imposing a lower bound on has a critical effect on what range of pathogen types are expected . We define the value of the lower bound on infectiousness below which infection type A is no longer found the critical infectiousness threshold . Evolutionary dynamics show a phase transition at this point , as can be seen in Fig . 6 which maps the areas of contact parameter space for which different infection types are seen for choices of the lower bound on just above and below the critical point . As discussed already , the transmission route is likely to be the most important determinant of the lower bound on , with STIs and other non-airborne pathogens , including those requiring a vector , being likely to achieve a much lower value of than respiratory pathogens ( as assumed in Fig . 5 ) . This is clear if one views as quantifying how much shed pathogen is typically wasted to achieve a single infectious contact . We therefore speculate that the critical infectiousness threshold may have a significant biological effect , with STIs — and also vector-borne infections — being within the sub-critical domain ( Fig . 6B ) , and with ChDs and FLIs — not necessarily relying on a respiratory transmission route — being super-critical ( Fig . 6A ) . Within the super-critical regime , the presence of low-diversity ChD-like type C infections depends less on the precise value of the critical infectiousness threshold and more on the contact rate and contact parameters . Infections of type C occur in contact networks with high cliquishness and low replacement rates — but not in the opposite case ( cf . presence of blue areas in Figs . 4 and 5A ) . Vector-borne infections ( representing contact networks of large neighborhood sizes or high replacement rates , and cliquishness not playing a role ) are thus excluded to be type C . At first sight they seem to be type A , because of large reproduction numbers . Large , however , can also be the result of large neighborhood sizes or high replacement rates — immediate from ( 6 ) and ( 8 ) . The quantity being important in this context is the lower bound on possible infectiousness values , which is small ( i . e . , sub-critical , ) — this identifies vector-borne infections as type B .
The work in this paper was motivated by a desire to understand why the most transmissible human pathogens — archetypal childhood diseases such as measles and rubella — show remarkably little antigenic variation , while less transmissible diseases — such as influenza ( and many other respiratory viruses ) and sexually transmitted diseases show substantial diversity . Addressing this question requires consideration of how evolvable parameters governing the natural history of infection within a host affect the transmission characteristics of a pathogen in the host population . We developed a relatively simple multi-strain model of the within-host dynamics of infection . Pathogen particle consume resource to replicate , and their replication is inhibited by a dynamically modeled immune response with two components: strain-specific immunity , and cross-immunity . Cross-immunity was assumed to be the key fitness cost of antigenic diversity within the host; the benefit is a much enhanced duration of infection ( and thus transmission ) . Pathogens which have a low rate of generating new antigenic variants are cleared from the host much faster than those with a high rate of antigenic diversification , but also maximize the initial peak level of parasite load reached prior to clearance ( cf . Methods ) . The second evolvable within-host parameter we considered was the within-host pathogen replication rate . Given the resource-dependent model of replication assumed , this has a more limited effect than in some models , but can set the timescale for pathogen load to initially peak and thus determine the effective latent period of the disease . At the between-host level , we assume a simple relationship between pathogen load and infectiousness which has been shown to be appropriate to model HIV transmissibility [12] , and incorporates the concept of a soft threshold level of pathogen load needed for a substantial level of transmissibility , . As argued above , this parameter is perhaps best viewed as the amount of excreted pathogen which is wasted to achieve an infectious contact . For a perfect pathogen , the value could correspond to a single pathogen particle , but in reality the physics of transmission will typically mean is much higher . We have considered to be an evolvable parameter , but introduced the concept of minimum possible value of which is transmission route specific — being intrinsically much higher for respiratory pathogens ( where transmission occurs via virus filling a three-dimensional volume around the infected individual ) , and potentially much lower for sexually transmitted diseases where transmission occurs over a two-dimensional contact surface . The final element we incorporate into the framework developed is contact between hosts , assumed to occur at some rate , within a contact network of hosts with a certain mean neighborhood size and cliquishness . We derive a simple model to calculate the reproduction number of a single infected host in this network allowing for local saturation effects in the network caused by clustering . It is the network-specific reproduction number we have used as our overall measure of pathogen fitness , and examine what within- and between-host pathogen characteristics maximize fitness for different types of transmission route and host contact network . Putting these elements together , we found that optimizing reproductive fitness in this way leads to well-defined infection types A , B , C , as contact rates ( and reproductive numbers ) increase ( cf . Fig . 5 ) . Type A and B both represent infections with low , with A being influenza-like and B mapping more to sexually transmitted diseases . When contact rates are very low , only one of these two types is evolutionary stable , with the stable type being determined by the assumed minimum infectiousness threshold . The latter serves as an order parameter and determines the mode of transmission . Consistently , type A corresponds to a high minimum infectiousness threshold whereas type B results from a low minimum threshold . The change of the transmission mode as a function of transmission threshold is phase transition-like . Infection type C represents childhood diseases with the highest values of . This regime is not possible for small network neighborhood sizes or low values of cliquishness ( i . e . random networks ) . It relies on the existence of large , persistent and highly clustered contact neighborhoods . In this context , maximizing the number of secondary infections ( and thus overall fitness ) requires a pathogen strain able to ( a ) infect as many of the index host's contacts as possible in as short a possible time , and ( b ) minimize the extent to which generations of infections overlap . The latter constraint is a result of the network clustering — if secondary cases become infectious while the index case is still infectious , they may deplete susceptible from the contact neighborhood before the index case has the chance to infect them . A latent period of the same or longer duration as the infectious period results in more discrete generations and maximizes the reproduction number of the index case . The need for a long latent period results in the evolutionary optimal value of the within-host replication rate , being relatively low for type C pathogens . The limited antigenic diversity and short infectious periods of type C pathogens are determined by the higher infectiousness threshold and the consequent need to maximize the peak pathogen load attained early in infection . When contact rates are high , the increase in duration of infection resulting from higher rates of antigenic diversity is insufficient to compensate for the reduction in peak pathogen load ( and therefore infectiousness ) caused by cross-immunity being generated against multiple pathogen strains simultaneously . A single strain pathogen generating a single immune response is able to generate a larger primary infection peak — though at the cost of being unable to sustain infection further . It is encouraging to see that the classification of infection types our model predicts closely corresponds to many of the pathogen regimes identified in other work [24] . However , our focus has been slightly different from that work , which focused more on the effect of different intensities of cross-immunity on between host phylodynamics . In contrast , we have focused more on examining how differences in transmission routes and contact rates ( ) determine pathogen characteristics — though the influence of different levels of cross-immunity could be explored in future work . Furthermore , it is interesting to note that in the context of our model only the concept of a minimal infectiousness threshold — introduced to characterize transmission modes — is necessary to explain the findings of [25] on tradeoffs between reproductive rate and antigenic mutability . Reference to the host's age is not needed here . The key limitation of our analysis is our highly simplified treatment of between-host transmission — namely using a network-corrected reproduction number as our measure of strain fitness . Doing so assumes evolutionary competition occurring in infinite ( non-evolving ) host populations in infinite timescales . It would clearly be substantially more realistic to explicitly simulate the transmission process in a large host population . The computational challenges are considerable — while large-scale simulations of influenza A evolution and transmission have been undertaken [7] , [26] , [27] , these have not included within-host dynamics , and have simulated evolution for decades rather than millennia . Other work [20] , [28] has simulated the evolution of pathogen strains on a contact network for longer time periods , but only in very small ( ) populations , and without modeling within-host dynamics . However , continuing advances in computing performance mean that it may now be feasible to explicit model multiple strains evolving within hosts and being transmitted independently in a large population . Such an approach would allow exploration of the relationship between antigenic diversity ( and cross-immunity ) within single hosts and strain dynamics at a population level . Perhaps even more importantly , it would allow extinction processes to be properly captured , while our current approach implicitly assumes fixation probabilities to be 1 even when fitness differences are marginal . Proper representation of finite population sizes and extinction will also allow the evolutionary emergence of childhood diseases ( such as measles ) as a function of early urbanization to be modeled . A second limitation is that we only consider a single , highly simplified within-host model . Future work to test the sensitivity of our results to the choice of within-host model would be valuable ( cf . Text S1 , Sect . A , which investigates an extension of the model here ) . That said , we would argue that the key qualitative feature of our within-host model driving the evolutionary results is the tradeoff — mediated by cross-immunity — between the maximum value of parasite load attained in initial infection and the degree of antigenic diversity ( and thus duration of infection ) . Also a conceptual simplification must be pointed out here: our model assumes that mutations , controlled by , directly affect antigenicity . For real-world pathogens , however , the link between genetic and antigenic change is less clear . Measles , for example , has a mutation rate typical of RNA viruses [29] , but its antigenic diversity is low . Instead of mutation rate controlling antigenic variability , a pathogen may evolve phenotypic robustness to genetic change . Further , we have not attempted to capture specialized strategies pathogens have adopted for persistence within infected hosts , such as use of refuges from immune responses ( HSV ) or hijacking the immune system ( HIV ) — the model only reflects tradeoffs which may have contributed to pathogens adopting the range of persistence strategies seen in nature . An interesting addition to future work would also be the incorporation of pathogen virulence [30] , which imposes an additional evolutionary constraint on within-host replication rates . A last area which is a clear priority for future research is the relationship between within-host parasite load and infectiousness . We have assumed a relationship which has some support in data ( Fig . 1 ) , and indeed the HIV system is perhaps the best explored in terms of the possible evolutionary tradeoffs inherent in maximizing transmissibility [31] . Unfortunately , little comparable data is available for other ( especially respiratory ) pathogens .
The within-host dynamics are simulated by the following system of ordinary differential equations ( see [7] for more details where this system is introduced without cross-reactive immunity ) : ( 1 ) ( 2 ) ( 3 ) representing ( 1 ) the load of pathogen strain , ( 2 ) the amount of the adaptive immune response specific to strain , and ( 3 ) the level of resource which all strains need to replicate; the number of equations , , corresponds to the number of strains present , where denotes the total pathogen load . For viral infections , for example , the load is assumed to represents the number of virions of strain , the immunity variable somehow the amount of specific antigen ( produced by B cells ) , and the resource target cells ( e . g . , epithelial cells for flu or T cells for HIV ) of maximal number . Saturation effects , modifying linear dependency on and , are modeled with the Hill function . The resource limitations act via , where , for large loads ( ) , growth is limited by the maximal pathogen capacity related with the resource , ; for small loads , the load is independent of the resource , . The adaptive immune response is given by the growth term , which increases in response to antigen quickly and reaches values at . For larger pathogen loads , growth stops slowly , limited by when . The parameter represents the critical load above which immunity saturates . Its value is chosen above the number of pathogen units released after one replication cycle per resource unit ( see below ) . Guided by values for RNA viruses , random mutations are assumed to occur with probability per pathogen replication , which happens at rate . Only a proportion of mutations generate new antigenic variants . We assume that all mutations not leading to new antigenic variants are deleterious . The emergence of new antigenic variants is modeled stochastically , where a Poisson distribution with expectation determines the number of mutant strains at time , with denoting the cumulative load . While back mutations are neglected in the equations above , they are taken account of in the numerical calculations . New antigenic variants generated at time induce a specific immune response , . This grows so long as , and declines for downwards , but never goes below . These characteristics are determined by the structure of ( 2 ) and the parameter choice , where and define the base rates at which immunity is produced and declines , respectively . We assume 5 loci with 3 alleles at each . ( These numbers are small but sufficient for our analyses , cf . Text S1 , Sect . C . ) The distance between strains , , is defined as the number of loci at which strains and differ . The immune-related clearance rate of strain is given by , where and for and 0 otherwise . Here is the degree of cross-immunity , and is the parameter governing homologous clearance rates . Independent of immunity , pathogen is cleared at a rate ( chosen smaller than ; cf . ( 5 ) below ) . Pathogen growth is limited by resource , where defines the saturation point . As pathogen grows at rate , resource is consequently depleted at rate . Resource is replenished at rate , and its total is modeled to never exceed ( chosen to represent a realistic number of target cells and thus give realistic pathogen loads; cf . Fig . 1 and the examples above ) . The differential equations are solved using a Runge-Kutta algorithm with the initial values , and , starting with 1 strain . New antigenic variants are generated potentially after each time step , each with initial pathogen load ( corresponding to 1 pathogen unit infecting 1 resource unit ) and specific immunity , if generated stochastically at . The infection ends once pathogen load drops below the value ( which is assumed to be the elimination threshold ) , or after 2 years ( the latter cutoff being chosen for computational simplicity ) . The parameter values ( essentially , and ) and the regions of pathogen space ( given by and ) have been chosen to produce load curves ( with significant resource depletion at the load peak , i . e . , ) that resemble measles characteristics ( with latency periods of up to 10 days and significant pathogen loads for similar periods; cf . Fig . 2F ) for small antigenic variation and small/intermediate reproduction rates . The duration of infection is adjusted by the strength of immunity ( i . e . , ) , with the value used here selected to give infections of over 1 year duration for maximal antigenic variation . This model is minimally complex , incorporating only the features essential to explain the tradeoff between transmissibility and antigenic diversity . A more realistic model is examined in Text S1 , Sect . A . However , the key diversity-transmissibility tradeoff arises as a simple consequence of within-host cross-reactive immune responses raised to individual new strains and competition between strains for a common resource for replication , and is relatively independent of the model-specific form of implementation of these mechanisms . The essential within-host dynamics of our combined within/between-host model is given by Eq . ( 1 ) , which links pathogen replication to two inhibitors — host immunity and resource limitation . This equation quantifies the tradeoff for increasing antigenic diversity ( the pathogen's survival strategy within the host ) — namely the smaller initial pathogen load peak seen in Fig . 3C ( and Fig . S1-2C in Text S1 , Sect . B . 1 ) . The specific realizations for the acquisition of immunity and the replenishment of resource ( modeled by Eqs . ( 2 ) and ( 3 ) , respectively ) are less important . Let us consider the pathogen load dynamics soon after infection with one initial strain . Our numerical simulations have shown that the initial strain is much more prevalent ( by orders of magnitude ) than mutant strains produced up to the first peak , . This observation clarifies that resource limitation ( as one inhibitor of pathogen growth ) cannot explain the tradeoff discussed here — being of low prevalence , mutant strains are unlikely to deplete resource to an extent which results in significantly lower loads , and in any case all strains have the same intrinsic replication rate and use the same resource . But the specific immune response to mutant strains , provided it is partially cross-reactive , is able to reduce both the load of the initial strain and other strains , and can thus lower the total pathogen load . This result is largely independent of model implementation and only depends on the strain-specific immune response being generated at relatively low strain-specific pathogen loads , and being sufficiently cross-reactive to slow overall growth of pathogen load . This verbal argument can be formalized . For simplicity we assume the load of the initial strain is a good approximation of the total pathogen load at the initial peak , . By applying as a condition for the initial peak , Eqn . ( 1 ) ( with ) then yields a relation for the initial peak load , ( 4 ) where defines the immune response with respect to the strain number . Provided cross-reactive immunity is implemented ( i . e . , for some , so that ) , the function is strictly increasing ( independently of how cross-immunity is defined via the strain-distance weight function and the parameter ) . This is based on the fact that , together with each newly generated strain , immunity is produced in a standardized way for the time period up to the initial peak when load is increasing and above a critical value , . This is the case in any setting where mutant strains have the same intrinsic replication kinetics as the initial strain . In our model , immune production happens at rates above ( and below ) as long as , independently of the concrete acquisition rule in ( 2 ) ; see the modifications ( Eqs . ( S1-1 , 2 ) ) and the corresponding result ( Fig . S1-2C ) in Text S1 for a more realistic but also more complicated mechanism . As a consequence of resource limitation ( i . e . , the reduced growth ) , ( 4 ) yields ( 5 ) Due to the monotony of , the function given by ( 5 ) is strictly decreasing . That means that the magnitude of the initial peak is inversely related to the number of ( mutant ) strains present . The result is independent of the specific functional form used for resource depletion ( in ( 1 ) ) and replenishment ( in ( 3 ) ) , as is confirmed by considering the limit of large pathogen loads , where ; the resulting peak height , , shows the same monotonic dependence on as ( 5 ) . Finally , we examine what would happen if cross-immunity or resource limitation were not implemented in the model . Without cross-immunity , , and the initial peak is thus independent of the strain number ( cf . Fig . S1-2I in Text S1 , Sect . B . 1 ) . Without resource limitation , ( 4 ) degenerates , and the initial peak load cannot be compared for different values of antigenic variation . As discussed in the text , we use the basic reproductive number of infected hosts as the measure of evolutionary fitness for infectious diseases [23] . For infections of finite duration , ( 6 ) where denotes the number of susceptible hosts in the neighborhood ( of assumingly constant size ) of one initially infected host , and is the transmission rate from the index case at time after infection . The pathogen-load dependence of the transmission rate is modeled by ( 7 ) where is the infectiousness threshold parameter and is the transmission coefficient , which critically depends on the contact rate . The parameter is the transmission probability per contact for a completely saturated pathogen load ( ) , and lies between 0 and 1 . This functional form is consistent with data for HIV ( Fig . 1 ) . The transmission dynamics in the entire susceptible contact neighborhood of an index case are given by ( 8 ) where . This equation models a local dynamic network ( derived in the section below ) , where defines the transitivity or cliquishness of the network ( proportion of neighbors of a node who are neighbors of each other ) and the per-capita rate at which hosts in the neighborhood of the index case are replaced by new susceptible hosts . Here represents convolution , with . This expression corrects for the depletion of the local contact neighborhood of the primary case by individuals infected by the index case then infecting shared contacts of the index before the index case herself does . Such local saturation of the susceptible population is entirely a network effect and vanishes for . It should be noted that the network dynamics are invariant for , bar a scaling of by . Enlarging the neighborhood size thus corresponds to effectively reducing cliquishness . This relation allows for incorporating vector-borne infections ( characterized by large ) into our classification ( as type B infections; cf . end of the section Infection types ) . Although our modeling framework has been designed for direct transmissions , it can formally be applied to vector-borne infections assuming that ( due to relatively low ) the transmission delay through the vector is less important . Here we derive Eqn . ( 8 ) of our between-host model , which also illustrates how the two parameters , and , characterize the host-contact network on local and on global scales , respectively . The transmission dynamic in an initially entire susceptible contact neighborhood of one index case and fixed size , , can be reconstructed approximately in terms of average numbers of infectives and susceptibles ( and , resp . ) , ( 9 ) counting the ( infinitesimal ) number of new infections caused by the index case at time . We have included direct infections and secondary infections which , we assume , occur with likelihood in the contact neighborhood . The time delays , , as reflected by the transmission rates relevant for secondary infections , correspond to primary infections at . The integral covers the secondary infections caused by new infectives up to time , respecting the changing transmission rates resulting from time-dependent pathogen loads ( cf . ( 7 ) ) . Written exclusively in terms of susceptibles ( while utilizing the notion of convolution ) , ( 9 ) reads ( 10 ) From here , Eqn . ( 8 ) is obtained by incorporating a constant ( global ) flow of individuals ( referring to the entire host population ) into the transmission model , quantified by the replacement rate of individuals in the considered neighborhood . This is readily confirmed by the formal replacement ( of the ordinary derivative by a covariant version ) , ( 11 ) which models the recruitment of new susceptibles in exchange for old infectives .
|
Infectious diseases vary widely in how they affect those who get infected and how they are transmitted . As an example , the duration of a single infection can range from days to years , while transmission can occur via the respiratory route , water or sexual contact . Measles and HIV are contrasting examples—both are caused by RNA viruses , but one is a genetically diverse , lethal sexually transmitted infection ( STI ) while the other is a relatively mild respiratory childhood disease with low antigenic diversity . We investigate why the most transmissible respiratory diseases such as measles and rubella are antigenically static , meaning immunity is lifelong , while other diseases—such as influenza , or the sexually transmitted diseases—seem to trade transmissibility for the ability to generate multiple diverse strains so as to evade host immunity . We use mathematical models of disease progression and evolution within the infected host coupled with models of transmission between hosts to explore how transmission modes , host contact rates and network structure determine antigenic diversity , infectiousness and duration of infection . In doing so , we classify infections into three types—measles-like ( high transmissibility , but antigenically static ) , flu-like ( lower transmissibility , but more antigenically diverse ) , and STI-like ( very antigenically diverse , long lived infection , but low overall transmissibility ) .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"computational",
"biology/evolutionary",
"modeling",
"public",
"health",
"and",
"epidemiology/infectious",
"diseases",
"infectious",
"diseases",
"public",
"health",
"and",
"epidemiology/epidemiology"
] |
2009
|
Antigenic Diversity, Transmission Mechanisms, and the Evolution of Pathogens
|
The intracellular protozoan parasite Leishmania donovani causes human visceral leishmaniasis . Intracellular L . donovani that proliferate inside macrophage phagolysosomes compete with the host for arginine , creating a situation that endangers parasite survival . Parasites have a sensor that upon arginine deficiency activates an Arginine Deprivation Response ( ADR ) . L . donovani transport arginine via a high-affinity transporter ( LdAAP3 ) that is rapidly up-regulated by ADR in intracellular amastigotes . To date , the sensor and its ligand have not been identified . Here , we show that the conserved amidino group at the distal cap of the arginine side chain is the ligand that activates ADR , in both promastigotes and intracellular amastigotes , and that arginine sensing and transport binding sites are distinct in L . donovani . Finally , upon addition of arginine and analogues to deprived cells , the amidino ligand activates rapid degradation of LdAAP3 . This study provides the first identification of an intra-molecular ligand of a sensor that acts during infection .
Leishmania donovani is a parasitic protozoan that causes visceral leishmaniasis ( kala-azar ) in humans that is almost always fatal [1] . This organism is an obligatory intracellular pathogen which cycles between the acidic phagolysosome of macrophages ( intracellular amastigote ) and the relatively alkaline mid-gut of female sand flies ( extracellular promastigote ) [2] . Shifting between the two distinct environments requires flexible adaptation mechanisms , and the ability to sense and rapidly respond to fluctuations in vector and host environments [3] . We use L . donovani as a model organism to investigate the molecular tools parasites have developed to overcome the harsh environments they encounter in the host and vector . Arginine is a semi-essential amino acid for mammals and is required for protein synthesis and other metabolic pathways like the synthesis of urea , polyamines and creatinine [4] . On the other hand , arginine is an essential amino acid for Leishmania , as the parasites cannot endogenously synthesize it and rely on its uptake from host arginine pools . Additionally , arginine is the sole precursor for the production of spermidine , and subsequently trypanothione , in Leishmania [5 , 6] . The parasite utilizes arginine for polyamine biosynthesis as it lacks both catalases and classical seleno-containing glutathione peroxidases and thus has to rely on trypanothione for maintaining redox balance [5] . Arginine is transported within L . donovani via a high-affinity arginine transporter ( LdAAP3 ) , which is specifically localized to the membrane of the flagellum and glycosomes [7] . LdAAP3 is a highly specific transporter and only a few compounds affect/inhibit it [8] . When over-expressed , LdAAP3 also localizes on the plasma membrane [8] . During infection with L . donovani , macrophages employ defense mechanisms such as nitric oxide ( NO ) and reactive oxygen species ( ROS ) production [9] . The synthesis of NO by macrophages requires arginine . The intracellular parasites increase arginase I activity in macrophages , which restricts the amount of arginine available for NO production , and also increase the production of trypanothione , thereby neutralizing ROS and facilitating a safe niche for themselves inside the macrophages [10 , 11] . Therefore , Leishmania is locked in competition with the host for arginine . L . donovani modulates infected host macrophages to stimulate arginine transport by up-regulating the expression of the cationic amino acid transporter 2- CAT2B ( SLC7A2 ) , thereby ensuring its own survival [12] . We have previously reported the presence of an arginine deprivation response ( ADR ) in L . donovani , which monitors arginine levels in its environment . Genome-scale transcriptomics identified six transporters that up-regulate upon ADR activation in L . donovani , which include , in addition to LdAAP3 , the pteridine , folate and three putative transporters [7] . Interestingly , we observed that ADR activation occurs in intracellular amastigotes , thereby supporting the notion that parasites undergo arginine deprivation during development in the phagolysosome . Most eukaryote amino acid sensors respond to sufficiency or deficiency by activating a mTOR signaling cascade [13] . In contrast , we have previously shown that the L . donovani arginine sensor responds to environmental arginine deprivation by activating a mitogen-activated protein kinase 2-mediated signaling pathway that within minutes up-regulates the expression of the ADR genes [7] . This arginine sensor was identified for the first time and was reported to induce a response in the absence of its ligand . Furthermore , the addition of arginine to ADR-activated promastigotes induced a rapid degradation of the LdAAP3 protein to its homeostatic level . This indicated that the arginine sensor activates two distinct pathways: the ADR and an arginine sufficiency response . To date , we have not identified the L . donovani arginine sensor or its ligand . This study identifies the regions on the arginine molecule which are essential for ADR activation and binding to the arginine transporter . We show that the conserved amidino group at the distal cap of the arginine side chain is the ligand that activates/suppresses ADR , both in axenic promastigotes and intracellular amastigotes . Using arginine analogues and additional compounds that contain this group , but lack the α amino group , we observed that arginine sensing and transport binding sites are distinct in L . donovani . Furthermore , these analogues affect arginine sensing of intra-lysosomal amastigotes . Sensor specificity is very high as any modification of the amidino group results in non-recognition of the arginine analogue by the sensor and thus has no effect on ADR . This study comprises the first in-depth analysis of arginine sensing in Leishmania .
L-arginine , D-arginine , Pentamidine , Canavanine , N-Methyl L-arginine acetate ( NMLAA ) , Nω-Nitro-L-arginine methyl ester ( L-NAME ) , Nω-Nitro-L-arginine ( L-NNA ) , L-citrulline , 3-Ureidopropionic acid and 4-{[5- ( 4-aminophenoxy ) pentyl]oxy}phenylamine were obtained from Sigma-Aldrich , USA . All other materials used in this study were of analytical grade and commercially available . Promastigote cultures of the L . donovani Bob strain ( LdBob strain/MHOM/SD/62/1SCL2D ) , initially obtained from Dr Stephen Beverley ( Washington University , St . Louis , MO , USA ) , and L . donovani 1S strain , ( MHOM/SD/00/1S ) were used in this study . Promastigotes were cultured at 26°C in M199 medium ( Sigma-Aldrich , USA ) , supplemented with 100 units/ml penicillin ( Sigma-Aldrich , USA ) , 100 μg/ml streptomycin ( Sigma-Aldrich , USA ) and 10% heat-inactivated fetal bovine serum ( FBS; Biowest ) . THP-1 cells , an acute monocytic leukaemia-derived human cell line ( ATCC , TIB-202TM ) were used for all experiments . They were cultured in RPMI-1640 ( Sigma-Aldrich , USA ) medium supplemented with 10% heat-inactivated FBS , 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C in a humidified atmosphere . For infection assays , 0 . 5 x 106 cells/ml were seeded in RPMI-1640 medium containing 10% FBS . The cells were treated with 50 ng/ml of phorbol 12-myristate 13-acetate ( PMA; Sigma-Aldrich , USA ) for 48 h to induce their differentiation into macrophage-like cells . Immediately before infection , the cells were washed once with phosphate buffered saline ( PBS ) and incubated in RPMI medium ( Sigma-Aldrich , USA ) containing 0 . 1 mM arginine ( unless stated otherwise ) , and supplemented with 10% heat-inactivated FBS , 100 units/ml penicillin , and 100 μg/ml streptomycin . Promastigotes in the late log-phase were added to cells at a ratio of 20:1 and incubated at 37°C in a humidified atmosphere for 5 h . Extracellular parasites were removed by washing the cells five times with PBS . Thereafter , the cells were incubated in RPMI medium containing 0 . 1 mM arginine ( unless stated otherwise ) at 37°C in a humidified atmosphere for 2 h , 24 h , or 48 h . Mid-log phase promastigotes ( 1×107 cells/ml ) were used for arginine deprivation studies . The cells were washed with Earl’s salt solution twice and re-suspended in arginine deficient Medium 199 ( Biological Industries Ltd . ) . Arginine deprivation was carried out at 26°C for specified time periods and was concluded by transferring the cells to ice . Arginine deprived cells were washed twice with Earl’s salt solution before being used for transport assays , Northern and Western blot analysis . THP-1 cells were cultured in DMEM medium with 10% FBS . Magnetic beads of 3 μm size were added to flasks containing THP-1 cells . The isolation of intact macrophage phagosomes was carried out as described by Kuhnel et al . [14] . The isolated phagosomes were then resuspended in 1 ml of ice-cold phosphate-buffered saline plus 1 ml of 5 N perchloric acid , vortexed , and incubated on ice for an additional 10 min . Perchloric acid lysates were centrifuged in a microcentrifuge at 14 , 000 rpm for 10 min at 4 °C , and 232 μl of 5 N KOH was added to the supernatant to titrate the sample pH to 7 . 0 . Additional centrifugation was performed under the aforementioned conditions , and 200 μl aliquots were analyzed for amino acid content by the method described by Fekkes et al . [15] . The analyses were carried out at the Medical Biochemistry Laboratory at Rambam Medical Center in Haifa . Uptake of 25 μM [3H]arginine ( 600 mCi/ mmol ) , by mid-log phase parasites was determined by the rapid filtration technique of Mazareb et al . as reported [16] . To determine initial rates of transport , transport measurements were performed on 1 x 108 promastigotes exposed to radiolabel for up to 2 min . The amount of radiolabel associated with the cells was linear with time over the 2-min time course of the transport assay . Total RNA from L . donovani promastigotes ( either deprived for arginine or non-deprived ) was prepared using the Tri-reagent protocol and subjected to Northern blotting for LdAAP3 as described before [17] . Probes were amplified using the following primers: LinJ . 31 . 900 LdAAP3 Forward: 5'-GCTGTGACGGGGTCAGTG-3' and . Reverse: 5'-GTACGTCGCCAGCCAGTG-3' . LdBPK_101450 . 1 pteridine transporter Forward: 5’- ATGACCGTTGGTCAGCAGA-3’ and Reverse: 5’- GCCGTGGTGACGCCGTACT-3’ . RNA quantification , cDNA preparation , and real-time PCR were performed as discussed previously [18] . Briefly , total RNA was isolated from cells by Tri reagent ( Ambion , Thermo Fischer Scientific , USA ) . The concentration and purity of RNA were determined by Nanodrop ( Thermo Fischer , USA ) . Two micrograms of RNA were treated with RNase-free DNase ( Promega , USA ) , and subsequently reversed transcribed into cDNA by using the First-strand cDNA synthesis kit ( Thermo Scientific , USA ) , as per the manufacturers’ instructions . Real-time PCR was performed on the resulting cDNA with gene-specific primers ( LdAAP3: Forward 5’ CGGTCGAAATGGTGCCAAAC 3’ , Reverse 5’ GGCTTCATCTTCCCTGCGTA 3’; LdPT: Forward 5’ AGGACGCTGCTCAACTCTTC 3’ , Reverse 5’ AAGGCGAACGTGTCACTCAA 3’; kinetoplast minicircle DNA ( control ) [19]: JW11 5’ CCTATTTTACACCAACCCCCAGT 3’; JW12 5’ GGGTAGGGGCGTTCTGCGAAA 3’ ) using the PowerUp SYBR Green Master Mix ( Applied Biosystems , USA ) . The PCR amplification program used was as follows: 50°C for 2 minutes ( min ) and 95°C for 10 seconds ( sec ) , followed by 40 cycles at 95°C for 15 sec , 59°C for 1 min , and 72°C for 20 sec . The amplification of the kinetoplast minicircle DNA of L . donovani was used as an internal control . The results were expressed as fold change of control ( 2 h infected cells ) using the method described by Pfaffl , 2001 [20] . The real time PCR primers did not amplify any products in uninfected macrophages . All sample analysis was performed in triplicate , and each experiment was performed three times . At the specified times following infection or treatment with inhibitors , THP-1 cells were washed two times with PBS . MTT [3- ( 4 , 5-dimethyl-2-thizolyl ) -2 , 5-diphenyltetrazolium bromide] dye solution ( Sigma-Aldrich , USA ) ( 5 mg MTT in 1 ml PBS ) was diluted 1:10 in RPMI medium ( normal or with 0 . 1 mM arginine ) . Uninfected or infected THP-1 cells were incubated in diluted MTT dye solution at 37°C in a 5% CO2-air atmosphere for 2 h and thereafter incubated with stopping solution which consisted of isopropanol containing 5% formic acid , at 150 rpm , 37°C for 20 min . L . donovani promastigotes were incubated with diluted MTT dye solution for 3 h at 37°C , and incubated with stopping solution comprising of isopropanol and 20% SDS in a 1:1 ratio , at 80 rpm , 37°C for 30 min . Absorption was then measured at 570 nm , and the percentage cell viability was calculated . THP-1 cells were seeded on glass coverslips ( 1 x 106 cells/well ) in a 6-well plate and treated with 50 ng/ml of PMA ( Sigma-Aldrich , USA ) for 48 h . They were infected as described above , and the intracellular parasite load was visualized by Giemsa staining . Western blot analysis of LdAAP3 was performed as described previously by Darlyuk et al . , 2009 [17] . Real-time PCR data were analyzed by GraphPad prism and represented as mean ± standard error of the mean ( S . E . M . ) . Student’s unpaired 2-tailed t-test was used to calculate significance . P value < 0 . 01–0 . 05 was considered statistically significant ( * ) , p < 0 . 001–0 . 01 was considered very significant ( ** ) , and p < 0 . 0001–0 . 001 was considered extremely significant ( *** ) . The northern blot image data was analyzed using ImageJ analysis software . Data from three independent experiments were analyzed and represented as mean ± standard error of the mean ( S . E . M . ) .
The first set of experiments aimed to identify the basal concentration of arginine required for the activation of the ADR in L . donovani promastigotes . The maximal concentration of arginine required for ADR activation was found to be 5 μM , which resulted in the up-regulation of LdAAP3 and threshold was lower at 0 . 5 μM for the pteridine transporter ( LdPT ) at mRNA level ( Fig 1A ) . However , unlike LdAAP3 the expression of LdPT mRNA was not linear dose dependent . Interestingly , this concentration of arginine is close to the apparent Km value of 2 . 4 μM for L . donovani LdAAP3 transport activity [8] , raising the question whether this transporter is also the sensor . Previously , we observed that 48 hours after infecting THP-1 macrophages , the expression of LdAAP3 in intracellular L . donovani increased almost two-fold as compared to promastigotes [7] . This suggested that during development inside phagolysosomes , parasites encountered a low level of arginine that activated ADR . In this case , increasing external arginine concentrations might help to maintain the phagolysosome arginine concentration above the threshold . To test this , we infected THP-1 macrophages that grew in media containing 0 , 0 . 1 , 0 . 5 and 1 . 5 mM arginine . At 48 hours post-infection , total RNA was extracted from infected macrophages , and the resulting cDNA was subjected to real-time PCR , using LdAAP3 and LdPT primers as probes , as mentioned in the methods section ( Fig 1B and 1C , respectively ) . As shown , the mRNA abundance of both genes increased as arginine concentration in the medium decreased . The results indicate that the arginine concentration in phagolysosomes of macrophages grown in a medium that contains arginine at a concentration of 0 . 1 mM and below activates ADR . A literature search indicated that the arginine concentration in human blood is ~80 μM [21] , thereby indicating that ADR activation in our experiments was achieved under physiological conditions . Additionally , the infectivity of L . donovani in THP-1 cells cultured in media containing different concentrations of arginine was ~40% ( S2A Fig ) . As control , THP-1 cells , either uninfected or infected with L . donovani in medium containing different concentrations of arginine were subjected to an MTT assay to determine their viability at 48 hours . As seen in S1A Fig , the macrophages were 85–100% viable in media containing different arginine concentrations , thus indicating that ADR activation in intracellular amastigotes was not detrimental to macrophage viability . Additionally , the expression of LdAAP3 remained unchanged in L . donovani promastigotes cultured in media containing 0 . 1 mM and 0 . 5 mM arginine ( S1B Fig ) , and promastigote viability was between 90–100% ( S1C Fig ) . This proves that extracellular parasites did not contribute to the observed activation of ADR in intracellular amastigotes . ADR activation in intracellular amastigotes was also determined in a time-course analysis of the infection cycle . THP-1 cells were infected with L . donovani in medium containing 0 . 1 mM arginine for 2 h , 24 h and 48 h post-infection and harvested at the end of each time point for RNA analysis ( Fig 1D and 1E ) . Real-time PCR showed that the up-regulation of LdAAP3 and LdPT started at 24 h post-infection and continued to increase at 48 h post-infection , thereby implying that the activation of ADR in intracellular amastigotes occurs between 24–48 hours post-infection . This infers that the initial arginine concentration in infected phagolysosomes is high and reduces with time , reaching ADR activation level at 24 h post-infection . However , it could also be possible that induction of the ADR is delayed due to the time taken to deplete intracellular parasite pools of arginine following their phagocytosis . As seen in Fig 2B , the infectivity of L . donovani in THP-1 cells at 2 h , 24 h and 48 h post-infection was between 28–38% . Previous studies have indicated that arginine concentration in the mammalian lysosomes is higher than that in their cytosol [22] and in Saccharomyces cerevisiae vacuoles [23] . Because lysosome volume was not determined in this study it did not provide accurate arginine concentrations . We have now determined the arginine concentration in THP-1 macrophage phagolysosomes that seems to agree with the observation of Harms et al . ( Fischer-Weinberger et al , in preparation ) . The initial arginine concentration in THP-1 phagolysosomes is 0 . 14 mM , a concentration that is higher than the concentration we found that activates ADR in axenic parasites . This further supports our findings that the late activation of ADR during infection is due to the time it takes for intra-phagolysosome parasites to utilize arginine and reduce its concentration to ≤5 μM . Arginine is a positively charged amino acid with an amidino group at the distal cap of its side chain . We hypothesized that this side chain of arginine is the ligand that binds the arginine sensor and transporter on the parasite surface . To determine this , we first analyzed various structural L-arginine analogues which were previously shown to be arginine transport inhibitors [24 , 25] . Canavanine has a guanidinoxy group with a conserved amidino group as in arginine , and N-methyl L-arginine acetate ( NMLAA ) is another structural analogue of L-arginine where the amidino group is modified by the addition of a methyl group . The methyl-amidino group of NMLAA has a net charge of zero at pH 7 . NMLAA has been previously shown to be an arginine transport inhibitor in L . donovani [24] and canavanine inhibits arginine transport in S . cerevisiae [26 , 27] and T . brucei [25] . To test the effect of these structural analogues on ADR , we determined the minimal concentration of each structural analogue that upon two hours treatment inhibited arginine transport in promastigotes but had no effect on cell viability ( Table 1 ) . This list includes 0 . 5 mM canavanine , and 1 mM NMLAA ( left column ) . We determined the effect of these compounds on ADR activation in promastigotes . This was carried out in arginine-depleted L . donovani axenic promastigotes cultured in M199 with or without arginine transport inhibitors followed by Northern blot analysis using gene-specific probes for LdAAP3 and LdPT . Fig 2A shows that Canavanine inhibited LdAAP3 ( 27%±1 . 9 ) and LdPT ( 61%±2 . 4 ) mRNA up-regulation . In contrast , NMLAA did not have any inhibitory effect on ADR as the expression of both the genes was found to be up-regulated upon arginine deprivation ( Fig 2A ) . Identical results were obtained at the protein level for LdAAP3 in a Western blot analysis for cananvanine and NMLAA ( Fig 2B ) . Hence , methylation of the amidino group in NMLAA retained binding to the LdAAP3 transporter but lost recognition by the sensor . Further , the effect of the enantiomer D-arginine as compared to L-arginine on ADR was determined . D-arginine is not a competitive inhibitor of L-arginine transport [24] . This was carried out in L . donovani promastigotes in M199 medium with or without D-arginine ( 1 mM ) and Northern blot was performed as mentioned above . Fig 3 shows that D-arginine had no effect on ADR and did not lead to the degradation of LdAAP3 and LdPT mRNA . This indicated that , the external arginine sensor in the parasite could distinguish between D-arginine and L-arginine . We further checked what is the effect of chemical compounds that have a conserved amidinio side chain group of arginine and one such compound was pentamidine . Pentamidine is a diamidine that has two amidino moieties previously shown to be competitive arginine transport inhibitors in L . donovani [8 , 28] . These amindino groups have a net positive at pH7 . Analysis of the effect of 100 μM pentamidine on axenic promastigotes showed that pentamidine drastically inhibited both LdAAP3 ( 81%±2 . 8 ) and LdPT ( 72%±3 . 2 ) up-regulation at 2 h post arginine deprivation ( Fig 2A ) . This suggests that the minimal molecular group necessary for recognition by the surface arginine sensor and transporter binding sites is the amindino group of the arginine side chain . Similar result was also seen in Western blot analysis where pentamidine inhibited the ADR-stimulated expression of LdAAP3 ( Fig 2B ) . This minimal amidino group of the arginine side chain is sufficient to inhibit ADR and arginine transport and it is direct proof that the α-carbon group is not necessary for arginine sensing as well as transport . Further , the analogues of L-arginine which have a modified amidino/guanidine group ( like NMLAA ) and their effect on ADR was analyzed . Nω-Nitro-L-arginine methyl ester ( L-NAME ) and Nω-Nitro-L-arginine ( L-NNA ) are arginine analogues where the amidino group is modified by the addition of a nitro-group ( nitro-amidino ) . The net charge of the nitroguanidinium group in L-NAME is neutral and the primary α-amino group is positively charged [29] , while the gaunidino group of L-arginine has a net positive charge . Both arginine analogues have been previously shown to be weak arginine transport inhibitors in L . donovani [24] . In addition to the above mentioned arginine transport inhibitors , other compounds which have a deamidated guanidino group were also tested for their effect on ADR . L-citrulline is an alpha amino acid and an intermediate in the urea cycle . It has a carbamoylamino group which is formed by the deamination of the guanidium group , and has previously been shown by our group to be a poor arginine transport inhibitor for LdAAP3 [8] . Citrulline is known to have no net charge at pH 7 unlike arginine which has a net positive charge . Similarly , 3-Ureidopropionic acid has an carbamoylamino group similar to citrulline but has a missing primary amine . Additionally , a pentamidine analogue 4-{[5- ( 4-aminophenoxy ) pentyl]oxy}phenylamine , which has the two amidino groups of pentamidine substituted with amines was also analyzed . All these compounds were used further to test their effect on ADR in L . donovani ( see all the molecular structures in Fig 4 ) . There was no inhibition of LdAAP3 and LdPT levels when L . donovani promastigotes were treated with the above mentioned compounds ( Fig 3 ) . This indicates that there are two distinct arginine binding sites in the ADR machinery in L . donovani . The complete list of molecular structures of all the arginine analogues ( structural and side chain analogues ) is shown in Fig 4 . These compounds were not tested for the regulation of LdAAP3 and LdPT at the protein level in promastigotes and at the RNA and protein levels in intracellular amastigotes as their effect at the RNA level was similar to that of NMLAA . Thus , NMLAA and other compounds with a modified guanidino/amindino group have a net charge of zero as compared to the net positive charge of the guanidino/amindino group of arginine . This suggests that in addition to the presence of a functional amidino or guanidino group , the positive charge on the R-group of arginine is also an important factor for the recognition of arginine by the L . donovani sensor . We have previously reported that the arginine sensor responds to both arginine deprivation and sufficiency and the addition of exogenous arginine to two hours arginine starved axenic promastigotes induces rapid degradation of the LdAAP3 protein to the level observed in un-deprived cells [7] . In order to further analyze arginine sufficiency response at the mRNA level , exogenous arginine ( 0 . 45 mM ) was added to two hours arginine deprived axenic promastigotes which induced the rapid degradation of LdAAP3 ( rate of t1/2 = 30 minutes ) and LdPT mRNA ( Fig 5A ) . This implies that the regulation of the LdAAP3 degradation signal occurs not only at the protein level as previously suggested [7] , but also at the post-transcriptional level . The minimal threshold of arginine sufficient to be detected by the arginine sensor as arginine sufficiency and thereby downregulate ADR was also checked . However , LdPT mRNA was degraded more rapidly as seen in Fig 5A and did not follow the same mRNA degradation kinetics as LdAAP3 . It was observed that 10 μM exogenous arginine did not activate arginine sufficiency signaling via arginine sensor , but 50 μM , 100 μM and 450 μM of exogenous arginine were detected as arginine sufficiency , thereby resulting in the downregulation of LdAAP3 levels ( Fig 5B ) . In order to ascertain whether the arginine sufficiency phenomenon also occurs in intracellular amastigotes , THP-1 macrophages were infected with L . donovani for 48 h in medium containing 0 . 1 mM arginine , and then excess arginine ( 1 mM or 5 mM ) was added for an additional 2 h . RNA was extracted from infected macrophages before and after the addition of excess arginine and subjected to real-time PCR . This resulted in the down-regulation of LdAAP3 and LdPT mRNA in intracellular amastigotes ( Fig 6A and 6B ) . This indicated that intracellular amastigotes , like the axenic , respond to arginine sufficiency by rapidly down-regulating ADR . As seen in Supp . Fig 2C , the differences in infectivity of L . donovani in THP-1 cells treated ot not with exogenous arginine were not significant . As it was observed that the arginine transport inhibitors ( pentamidine and canavanine ) inhibited ADR , we subsequently checked the effect of these inhibitors on arginine sufficiency . L . donovani axenic promastigotes were incubated for 2 h in M199 medium lacking arginine . Thereafter , arginine transport inhibitors ( pentamidine ( 100μM ) , canavanine ( 500μM ) , NMLAA ( 1mM ) , D-arginine ( 1mM ) , L-NAME ( 1 mM ) , L-NNA ( 1 mM ) , L-citrulline ( 1 mM ) , 3-Ureidopropionic acid ( 1 mM ) and 4-{[5- ( 4-aminophenoxy ) pentyl]oxy}-phenylamine ( 100 μM ) ) were added to arginine deprived cells , and the cells were harvested at different time-points . Upon Northern blot analysis , it was observed that only pentamidine and canavanine down-regulated ADR as evidenced by the rapid degradation of LdAAP3 ( Fig 7A ) and LdPT ( Fig 7B ) in a time-dependent manner . However , NMLAA ( Fig 7A ) and the other analogues ( Fig 8 ) had no effect on ADR . In order to ascertain whether the same holds true in THP-1 cell-derived intracellular amastigotes , the first step was to determine the ideal concentration of pentamidine required to suppress ADR in intracellular amastigotes . As seen in S3 Fig , a dose-response analysis revealed that 100 μM of pentamidine inhibited the expression of LdAAP3 mRNA . Treatment with 100 μM pentamidine and 500 μM canavanine led to a decrease of both LdAAP3 and LdPT mRNA levels in intracellular amastigotes , while 1 mM NMLAA did not have any significant effect on their expression ( Fig 9A , 9B and 9C ) . Additionally , 100 μM pentamidine , 500 μM canavanine and 1 mM NMLAA did not significantly affect the infectivity of L . donovani in THP-1 cells ( S2D Fig ) . In order to verify if ADR inhibition by arginine analogues in intracellular amastigotes was not due to changes in cell viability , THP-1 cells were treated with different concentrations of pentamidine , canavanine or NMLAA for 24 h and 48 h , following which they were subjected to MTT assay in order to determine their viability . Cells treated with 100 μM pentamidine , and 1 mM canavanine or NMLAA exhibited 80–100% viability , which was the concentration used for treating infected THP-1 cells for ADR inhibition ( S4A–S4C Fig ) . In conclusion , the above results confirmed that the amidino group is the ligand that binds the arginine sensor . This means that pentamidine and canavanine are recognized by the sensor as “arginine” .
In this study , we have identified that the amidino group on the arginine side chain is the specific ligand that binds the L . donovani surface arginine sensor , and thereby activates an arginine deprivation response ( ADR ) . We also show that arginine transporter and sensor binding sites are distinct in both axenic and intracellular L . donovani . Our analysis has indicated that the sensor is more selective in terms of its ligand as compared to the transporter . The arginine sensor not only detects the lack of arginine in the environment but also responds to excessive extracellular arginine by inducing rapid degradation of the LdAAP3 protein and mRNA . This study provides the first identification in Leishmania of an intermolecular region or functional group of arginine that interacts with a receptor . In all other sensors identified to date , such intermolecular recognition has not yet been described . In our present study , it was observed that the arginine structural analogue canavanine and diamidine pentamidine inhibit not only arginine transport but also ADR , in axenic promastigotes as well as intracellular amastigotes . Arginine is a positively charged amino acid with an amidino group ( pKa = 13 . 8 ) at the distal cap of its side chain [30] . Pentamidine , a diamidine , and a potent anti-protozoal agent , possesses two positively charged amidino moieties which are a part of the guanidium group and have a pKa of 12 . 1 [28 , 31 , 32 , 33] . Canavanine , on the other hand , is a structural analogue of arginine that has a deprotonated guanidinooxy group ( pKa = 6 . 6 ) [25 , 34] . Considering that there is no similarity in the backbone structure of pentamidine and arginine other than the amidino moiety , each of these amidino moieties is the minimal structure which is required to be recognized by the arginine sensor . Any modification of the amidino group , such as the replacement of hydrogen with methyl-group ( in case of NMLAA ) or replacement of hydrogen with any other groups as in the case of the amidino-modified arginine side-chain analogues and compounds , resulted in their non-recognition by the arginine sensor even when these compounds have an arginine backbone . Among the nine compounds tested in the present study , only pentamidine , canavanine and NMLAA inhibited arginine transport in Leishmania , while the others are known to be poor transport inhibitors [8 , 24] . This provides conclusive evidence that the arginine sensor of L . donovani is highly specific and exclusive to the amidino group of arginine and also implies that arginine sensing and transport binding sites are distinct in Leishmania parasites , in axenic promastigotes and intracellular amastigotes . However , the arginine transporter site is still permissive to change ( unlike the sensor site ) which does not affect the net charge of the transported compound . This finding opens up the possibility of employing new arginine analogues containing the amidino group as therapeutic agents in leishmaniasis . D-Arginine is an enantiomer of L-Arginine . The parasite arginine sensor seems to have two levels of ligand recognition: i ) the enantiomer ( D/L ) of the primary amine of arginine and ii ) the amidino group . Thus , the presence of D-Arginine may not activate ADR even though it has an amidino group . When the chemical ligand lacks primary amine as in the case of pentamidine , the arginine sensor recognizes the presence of amidino group and hence triggers ADR . The up-regulation of various members of the ADR pathway including LdAAP3 , pteridine transporter , folate/biopterin transporter among others upon arginine starvation of L . donovani suggests that the ADR pathway not only regulates the expression of LdAAP3 but also other transporters of essential compounds such as vitamin B9 ( folate ) and pteridine which are essential for Leishmania . The variation in the RNA degradation profile seen during arginine sufficiency suggests that LdAAP3 and other members of the ADR pathway including LdPT are regulated differently . The mRNA degradation of LdAAP3 was slower as compared to LdPT as it is an central gene in the ADR pathway . Macrophage phagolysosomes evolve from late endocytic compartments [35] and are the sites of Leishmania amastigote differentiation [36] . The arginine concentration in phagolysosomes was found to be 140 μM ( Fischer-Weinberger et al , in preparation ) . We also found that ADR activation in intracellular L . donovani amastigotes occurred between 24 and 48 h post-infection under physiological levels of arginine ( ~100 μM ) , which is considerably higher than the concentration which induces ADR in axenic parasites ( 5 μM ) . Thus , intracellular arginine starvation builds up between 24 and 48 h post-infection , which may be the time taken for the depletion of arginine levels from 140 μM ( the physiological concentration in phagolysosomes ) to 5 μM ( which is sufficient for ADR activation in axenic L . donovani ) . Hence , it is noteworthy that the axenic Leishmania parasite model system established by various research groups including ours [37] is well-suited for deciphering molecular mechanisms in Leishmania as this is a clean system without host-protein interference . Our present study is the first to elucidate the specificity of the parasite arginine sensor , as well as its arginine sufficiency and deficiency responses , both in axenic and intracellular parasites . Sensing nutrient availability in vector or host environment may be essential for parasite survival and growth . Thus , nutrient sensing and transport pathways can be promising drug targets in the protozoan parasite Leishmania [38] . Our study also provides evidence that the L . donovani arginine sensor has a dual function of response to not just arginine deprivation as we have reported earlier [7] , but also to arginine sufficiency , and thus maintains homeostasis . Transceptors are dual function solute transporters that concomitantly sense and translocate their substrates , and localize to cell surface and organelle membranes [39] . A well-characterized example of the transceptor is the SSY1 transceptor in Saccharomyces cerevisiae [40] . It is part of a membrane-sensing system that detects extracellular amino acids by binding to it [41 , 42] and signal transmission for detecting the presence of extracellular amino acids is initiated at the cell membrane . Mammals possess the System A amino acid transporter 2 ( SNAT2 ) , which also acts as a transceptor that signals and senses neutral amino acid availability [43] . More recently , an arginine trans-membrane sensor ( SLC38A9 ) has been identified on the membrane of mammalian lysosomes [44 , 45] . However , signaling is initiated at the cell membrane where the arginine sensor is most likely localized . Whether the arginine sensor and transporter sites in L . donovani are located in the same protein or different proteins remains an open question . Also , the existence of multiple arginine transporters , sensors or transceptors implies that arginine is evolutionarily conserved and indispensable in both lower and higher organisms . In recent years , research on flagella-localized transporters in Leishmania suggest that these transporters may be involved in ligand sensing . Some of the well-known transporters include glucose transporter 1 ( GT1 ) from L . mexicana [46] and aquaglyceroporin ( AQP1 ) from L . major involved in osmoregulation [47] . The phenomenon of sensors localization on the flagella has recently been discussed [48] . In light of this , the fact that arginine is an essential amino acid in Leishmania , and semi-essential in mammals , represents its global role in various cellular processes . In a commentary to our previous paper on the discovery of ADR , McConville [49] suggested that the arginine sensing phenomenon is a metabolic crosstalk between Leishmania and the macrophage host . Our finding in this work that ADR is activated in parasites inside phagolysosomes , early in infection , strongly supports this idea . Furthermore , the identification of the arginine intermolecular ligand is unprecedented and provides an efficient tool to further explore host-parasite interaction .
|
Leishmania donovani , the causative agent of visceral leishmaniasis , leads a digenetic life cycle as a flagellated promastigote in the vector sandfly and aflagellated amastigote within phagolysosomes of infected macrophages . Arginine is an essential amino acid for Leishmania which possesses a high specificity arginine transporter ( LdAAP3 ) , a protein that imports the amino acid into parasite cells . Arginine is primarily utilized in de novo protein synthesis and for biosynthesis of trypanothione via the polyamine pathway . It was previously reported by our group that L . donovani senses lack of arginine in the surrounding micro environment and activates a unique arginine deprivation response ( ADR ) pathway , thus upregulating the expression of LdAAP3 as well as other transporters . In the present study , we identified the region on the arginine molecule which is the ligand that activates ADR . We show that the conserved amidino group at the distal cap of the arginine side chain is the ligand that activates/suppresses ADR . Using arginine analogues that contain this group we observed that arginine sensing and transport are distinct in L . donovani , both in axenic promastigotes and intracellular amastigotes . Additionally , the arginine sensor responds to both arginine starvation and sufficiency .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"chemical",
"compounds",
"immunology",
"microbiology",
"parasitic",
"protozoans",
"organic",
"compounds",
"protozoan",
"life",
"cycles",
"developmental",
"biology",
"protozoans",
"leishmania",
"basic",
"amino",
"acids",
"amino",
"acids",
"bioassays",
"and",
"physiological",
"analysis",
"molecular",
"biology",
"techniques",
"promastigotes",
"research",
"and",
"analysis",
"methods",
"transport",
"inhibition",
"assay",
"white",
"blood",
"cells",
"artificial",
"gene",
"amplification",
"and",
"extension",
"animal",
"cells",
"proteins",
"life",
"cycles",
"chemistry",
"leishmania",
"donovani",
"molecular",
"biology",
"amastigotes",
"biochemistry",
"macrophages",
"eukaryota",
"arginine",
"organic",
"chemistry",
"cell",
"biology",
"polymerase",
"chain",
"reaction",
"biology",
"and",
"life",
"sciences",
"protozoology",
"physical",
"sciences",
"cellular",
"types",
"organisms"
] |
2019
|
The arginine sensing and transport binding sites are distinct in the human pathogen Leishmania
|
RNA viruses such as poliovirus have high mutation rates , and a diverse viral population is likely required for full virulence . We previously identified limitations on poliovirus spread after peripheral injection of mice expressing the human poliovirus receptor ( PVR ) , and we hypothesized that the host interferon response may contribute to the viral bottlenecks . Here , we examined poliovirus population bottlenecks in PVR mice and in PVR mice that lack the interferon α/β receptor ( PVR-IFNAR−/− ) , an important component of innate immunity . To monitor population dynamics , we developed a pool of ten marked polioviruses discriminated by a novel hybridization-based assay . Following intramuscular or intraperitoneal injection of the ten-virus pool , a major bottleneck was observed during transit to the brain in PVR mice , but was absent in PVR-IFNAR−/− mice , suggesting that the interferon response was a determinant of the peripheral site-to-brain bottleneck . Since poliovirus infects humans by the fecal–oral route , we tested whether bottlenecks exist after oral inoculation of PVR-IFNAR−/− mice . Despite the lack of a bottleneck following peripheral injection of PVR-IFNAR−/− mice , we identified major bottlenecks in orally inoculated animals , suggesting physical barriers may contribute to the oral bottlenecks . Interestingly , two of the three major bottlenecks we identified were partially overcome by pre-treating mice with dextran sulfate sodium , which damages the colonic epithelium . Overall , we found that viral trafficking from the gut to other body sites , including the CNS , is a very dynamic , stochastic process . We propose that multiple host barriers and the resulting limited poliovirus population diversity may help explain the rare occurrence of viral CNS invasion and paralytic poliomyelitis . These natural host barriers are likely to play a role in limiting the spread of many microbes .
RNA viruses undergo error-prone replication and exist as quasispecies due to the high error rate of RNA-dependent RNA polymerases ( RdRp ) . Within these complex viral populations , genomes can differ by one to many nucleotides resulting from approximately one mutation incorporated per 10 , 000 nucleotides [1] , [2] , [3] . For poliovirus , a mutant virus with a high fidelity RdRp attenuated the virus in mice suggesting that a diverse quasispecies is required for full virulence [4] , [5] . Genetic recombination also contributes to quasispecies diversity , and has been detected in poliovirus isolated from patients with paralytic poliomyelitis [6] . Mutation and genetic recombination may contribute to greater viral population diversity leading to increased virulence [1] , [6] , [7] . Poliovirus is an enterovirus spread by fecal-oral transmission and can cause poliomyelitis in humans . Only ∼1% of people infected with poliovirus develop paralytic poliomyelitis from viral invasion of the central nervous system ( CNS ) [8] , [9] , [10] . Reversion of the live-attenuated Sabin oral polio vaccine ( OPV ) by mutation or recombination occurs rather frequently , but only causes vaccine-associated paralytic poliomyelitis ( VAPP ) in a very small percentage ( 0 . 0001% ) of people that receive OPV [11] , [12] , [13] , [14] . The reason for such a low incidence of paralytic poliomyelitis and VAPP remains unclear . Interestingly , in human VAPP patients , viral isolates found in the CNS are a minor subset of those found in feces , suggesting viral transit from the gut to the CNS may be difficult in humans [15] . Poliovirus receptor ( PVR ) -expressing mice are susceptible to poliovirus via intravenous , intraperitoneal , intracerebral , and intramuscular routes [16] , [17] , [18] . Following intramuscular injection , poliovirus traffics to the CNS by retrograde neuronal transport [19] , [20] . Intravenously injected poliovirus is thought to reach the CNS by the blood route , independent of the presence of PVR [21] . Intraperitoneally injected poliovirus may reach the CNS by blood or neural routes . However , these injection models may not mimic the natural fecal-oral route of infection since PVR mice are not orally susceptible . Recently , PVR mice lacking the interferon α/β receptor ( PVR-IFNAR−/− ) , a major component of innate immunity , demonstrated oral susceptibility to poliovirus [22] , [23] . Oral poliovirus infection in PVR-IFNAR−/− mice resulted in dissemination of virus to many tissues such as esophagus , nasopharynx-associated lymphoid tissue , small intestine , spinal cord , and plasma , as measured by viral titer assay [23] . Viral titers in PVR-IFNAR−/− mice were typically 100 to 10 , 000-fold higher than titers in PVR mice expressing IFNAR . Here , we use PVR-IFNAR−/− mice to measure bottlenecks faced by the viral population during trafficking inside a host . Previously , we identified bottlenecks in PVR mice that limited poliovirus population diversity after peripheral injection by intravenous ( IV ) , intraperitoneal ( IP ) , and intramuscular ( IM ) routes . An artificial quasispecies of four viruses with distinct genomic restriction enzyme site tags were injected , and upon disease onset , brains contained an average of 1 . 7 input viruses suggesting that an intra-host bottleneck was encountered during trafficking to the CNS [24] . Barriers encountered during spread of microbes are common for many pathogens . Bottlenecks have been described for plant RNA viruses [25] , fungi [26] , and bacteria such as Salmonella and Yersinia [27] , [28] , [29] . Interestingly , the picornavirus foot-and-mouth disease virus , may encounter inter-host and intra-host bottlenecks [30] , [31] , [32] . In this study , we introduce a new system for monitoring viral quasispecies trafficking in a murine host orally susceptible to poliovirus . We developed a hybridization-based assay for detection of a population consisting of ten marked viruses . To corroborate our previous work , we examined viral trafficking following peripheral injection of PVR mice vs . PVR-IFNAR−/− mice . In addition , we orally inoculated PVR-IFNAR−/− mice to follow viral trafficking from the initial inoculation site , the oral cavity , to the gastrointestinal ( GI ) tract , blood , and brain . We identified several bottlenecks that limit poliovirus spread following oral inoculation , and found means of overcoming some of these barriers by use of a colon-damaging agent .
Bottlenecks were previously studied using restriction enzyme site markers in the genomes of four distinct viruses; however , this assay was labor intensive and only included four pool members [24] . To overcome these drawbacks , we developed a more streamlined assay based on signature-tagged mutagenesis technology used in bacterial pathogenesis studies [33] . Hybridization-based detection , 96-well format , and an increased number of pool members are advantages of the new assay . The artificial quasispecies pool of ten members was engineered by incorporating silent mutations into the VP3 capsid-coding region of the genome , and oligonucleotide probes were designed for specific recognition of each variant ( Figure 1A , Figure S1 ) . To determine the specificity of the new assay , HeLa cells were infected with individual viruses or a pool of all ten viruses , RNA was isolated after one replication cycle , and RT-PCR products derived from the RNA were blotted on a nylon membrane using a 96-well vacuum manifold . Oligonucleotide probes were 32P-labeled and hybridized to each blotted membrane individually ( Figure 1B ) . Each blot was hybridized with only one labeled probe; therefore , ten blots were performed for each sample . Figure 1C displays the probe hybridization specificity following infection of HeLa cells and probing all samples with each probe . All oligonucleotide probes proved specific for their cognate virus . To ensure the viruses had no detectable growth defects , single-cycle growth curves were performed for each virus and no differences in growth were observed ( Figure S2 ) . Additionally , a serial passage competition experiment was performed by infecting HeLa cells with a mixture of the ten viruses and then passaging the virus mixture five times , followed by assessment of input virus loss over time . All ten viruses were maintained throughout the passages , and therefore , no major growth defects of the marked viruses were detected in vitro ( Figure 1D ) . For each hybridization assay , normalization was performed to eliminate cross-reactivity of nonspecific probes ( Figure 1E ) . Perfectly matched product ( PCR product specific for the probe ) and mismatched products ( all PCR products except for the one specific for the probe ) were loaded on each membrane as controls . The image intensity level of the blots was adjusted until the mismatched product signal became undetectable , revealing only legitimate signals . Validation of the new hybridization assay confirmed the bottleneck effect observed in previous experiments [24] . PVR mice were inoculated with 2×107 plaque-forming units ( PFU ) of a pool of all ten viruses ( 2×106 PFU each; viruses 2 through 11 ) by intramuscular ( IM ) or intracerebral ( IC ) injection . Brains of mice inoculated with 2×107 PFU by the IC route contained most , if not all , input viruses upon disease onset; however , the brains of IM-injected mice contained 10% to 30% of the input viruses ( Figure 2A , 2B ) . For IM-injected mice , all ten viruses were present at the inoculation site , muscle . Brains of PVR mice inoculated by the intraperitoneal ( IP ) route with 1×108 PFU of the ten-virus pool contained only 10% of the input viruses . These experiments validated the new assay and confirmed our previous results [24] . Next , we measured viral population diversity in PVR-IFNAR−/− mice , which are hyper-susceptible to poliovirus [22] , [23] . We hypothesized that innate immunity may contribute to the bottleneck , and therefore , we predicted increased population diversity in the brains of PVR-IFNAR−/− mice . PVR-IFNAR−/− mice were injected intramuscularly with 2×107 PFU of the ten-member pool . As shown in Figure 2A and 2B , the brain bottleneck was greatly diminished in PVR-IFNAR−/− mice , with 40% to 100% of the input viruses detectable in the brain . In fact , the brains of IM-injected PVR-IFNAR−/− mice contain an average of 70% of the input viruses , a result comparable to PVR mice injected IC with 2×107 PFU . Similarly , brains of IP-inoculated PVR-IFNAR−/− mice contained 80% of the input viruses . The diminished bottleneck in PVR-IFNAR−/− mice may be the result of increased peripheral titers in PVR-IFNAR−/− mice , essentially increasing the viral dose , physical barrier differences caused by the lack of IFNAR , such as alteration of neurons or the blood-brain barrier that affect viral trafficking , or , perhaps , a brain-specific IFNα/β response established by the first virus ( es ) to enter the brain contributes to the bottleneck observed in PVR mice . To determine whether the amount of virus entering the brain influences viral diversity , PVR and PVR-IFNAR−/− mice were inoculated by the IC route with a low dose of the ten virus mixture , 2×103 PFU , which corresponds to 200 PFU of each pool member . Using this low input dose , viral diversity was low in the brains of both PVR and PVR-IFNAR−/− mice ( 13% and 24% of input viruses present , respectively ) ( Figure 2B ) . These results suggest that the bottleneck we observe is affected by the quantity of virus entering the brain . The viral bottlenecks we observe are independent of selective advantages possessed by a particular marked virus . Based on a compilation of 479 hybridization signals , all ten viruses were approximately equally represented in a variety of tissues from over 25 mice , although virus 3 showed reduced representation , possibly indicating a slight growth defect ( Figure 1D; Figure S3 ) . However , statistical analysis revealed that none of the viruses , including virus 3 , were significantly under- or over-represented in mouse tissues ( p = 0 . 07 to p = 1 , Student's t test ) . This apparent random sampling of population members was also observed in our previous study [24] . Unlike PVR mice , PVR-IFNAR−/− mice are orally susceptible to poliovirus [22] , [23] . Although the peripheral site-to-brain bottleneck was reduced in PVR-IFNAR−/− mice ( Figure 2A , 2B ) , we sought to determine whether bottlenecks exist following oral inoculation . Because the gut is a complex environment composed of many unique cell types and processes , barriers to viral spread may be encountered in PVR-IFNAR−/− mice despite the hyper-susceptibility of these animals to poliovirus . We orally inoculated PVR-IFNAR−/− mice with 2×107 PFU of a mixture of the ten-member virus pool . Following oral inoculation , PVR-IFNAR−/− mice developed encephalitis rather than paralysis observed in injected mice , and disease onset was delayed , with symptoms developing on days five through ten or later , in agreement with published data [23] . Feces were harvested daily from individual mice , and tissues were collected upon disease onset . Viruses isolated from stomach , small intestine , colon , feces , and blood were amplified for approximately three replication cycles in HeLa cells to increase detection , as the detection limit of the hybridization assay is ∼5 , 000 PFU ( data not shown ) . In vitro amplification does not significantly affect diversity of virus extracted from tissues . For example , in the brain , where viral titers were high enough to perform the hybridization assay with or without amplification , viral diversity was equivalent in amplified and unamplified viral stocks ( data not shown ) . Therefore , in vitro amplification allows detection without significantly altering the composition of the viral population . Three major poliovirus bottlenecks were observed in orally inoculated PVR-IFNAR−/− mice . First , a major bottleneck occurred between the inoculation site ( mouth ) and gut tissues ( Figure 3 ) . Gut tissues were harvested upon disease onset , and lumenal contents were removed . An average of approximately 20% of input viruses were present in the stomach , small intestine , and colon ( Figure 3B ) . Notably , virus was detectable in the stomach late in infection upon disease onset , suggesting that non-input replicating virus was present . These results support the notion that poliovirus is resistant to stomach acid and digestive enzymes , although it is possible that viruses entered the bloodstream and re-seeded organs later in the disease course . Interestingly , viruses found in one GI tract tissue did not always correlate with those detected in other GI tract tissues within the same animal ( e . g . mouse 9-1 , Figure 3 ) . Second , a major bottleneck occurred between the mouth and blood ( Figure 3 ) . It is unclear how poliovirus enters the bloodstream , with evidence supporting upper GI and lower GI routes [34] , [35] , [36] . We found that less than 50% of mice had detectable virus in blood harvested at disease onset , with an average of 9% of input viruses present ( Figure 3 ) . Because it is likely that viremia occurred earlier in the disease course , we assessed viral population diversity in blood from a separate set of animals bled at several time points . Similar to the results obtained by sampling blood at disease onset , less than 60% of day three blood samples contained detectable virus , with an average of 17% of input viruses present ( data not shown ) . Third , a major bottleneck occurred between inoculation site and brain , with an average of 21% of input viruses detected in the brain , harvested upon disease onset ( Figure 3 ) . Surprisingly , viruses found in the brain did not always correlate with those detected in other tissues within the same animal . Interestingly , the timing of disease onset and viral population diversity were associated , such that earlier disease onset correlated with higher diversity . Mice developing symptoms prior to day seven had 3 . 3-fold ( p = 0 . 025 ) more input viruses in the brain than those developing symptoms after day seven , according to mean viral diversity comparison ( Figure 3B ) . Higher diversity was also observed in blood and gut tissues of the early onset mice , with 9-fold higher diversity in blood ( p = 0 . 042 ) , and 2 . 1 to 3 . 3-fold higher diversity in gut tissues ( stomach , small intestine , and colon; p<0 . 05 ) . With the finding that major bottlenecks occurred during viral trafficking from the mouth to other mouse tissues , it became important to determine whether transit through the gut environment is difficult for poliovirus populations . Interestingly , only a minimal bottleneck occurred between inoculation site ( mouth ) and feces ( Figure 3A & 3B ) . For the population diversity assay , we analyzed fecal samples collected at 24 hours post-inoculation because relatively high viral titers were detected at this time . On average , more than 80% of input viruses were detected in feces ( Figure 3B & 6A ) . Many of the mice ( 5/13 ) shed all ten input viruses in feces . Because viral diversity was high in feces , we sought to determine whether the 24-hour fecal samples contained replicated virus , non-replicated/input virus , or both . First , we monitored viral transit time through the GI tract by measuring fecal titers at several time points , and transit time of a dye . Mice were orally inoculated with 2×107 PFU of poliovirus , or Evan's Blue dye as a tracer . Fresh feces were harvested at regular intervals . Viral titers were determined by standard plaque assay using HeLa cells , and transit time of Evan's Blue was determined by scoring the relative dye intensity of fecal samples . As shown in Figure 4A , very high fecal titers were present at 2 hours post-inoculation for some animals . Since this time point is within the eclipse period of the viral replication cycle ( see Figure S2 ) , we presumed that virus shed at 2 hours post-inoculation was input/non-replicated virus . Viral titers remained relatively high from 5–12 hours post-inoculation , and then declined at later time points . This rise and decline of viral titers correlated well with the transit time of Evan's Blue dye through the mouse GI tract ( Figure 4A ) . Although the results from the fecal virus kinetics study suggested that virus shed at early time points is input/non-replicated virus , the presence of replicated virus could not be excluded; therefore , we monitored the transit of light-sensitive poliovirus to directly measure the amount of replicated vs . non-replicated virus present in feces . Poliovirus grown in the presence of neutral red ( NR ) is sensitive to inactivation by light exposure due to dye incorporation and concentration in virions [37] , [38] , [39]; hence , these viruses must be handled in the dark , using a red safety light . Upon replication in the absence of NR , viruses lose this light sensitivity . Therefore , the presence or absence of light-sensitive poliovirus in feces was utilized to monitor whether replication had occurred in the GI tract of orally inoculated mice . In the dark , mice were orally inoculated with 2×107 PFU of light-sensitive NR-poliovirus , and feces were harvested in the light or in the dark . As a control , 6-hour feces harvested in the dark were subjected to titer analysis in light vs . dark conditions . The non-light exposed samples demonstrated high titers: viable virus titers were ∼40% of non-NR poliovirus titers harvested at the 6 hour time point in Figure 4A . We presume that these NR-virus titers were not 100% of the non-NR titers due to intrinsic variability in the animal experiments and/or subtle defects in NR-containing virions . Upon exposure to light , <0 . 1% of the non-NR poliovirus titer was obtained , indicating a very low level of light-insensitive viruses in the population ( Figure 4B , right ) . Fecal samples exposed to light contained negligible viral titers until after 10 hours post-inoculation , suggesting that prior to 10 hours , feces contain input/non-replicated virus ( Figure 4B , left ) . However , at 24 hours post-inoculation , feces contained light-insensitive/replicated virus , although only ∼14% of the non-NR poliovirus titer was obtained . Therefore , the 24-hour fecal samples used for our population diversity analysis contained a mixture of replicated and non-replicated/input virus . We hypothesized that the colonic mucosal epithelium and/or stomach acidity may create barriers that contribute to viral bottlenecks . Therefore , we treated mice with agents that damage the colonic mucosa or neutralize stomach acid and determined the effects on poliovirus titer and diversity . Damage to the colonic mucosa was induced by treating mice with dextran sulfate sodium ( DSS ) in drinking water . DSS directly damages colonic epithelia resulting in ulceration , immune infiltration , and bloody feces [40] , [41] , [42] , [43] . We measured viral titers in feces ( Figure 5A , 5B ) , blood ( Figure 5C ) , and brain ( Figure 5D ) following oral inoculation performed +/− DSS pre-treatment . High-dose ( 5% ) DSS treatment increased 72-hour fecal titers 56-fold ( p = 0 . 000163 ) . Day one fecal titers were 16-fold higher in 5% DSS-treated mice compared to untreated mice . Blood titers for 5% DSS-treated mice were increased 66-fold , and virus was detected in the blood of all 5% DSS-treated mice ( Figure 5C ) compared to untreated mice , where less than 50% of animals had detectable virus in blood . Treatment with 3% DSS did not have an effect on viral titers suggesting that 3% DSS may not induce sufficient damage . We next assessed the role of stomach acid in establishing the poliovirus bottleneck . The mouth-to-feces bottleneck is minor since the majority of the ten input viruses were detected in feces . However , Ohka and colleagues showed that sodium bicarbonate , an acid-neutralizing agent , increased poliovirus titers in a ligated stomach model following oral inoculation of PVR-IFNAR−/− mice [23] . We orally inoculated PVR-IFNAR−/− mice with a virus/5% sodium bicarbonate mixture . Our results revealed no titer differences between sodium bicarbonate-treated and untreated animals ( Figure 5A–D ) . Since 5% DSS-treated poliovirus-infected mice demonstrated increased viral titers , we reasoned that viral population diversity may be increased in these mice . Therefore , we performed the viral population diversity assay for samples from 5% DSS-treated , orally inoculated PVR-IFNAR−/− mice . As expected , viral diversity in feces was high for all mice , regardless of treatment ( Figure 6 ) . Viral diversity in the stomach of 5% DSS-treated mice increased 1 . 8-fold ( p = 0 . 0218 ) , diversity in the small intestine increased 2 . 2-fold ( p = 0 . 00865 ) , and diversity in the colon increased 2 . 8-fold ( p = 0 . 0000497 ) compared to untreated controls ( Figure 6 ) . Additionally , viral diversity in blood increased 3 . 5-fold ( p = 0 . 0101 ) . Interestingly , viral diversity in the brain was unaffected by DSS treatment ( Figure 6A ) . Again , we found that viruses present in the brain do not necessarily correlate with those present in blood or gut tissues ( Figure 6B ) . Viral diversity in tissues of mice treated with 3% DSS or sodium bicarbonate did not differ from untreated mice ( Figure 6A ) . Initially , one might assume that viral titer and viral diversity are linked , with high titer sites containing high population diversity , and vice versa . However , this is not the case , especially when bottlenecks are present [25] , [44] . Figure 7 compares viral titer vs . diversity for feces , blood , and brain viruses from untreated mice orally inoculated with the ten-virus mixture . Fecal samples contained low to moderate titers of ∼5–300 PFU/mg , but contained moderate to high population diversity . Titer and diversity may be linked before a major bottleneck is encountered , as in feces , in which higher titers correlate with higher diversity . These results confirmed that the bottleneck between mouth and feces is minor . Brain samples had the highest titers ( ∼2 , 000–100 , 000 PFU/mg ) , but contained low diversity , which is characteristic of a major bottleneck . We propose that entry into the brain is difficult , but once in the brain , founder viruses undergo robust replication . Blood samples had low to moderate titers ( ∼1–700 PFU/mg ) and low diversity . Therefore , our data confirm that titer and diversity are not linked following bottlenecks .
We have developed a new diversity assay that has allowed us to uncover barriers to viral trafficking that would be missed by standard viral titer assays . Using our hybridization-based assay , we demonstrated bottleneck barriers by monitoring marked polioviruses . We confirmed a previously observed bottleneck between peripheral injection sites and brain ( Figure 2 ) . As before , random sampling was revealed , in which no pool member had an apparent selective advantage over the others ( Figure 1D; Figure S3 ) . The previous assay employed four viruses , and one to three were found in the brain ( average ∼50% ) following IM injection [24] . Here we found that , on average , ∼20% of our ten marked viruses reached the brain , suggesting that this bottleneck was more severe than that previously observed . One possible explanation is that the previous study was performed using ICR-PVR mice [18] , while this study was performed using C57/BL6-PVR mice [45] . Additionally , the observed increase in bottleneck severity could be a result of our increased artificial quasispecies sample size . Because interferons ( IFN ) play an important role in controlling viral infections , prior to this study , we proposed that the IFNα/β response may contribute to viral bottlenecks . In PVR-IFNAR−/− mice , the bottleneck following IM or IP injection was largely absent with an average of 70% or 80% of input viruses detected in the brain , respectively . In fact , direct injection of a large inoculum ( 2×107 PFU ) of the virus pool into the brains of PVR mice resulted in an average of 76% of input viruses detected in the brain , confirming the absence of a major bottleneck in peripherally-injected PVR-IFNAR−/− mice . We propose several possible reasons for the diminished bottleneck in peripherally-injected PVR-IFNAR−/− mice: 1 ) The first viruses to enter the brain in PVR mice established an anti-viral state which limited the spread of later viruses , resulting in a bottleneck effect . The lack of IFNα/β response in PVR-IFNAR−/− mice , therefore , facilitated higher brain diversity . 2 ) Increased peripheral titers in hyper-susceptible PVR-IFNAR−/− mice may have essentially increased the poliovirus dose . This effect could be unique to PVR-IFNAR−/− mice since our previous work determined it was very difficult to overcome the bottleneck by increased dose in PVR mice [24] . 3 ) Physical barriers in PVR-IFNAR−/− mice may have been altered due to lack of the type I IFN environment . Perhaps lack of IFNAR created differences in neurons or the blood-brain barrier that may have contributed to higher viral brain diversity in PVR-IFNAR−/− mice . Importantly , data from our oral inoculation studies demonstrated a bottleneck exists between mouth and brain in PVR-IFNAR−/− mice ( Figure 3 ) . Therefore , the lack of the IFNα/β response in the brain was not the sole cause for the diminished bottleneck in peripherally-injected PVR-IFNAR−/− brains . Additionally , PVR and PVR-IFNAR−/− mice injected by the IC route with a low dose of the virus pool ( 2×103 PFU ) demonstrated comparable low levels of viral diversity in the brain ( 13% and 24% of input viruses , respectively ) . These results suggest that viral diversity in the brain is governed by the amount of virus that enters the brain , and that elevated peripheral titers in injected PVR-IFNAR−/− mice contribute to the elevated viral diversity in the brains of these animals . Following oral inoculation of PVR-IFNAR−/− mice , poliovirus moves through the GI tract without much difficulty . Relatively high amounts of virus were shed in feces , including input/non-replicated viruses and replicated viruses , depending on the sampling time ( Figure 4 ) . Population diversity in feces was relatively high with an overall average of 81% of input viruses present ( Figures 3 and 6A ) , suggesting only a minor bottleneck was encountered during GI lumenal passage . Although we consider this bottleneck minor , it could actually represent the successful passage of just 0 . 025% ( 5×103 PFU ) of the input virus , which would still allow detection of all pool members in our system . Regardless , this mouth-to-feces bottleneck was minor in comparison to other bottlenecks we observed . Our experiments identified three major bottlenecks following oral inoculation of PVR-IFNAR−/− mice: mouth-to-gut tissues , mouth-to-blood , and mouth-to-brain . First , a major bottleneck existed between the mouth and gut tissues . Of the ten viruses , an average of 16% of input viruses were present in the stomach , and 19% of input viruses were present in the small intestine and the colon ( Figures 3 and 6 ) . We presume that virus must be replicating in these tissues to be detected late in infection when the tissues were harvested ( day 5–10 ) . However , gut tissues could have been re-seeded by virus in the blood . We identified a second bottleneck between mouth and blood . Blood titers were moderate , but diversity was very low ( avg . = 9% of input viruses ) ( Figures 3 , 6 , and 7 ) . We are uncertain how the virus is traveling from the inoculation site into the blood , but possibilities include drainage from lymph , mucosal passage to the blood , and entrance into the bloodstream at sites of mucosal micro-damage . Viruses may have entered the bloodstream early in infection [34] , [46] , [47] . Third , a prominent bottleneck existed between the mouth and the brain . Viral trafficking between the mouth and brain could have occurred through blood or neural routes . Historically , poliovirus invasion of the brain has been presumed to occur through the blood route because neutralizing antibodies are protective and IV-injected radiolabeled virions readily enter the murine brain [21] , [48] , [49] . However , viral trafficking in neurons may also occur and contribute to pathogenesis [10] , [19] , [20] , [50] . Surprisingly , 93% of viruses found in the brain were present in gut tissues of a given mouse , but only 35% were detected in blood ( Figure 3B ) . This suggests that a gut tissue-to-brain pathway was involved in viral spread . Virus may have entered the blood from gut tissues and trafficked to the brain , or virus may have infected neurons associated with the GI tract and reached the brain by retrograde transport . Trafficking via neurons has been demonstrated by sciatic nerve transection experiments following poliovirus infection [19] , [20] . Although our data did not definitively discriminate between blood and neural routes , our data did show that absolute match of blood and brain viruses was rare . In some instances , there was no overlap between viruses present in the brain and blood ( Figure 3B , 9-6; Figure 5 , 9-1 ) . These results indicated that virus may enter the brain by a non-hematogenous route , such as neurons , that low-abundance viruses in blood seeded the brain , or that virus found in the blood at disease onset differed from the virus in the blood at earlier time-points . Interestingly , the blood/brain virus mismatch was confirmed in an experiment where blood was collected at day three post-inoculation and upon disease onset , and then blood diversity was compared with brain diversity . This experiment revealed that only 44% of viruses found in the brain are present in blood at day three post-infection , suggesting that not all viruses may spread to the brain via a blood route ( data not shown ) . Aside from possible GI neuronal trafficking , it is likely that virus moves into and out of blood throughout infection by seeding other tissues with subsequent re-seeding of the blood . In humans , it is thought that a primary asymptomatic viremia may seed tissues , with a subsequent secondary viremia contributing to minor or major illness , which can lead to CNS invasion and paralytic poliomyelitis [34] , [35] , [36] . Our results suggested that disease onset and viral diversity are linked . Earlier disease onset correlated with greater viral diversity in the gut tissues , blood , and brain . Mice that developed symptoms before day seven had 3 . 3-fold ( p = 0 . 025 ) higher brain diversity than those that developed symptoms later ( Figure 3B ) . These early onset mice also had higher blood diversity ( 9-fold , p = 0 . 042 ) and gut diversity ( 2 . 1–3 . 3 fold , stomach: p<0 . 05 ) . Greater diversity upon earlier onset was not simply due to a tissue sampling time bias , because a separate study demonstrated very low population diversity in tissues harvested on days one and three post-inoculation ( Figure S4 ) . There are several possible explanations for the correlation of disease onset and viral diversity . First , since high population diversity and virulence are linked [4] , [5] , higher viral diversity may contribute to faster disease progression . Second , some component of host immunity may have developed later in infection , which limited viral replication , and ultimately , viral diversity . Interestingly , two of the three major bottlenecks could be overcome by pre-treating the mice with a colonic epithelial-damaging agent , DSS . The first ( mouth-to-gut tissues ) and second bottlenecks ( mouth-to-blood ) were affected by DSS treatment: gut tissue and blood diversity increased ∼2–3-fold and 3 . 5-fold , respectively . Additionally , blood titers increased 66-fold in the presence of colonic damage . Importantly , the mouth-to-brain bottleneck was unchanged in DSS-treated mice compared to untreated mice . These results suggest that either virus trafficked to the brain via a non-blood route , which was unaffected by DSS treatment , or virus trafficked to the brain via a blood route , but spread to the brain was limited by another barrier , such as the blood-brain barrier . Viral titer and diversity were not linked after a bottleneck was encountered ( Figure 7 ) . By monitoring diversity , we uncovered limitations on viral trafficking that would be missed by viral titer analysis . For example , blood and fecal titers were similar; therefore , one might conclude that transit from the gut to blood was not difficult . Our assay allowed us to conclude that a major bottleneck exists since blood diversity was low . We presume that virus was replicating in blood and/or other tissues that seed blood , thus increasing the blood titer post-bottleneck encounter , resulting in founder effects . We found that viral population trafficking was a very dynamic , stochastic process . Using virus 2 as an example , a given virus might be present in all tissues ( Figure 3 , 9-1 ) , in colon and feces only ( Figure 3 , 9-2 ) , in brain and feces only ( Figure 3 , 9-6 ) , or in other differing combinations . Similar random trafficking patterns have been observed in several microbial systems , including animal and plant viruses , bacteria , and fungi [25] , [26] , [27] , [28] , [29] , [30] , [31] , [32] . For highly mutable RNA viruses , host barriers likely play an important role in shaping viral populations and determining virulence [51] . The random distribution of viral populations makes predicting VAPP ( vaccine-associated paralytic poliomyelitis ) impossible , because a viral isolate from the CNS of one person may not invade the CNS of another due to bottleneck effects and stochastic trafficking . Notably , in human VAPP patients , fecal virus does not always correlate with virus found in the CNS [15] . Perhaps physical barrier disruption and/or a defective innate immune response increased susceptibility to inadvertent poliovirus CNS invasion in individuals afflicted with paralytic poliomyelitis . We have shown that this artificial quasispecies system mimics the stochastic poliovirus trafficking observed in humans , and can be used to understand RNA virus population dynamics in an infected host .
The ten viral plasmids ( 2 through 11 ) were made using silent site-directed mutagenesis of the Mahoney serotype 1 viral cDNA clone beginning with nucleotide 2425 and ending at 2443 ( Figure 1A ) [52] . Two unique silent restriction sites were added , Bgl II at nucleotide 5601 and Mlu I at nucleotide 7550 , in order to facilitate cloning . Each PCR-generated region was confirmed by sequencing ( Sequencing Core , UT Southwestern Medical Center , Dallas , TX ) . All poliovirus work was done in WHO-approved elevated BSL2/poliovirus conditions . Cell culture infections and propagation of virus was performed from a single poliovirus plaque using HeLa cells grown in Dulbecco's modified Eagle's medium with 10% calf serum as previously described [53] . For the viral serial passage experiment ( Figure 1D ) , the ten viruses were combined at equivalent amounts and single-cycle infections beginning with a MOI of 0 . 1 , were performed as described [4] . Virus stocks were titered using plaque assays in HeLa cells as previously described [53] . A neutral red ( NR ) -poliovirus stock was prepared by infecting HeLa cells with wild-type poliovirus in the presence of 10 ug/ml neutral red ( Sigma ) in the dark , using a red safety light [37] , [38] , [39] . NR-poliovirus stocks were light inactivated by exposure to a fluorescent light bulb at a distance of 3 inches for 10 minutes . The ratio of light-insensitive to light-sensitive PFU in the NR-poliovirus stock was 1 to 1 . 27×106 . All animal work was performed according to protocols approved by the UT Southwestern Medical Center IACUC . C57/BL6 PVR-Tg21 ( PVR ) mice and C57/BL6 PVR-IFNAR−/− ( PVR-IFNAR−/− ) mice were obtained from S . Koike ( Tokyo , Japan ) , and maintained in specific pathogen free conditions [22] . Intramuscular ( 50 µl volume ) and intracerebral ( 15 µl volume ) injections were done as previously described [4] using 2×107 PFU total ( 2×106 PFU of each of the 10 viruses ) , or 2×103 PFU total for low-dose IC injections . For intraperitoneal injections , 1×108 PFU total of the 10 viruses were injected in a volume of 50 µl . It should be noted that inocula for all experiments in this study were based on viral titers obtained using HeLa cells . We have shown previously that poliovirus titers in PVR-derived mouse embryo fibroblasts ( PVR-MEFs ) are approximately 300-fold lower than those obtained in HeLa cells [4] . Therefore , in terms of poliovirus titers in mouse cells , mice were actually inoculated with 6 . 67×104 PFU for the “2×107 PFU” inoculations . Oral inoculations were performed by dispensing 15 µl of virus , by pipette tip , in the mouth . Each mouse was euthanized at first signs of disease , which included encephalitis , ruffled fur , lethargy , and paralysis . In our experience , once symptoms develop , the mice die within a day . For DSS treatments , mice were pre-treated with DSS ( molecular weight 36 , 000–50 , 000; MP Biomedicals LLC , Solon , OH ) in their drinking water prior to oral inoculation [40] . Mice receiving 3% DSS were pre-treated for three days , and mice receiving 5% DSS were pre-treated for five days . Once infections were performed , the mice were provided with regular drinking water for the course of the experiment . Sodium bicarbonate was added to virus to make 5% mixtures immediately prior to oral infections [23] . Mice were housed in individual cages and feces were collected at 24-hour intervals with subsequent bedding changes . A combination of moist , freshly acquired feces and dry feces were combined to generate the fecal samples for the population diversity assay . For kinetics of viral shedding experiments ( Figure 4 ) , fresh feces were harvested from each mouse . For Evan's Blue dye transit experiments ( Figure 4A ) , feces were weighed , resuspended in 6 volumes of PBS , freeze-thawed three times , and samples were centrifuged at 13 , 000 rpm for 1 minute . “Evan's Blue Score” was determined by assessing the level of blue color in the feces: slightly blue = 1 , light blue = 2 , moderate blue = 3 , dark blue = 4 , intense blue = 5 . Upon euthanasia , blood , stomach , small intestine , colon , and brain were harvested and stored at −80°C prior to use . During tissue harvests , lumenal contents were removed from gut tissues . Tissues ( brain , stomach , small intestine , colon ) were homogenized under liquid nitrogen using a mortar and pestle . For brain RNA extractions , 1 ml of TRIZOL ( Invitrogen , Carlsbad , CA ) was added to approximately 300mg of tissue , and extractions and RT-PCR were performed as previously described [4] . BN2 antisense primer 5′ ATGCTTTCAAGCATCTGACCTAACC 3′ and NdeI sense primer 5′ AAACTGTTGGTGTCATATGCGCCTCCTGGAG 3′ were used for RT-PCR and PCR . To amplify virus from tissues , homogenized tissues were weighed and resuspended in 3 volumes of PBS+ ( 1× PBS supplemented with 100 µg/ml MgCl2 and CaCl2 ) , and freeze-thawed 3 times . Feces were weighed , resuspended in PBS , and freeze-thawed three times . Each tissue slurry was dounce homogenized and centrifuged at 13 , 000 rpm for 1 minute , and supernatants were kept as virus stocks . To limit microbial contamination , virus from gut samples ( stomach , small intestine , colon , and feces ) were chloroform extracted by adding 1/10 volume of chloroform , centrifuged at 13 , 000 rpm for 2 minutes , and the supernatant was kept as the virus stock . Virus was amplified for 2–3 rounds of replication ( 12–16 hours ) at 37°C in HeLa cells and the cells were harvested , resuspended in 50–100 µl of PBS+ , freeze-thawed , and kept as amplified virus stock . Half of the amplified virus stock was added to 1ml of TRIZOL for RNA extractions and RT-PCR . PCR was performed in quadruplicate and products were combined before they were run on an agarose gel and quantitated by standards of known concentrations . These concentrations were used to normalize the amount of PCR product blotted to 50–100 ng of PCR product for each sample . DNA was blotted onto Hybond N+ membranes ( GE Healthcare , Buckinghamshire , UK ) using a 96-well vacuum manifold , and membranes were pre-hybridized and hybridized following standard procedures [54] . Optimal hybridization annealing temperature was empirically determined to be 59°C ( data not shown ) . Probes were made by kinase treatment of specific primers ( see Figure S1 ) with [γ-32P] ATP and excess nucleotides were removed with the Qiagen Nucleotide Removal kit ( Qiagen , Valencia , CA ) [4] . Membranes were exposed to PhosphorImager screens and scanned by Stormscan . Scanned blots were normalized by comparison of equivalently loaded products of perfectly matched PCR product to probe or mismatched PCR products to probe . Blot image intensities were adjusted such that any apparent mismatch dot was no longer visible , thus eliminating the minimal level of cross-reactivity of the probes with non-matched PCR products ( Figure 1E ) . The Genbank ( http://www . ncbi . nlm . nih . gov/entrez ) accession number for serotype 1 poliovirus ( Mahoney strain ) is NC002058 .
|
RNA viruses are highly error prone , and can use their replication infidelity to adapt to complex environments within an infected host . However , viral populations may experience bottlenecks , which limit their diversity and potentially reduce their virulence . We hypothesized that natural barriers may limit the spread of RNA viruses within an infected host . To test this hypothesis , we engineered a pool of ten marked polioviruses identifiable by a novel assay , infected susceptible mice by injection or oral inoculation , and determined the percentage of the ten viruses that successfully spread to various body sites , including the brain . We found that , on average , only 10%–20% of the input viruses were found in most tissues , suggesting that barriers prevented the spread of the whole population . The importance of one such physical barrier , the colonic epithelium , was demonstrated in experiments where the colon was damaged prior to oral inoculation . Under these conditions , 30%–50% of the input viruses successfully spread to various body sites . We propose that host barriers limit viral spread , and this could possibly explain the rare incidence of paralytic poliomyelitis due to central nervous system invasion .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"virology/virus",
"evolution",
"and",
"symbiosis",
"gastroenterology",
"and",
"hepatology/gastrointestinal",
"infections",
"virology/animal",
"models",
"of",
"infection",
"virology",
"virology/mechanisms",
"of",
"resistance",
"and",
"susceptibility,",
"including",
"host",
"genetics",
"virology/host",
"antiviral",
"responses"
] |
2008
|
Multiple Host Barriers Restrict Poliovirus Trafficking in Mice
|
Nonenveloped viruses undergo conformational changes that enable them to bind to , disrupt , and penetrate a biological membrane leading to successful infection . We assessed whether cytosolic factors play any role in the endoplasmic reticulum ( ER ) membrane penetration of the nonenveloped SV40 . We find the cytosolic SGTA-Hsc70 complex interacts with the ER transmembrane J-proteins DnaJB14 ( B14 ) and DnaJB12 ( B12 ) , two cellular factors previously implicated in SV40 infection . SGTA binds directly to SV40 and completes ER membrane penetration . During ER-to-cytosol transport of SV40 , SGTA disengages from B14 and B12 . Concomitant with this , SV40 triggers B14 and B12 to reorganize into discrete foci within the ER membrane . B14 must retain its ability to form foci and interact with SGTA-Hsc70 to promote SV40 infection . Our results identify a novel role for a cytosolic chaperone in the membrane penetration of a nonenveloped virus and raise the possibility that the SV40-induced foci represent cytosol entry sites .
Nonenveloped viruses must penetrate a biological membrane to infect cells . As they lack a surrounding lipid bilayer , membrane penetration by nonenveloped viruses must be fundamentally different from enveloped viruses , which normally gain access to the host cell by membrane fusion . Although the precise membrane transport mechanism for nonenveloped viruses is not entirely clear , a general principle is emerging . These viruses enter host cells by endocytic internalization in order to arrive at a precise cellular environment necessary for productive infection [1] . Upon reaching this proper environment , important conformational changes are induced by specific cellular triggers including low pH , proteases , or chaperone activities [2] . These conformational changes in turn generate a hydrophobic viral particle or cause the release of a lytic peptide hidden in the intact virus . Engagement of the hydrophobic particle or lytic peptide with the limiting membrane disrupts the membrane integrity and initiates membrane penetration . For example , the nonenveloped reovirus , parvovirus , and adenovirus become internalized and traffic to endosomes where the low pH or proteases trigger viral conformational changes that allow them to penetrate the endosomal membrane [3]–[6] . In these cases , membrane penetration is thought to involve virus-induced pore formation or disruption of overall membrane integrity . Currently , absent in this model is a role for any cytosolic factors directly influencing membrane penetration . Polyomaviruses are unique among nonenveloped viruses in that they traffic beyond the endosomal system to reach the endoplasmic reticulum ( ER ) for membrane penetration [7]–[13] . This virus family consists of a growing list of important human polyomaviruses known to cause devastating diseases in immunocompromised individuals [14] , [15] . Simian virus 40 ( SV40 ) has traditionally served as an excellent model member of this family; it has genetic and structural similarity to human polyomaviruses , yet is easy to propagate and study in cells . To cause infection , SV40 engages the ganglioside receptor GM1 at the cell surface to initiate internalization [16] , [17] . Caveolae-dependent endocytosis brings SV40 particles attached to lipid rafts into the cell where they travel through endosomes before being sorted to the ER [7] , [12] . Once inside the ER lumen , SV40 is faced with the task of penetrating the ER membrane to reach the cytosol prior to nuclear import [18] , [19] . In the nucleus , transcription and replication of the viral genome are initiated , leading to lytic infection or cellular transformation . The ER provides an ideal environment for inducing important conformational changes to the structure of SV40 . The outer surface of each viral particle contains 360 copies of the major coat protein VP1 arranged as 72 pentamers . A single hydrophobic minor coat protein VP2 or VP3 resides beneath each VP1 pentamer [20] . VP1 molecules are stabilized by interpentameric disulfide bonds , with bound calcium ions and hydrophobic interactions providing additional capsid support [21] , [22] . Protein disulfide isomerase ( PDI ) -family members appear to be broadly important during entry of polyomaviruses for either their ability to disrupt viral disulfide bonds by using their redox/isomerase activities , or to impart conformational changes by using their chaperone functions [10] , [23]–[26] . PDI proteins exert these activities on SV40 , likely in concert with other ER factors , resulting in VP2 exposure [9] , [24] , [27] , [28] . Due to its hydrophobic N-terminus , exposure of VP2 renders the virus itself hydrophobic . As a result , the virus binds and integrates into the ER membrane to initiate membrane penetration [27] , [29] , [30] . In the next stage of the membrane penetration process , a critical Glu residue in VP2's N-terminus embedded in the ER membrane is hypothesized to act as a charged irregularity in the ER membrane and recruits cellular factors involved in ER-associated degradation ( ERAD ) [29] . The ERAD pathway utilizes large multi-protein complexes to eliminate misfolded or incorrectly assembled ER proteins by facilitating their retro-translocation into the cytosol for degradation by the ubiquitin-proteasome system [31] , [32] . SV40 and other polyomaviruses co-opt several ERAD membrane components including the J-proteins DnaJB14 ( B14 ) and DnaJB12 ( B12 ) to reach the cytosol and infect cells [10] , [11] , [23] , [29] , [33]–[35] . B14 and B12 both span the ER membrane once and display their functional J domain in the cytosol [36]–[38] . By virtue of this domain , J-proteins stimulate the ATPase activity of Hsp70 chaperones to promote substrate-Hsp70 interaction [39] . However , the precise mechanism by which these membrane components facilitate ER-to-cytosol transport of a large viral particle [40] is not clear . Whether any cytosolic components provide the driving force to extract the hydrophobic ER membrane-embedded SV40 into the cytosol is also unknown . In this context , the cytosolic ATPase p97 ( also called VCP ) involved in the mobilization of ERAD substrates into the cytosol [41] was shown to be dispensable for SV40 infection [29] . Additionally , while chemical inhibition of the cytosolic proteasome perturbs infection of SV40 and other polyomaviruses [11] , [23] , [34] , [40] , this effect may be indirect [29] . Thus , the potential roles of cytosolic factors that directly promote membrane penetration of polyomaviruses , as well as other nonenveloped viruses , remain enigmatic . Here , we demonstrate that the cytosolic chaperone SGTA ( small glutamine-rich tetratricopeptide repeat-containing protein α ) is critical for transport of SV40 from the ER membrane into the cytosol . SGTA associates with a chaperone complex containing B14 and B12 ( B14-B12 ) at the ER membrane and is therefore positioned to act at the site of membrane penetration . Using a combination of cell-based and biochemical assays , we found that SGTA binds directly to SV40 and promotes its ER membrane penetration and infection . During membrane penetration , SGTA is released from the B14-B12 complex , suggesting SV40 alters localization or structural characteristics of these factors . Consistent with this idea , we found that SV40 causes the B14-B12 complex to reorganize into discrete foci in the ER membrane . As formation of these foci coincides temporally with SV40's cytosolic arrival , they may represent ER exit sites where SGTA engages the virus to complete membrane penetration .
To clarify how B14 and B12 promote SV40 ER membrane penetration [33] , we first characterized their biochemical properties . Gel-filtration analysis of detergent-solubilized cell extracts followed by subsequent immunoblotting of the individual fractions demonstrated that a substantial pool of B14 eluted in high molecular weight fractions ( >150 kDa ) , suggesting B14 is part of a complex larger than its monomeric size of 42 kDa ( Figure 1A ) . Probing for Hrd1 and ERp29 confirmed the fractionation of large and small complexes , respectively . B12 co-fractionated identically with B14 , raising the possibility that these proteins interact with each other . Co-immunoprecipitation experiments revealed that endogenous B14 and B12 bind to each other with high efficiency in both HeLa and CV-1 cells ( Figure 1B and 1C ) . These data demonstrate that these J-proteins interact with each other and are likely part of a stable membrane complex . To identify other components of the B14 complex , we used an unbiased strategy of immunoprecipitation followed by mass spectrometry ( MS ) . We constructed 293T cells that allowed for tetracycline-inducible expression of B14-3xFLAG . To minimize overexpression artifacts , B14-3xFLAG expression was maintained at a level similar to endogenous B14 by providing a low concentration of tetracycline ( Figure 2A ) . Under this condition , cells were lysed with a low concentration of digitonin to maintain protein-protein interactions during immunoprecipitation with anti-FLAG conjugated agarose beads . To control for nonspecific binding to the agarose beads , an identical lysate was incubated with anti-FLAG conjugated agarose beads pre-blocked with 3xFLAG peptide . After washing and elution by addition of 3xFLAG peptide , the eluted material was concentrated and subjected to SDS-PAGE followed by silver staining ( Figure 2B ) . Co-immunoprecipitated proteins reproducibly observed by silver staining were excised and analyzed by MS . Not surprisingly , a band migrating at approximately 72 kDa was identified as Hsc70 , a common interacting partner of J-proteins . Interestingly , another band corresponding to approximately 38 kDa was identified as the cytosolic chaperone SGTA . We focused our study on this protein as it was recently implicated in the ERAD pathway [42] . While several components of the SMN complex involved in spliceosomal snRNP assembly and pre-mRNA processing were also identified during MS analyses , they remain to be tested as authentic B14 binding partners . Eluted samples prepared as in Figure 2B were immunoblotted with specific antibodies to confirm the presence of SGTA , Hsc70 , and B14-3xFLAG ( Figure 2C ) . Importantly , when endogenous SGTA was immunoprecipitated from standard 293T or CV-1 cells , endogenous B14 but not the abundant membrane protein BAP31 was detected in the precipitate ( Figure 2D , left and right panels ) . As B12 complexes with B14 ( Figure 1 ) , we asked whether endogenous B12 also binds SGTA . Indeed , B12 also co-precipitated with SGTA . ( Figure 2D , right panels ) . We next investigated whether the interaction between SGTA and B14-B12 was direct or mediated by a previously reported SGTA binding partner . SGTA has been observed to interact with Hsc70 as a co-chaperone [43] , [44] . Additionally , a more recent study demonstrated that SGTA binds to the cytosolic Bag6 ( BAT3 , scythe ) -Ubl4a-Trc35 complex via Ubl4a [42] . We therefore tested whether previously characterized SGTA mutants defective in their ability to bind to either Hsc70 ( K160E/R164E ) or Ubl4a ( D27R/E30R ) [42] could interact with B14 . CV-1 cells were transfected to express FLAG-tagged WT or a mutant form of SGTA . Lysates derived from these cells were immunoprecipitated with anti-FLAG conjugated agarose beads . Endogenous B14 was observed to precipitate with WT-SGTA and the Ubl4a-binding defective mutant ( D27R/E30R ) ( Figure 2E ) . By contrast , the Hsc70-binding defective mutant ( K160E/R164E ) was unable to interact with B14 , despite substantially more mutant in the precipitation when compared to WT SGTA ( Figure 2E ) . This interaction was specific to B14's cytosolic J domain as ERdj5 , a lumenal J-protein , did not precipitate with any form of SGTA ( Figure 2E ) . We performed the converse analysis by transfecting FLAG-tagged WT B14 or a mutant B14 defective in coupling to Hsc70 ( H136Q ) [37] . WT B14 but not H136Q B14 precipitated endogenous SGTA ( Figure 2F ) . Together these data demonstrate SGTA interacts with the B14-B12 complex in a Hsc70-dependent manner . B14 and B12 are two key ER membrane components crucial for SV40 and human BK polyomavirus ( BKPyV ) infection [33] . Since our findings revealed SGTA is a binding partner of the B14-B12 complex , we first asked whether SGTA is also important for SV40 infection . Expression of the virally-encoded large T antigen protein ( TAg ) in the host nucleus reflects successful viral infection . We monitored infection in this way by scoring cells for the presence or absence of TAg using immunofluorescence microscopy . Three distinct siRNA oligonucleotides effectively downregulated SGTA expression in CV-1 cells by over 80% ( Figure 3A; quantified in 3B ) . Using a low M . O . I . ( i . e . 0 . 5 ) , silencing SGTA robustly inhibited SV40 infection by 70–80% when compared to the scrambled control ( Figure 3C , black bars ) . When a higher M . O . I . ( i . e . 5 ) was used , SGTA knockdown decreased infection by approximately 40–50% ( Figure 3C , white bars ) . A striking decrease in TAg expression was also evident by immunoblot analysis of whole cell extracts derived from infected cells ( Figure 3D ) . Similarly , these knockdown conditions markedly blocked the expression of BKPyV TAg as assessed by immunofluorescence and immunoblot analyses ( Figure 3E and 3F ) . We conclude that SGTA plays an important role during SV40 and BKPyV infection . SGTA is involved in multiple aspects of protein quality control [42] , [45] . Recent evidence suggests one function of SGTA is to facilitate ERAD substrate loading on Bag6 , a holdase that prevents substrate aggregation prior to proteasomal degradation [42] , [46] . Based on these findings , we tested whether Bag6 was important for SV40 infection and found that Bag6 knockdown did not block expression of SV40 TAg ( Figure S1 ) , indicating that SGTA promotes SV40 infection independent of Bag6 . As SGTA is a cytosolic chaperone complexed with the ER membrane J-proteins B14 and B12 , we asked whether this chaperone facilitates the arrival of SV40 into the cytosol from the ER . Our laboratory previously developed a cell-based assay to monitor SV40 ER-to-cytosol transport [40] . A similar assay has also been reported [29] . Briefly , cells infected with SV40 are harvested , selectively permeabilized with a low digitonin concentration , and centrifuged to generate a supernatant and a pellet fraction; the supernatant harbors cytosolic material while the pellet contains intracellular membranes including the ER . These fractions are subsequently analyzed by immunoblotting for fractionation markers and VP1 . VP1 in the supernatant therefore represents SV40 that reached the cytosol , and VP1 in the pellet reflects virus within intracellular organelles . We applied this assay to cells transfected with scrambled or SGTA siRNAs . The cytosolic marker Hsp90 was predominantly detected in the supernatant whereas the ER lumenal marker BiP was found exclusively in the pellet ( Figure 4A ) , confirming the integrity of the fractionation procedure . Strikingly , the VP1 level in the supernatant was significantly reduced in cells with SGTA downregulated when compared to the scrambled control ( Figure 4A; quantified in 4B ) . We note that the severity of the ER-to-cytosol transport defect , while consistent with the robust block in infection when cells were incubated with a low M . O . I . ( Figure 3B , black bar ) , appears to be greater than the perturbation in infection when a high M . O . I . was used ( Figure 3B , white bars ) . This difference is likely attributed to the remaining low amounts of virus in the cytosol that are sufficient to reach the nucleus to express TAg [47] . We also assessed whether downregulating SGTA affects arrival of SV40 to the ER . Successful SV40 ER arrival can be monitored by examining the amount of Triton X-100 soluble VP1 present in the pellet ( i . e . membrane fraction ) . We previously found that only viruses which arrive in the ER are released from Triton X-100 insoluble lipid rafts into the ER lumen whose content can be extracted by Triton X-100 [40] . As expected , when cells were infected in the presence of brefeldin A ( BFA ) , a drug that blocks COPI-dependent retrograde transport from the plasma membrane to the ER , there was a robust block in ER arrival of SV40 ( Figure 4C; quantified in 4D ) [13] . However , no detectable loss of SV40 ER arrival was apparent when SGTA was downregulated ( Figure 4C; quantified in 4D ) . Thus SGTA does not significantly regulate retrograde trafficking of SV40 to the ER . Given that a pool of SGTA resides on the cytosolic side of the ER membrane due to its interaction with B14 and B12 , we propose this factor facilitates SV40 infection primarily by completing the membrane penetration step during entry . We assessed whether SGTA knockdown causes massive ER stress , which may explain the observed disruption of SV40 transport across the ER membrane . Monitoring the splicing of XBP1 mRNA ( which encodes a stress-responsive transcription factor ) is a sensitive method for detecting ER stress induction . When SGTA was silenced , we detected modest splicing of this transcription factor mRNA ( SGTA siRNA #2 ) , in contrast to cells exposed to the chemical ER stress inducer dithiothreitol ( DTT ) ( Figure 4E ) . This finding indicates that significant ER stress was not triggered by silencing SGTA . While this result varies slightly from a recent report in which 293T cells expressing SGTA shRNA exhibited a strong ER stress response [42] , this difference may be due to our use of a transient knockdown system in CV-1 cells . Cholera toxin ( CT ) is another pathogenic factor that becomes internalized and traffics to the ER . From the ER , the catalytic A1 subunit of CT ( CTA1 ) retro-translocates to the cytosol to induce cytotoxicity . In fact , CTA1 retro-translocation requires some of the same ERAD machinery co-opted by SV40 [48]–[50] . Unlike SV40 , we found that SGTA knockdown did not inhibit transport of CTA1 to the cytosol ( Figure 4F ) , whereas BFA treatment potently blocked CTA1 retro-translocation ( Figure 4F ) , as previously reported [50] . These data indicate that defective SV40 ER-to-cytosol transport and infection caused by SGTA silencing was not due to poor cell health or general ER dysfunction . If SGTA extracts SV40 from the ER membrane into the cytosol , we reasoned that SGTA itself might dissociate from the B14-B12 complex during virus transport . We first monitored SGTA-B14 interaction using co-immunoprecipitation assays during SV40 entry . Cells were uninfected or infected for 2 or 8 h , harvested , followed by cross-linking of the intact cells . Similar to previous experiments ( Figure 2D ) , when endogenous SGTA was immunoprecipitated , a significant amount of B14 was detected ( Figure 5A ) . This interaction remained intact at 2 h post infection ( h . p . i . ) ( Figure 2D ) , a time point when SV40 particles are largely present in endosomes and have not yet reached the ER [7] , [12] . By contrast , a stable SGTA-B14 interaction was completely lost at 8 h . p . i . ( Figure 5A ) when SV40 has reached the ER and initiated ER-to-cytosol transport [29] , [40] . When Hsc70 was immunoprecipitated under similar conditions , a loss of stable B14 interaction was also detected ( Figure 5B ) . However , the interaction between Hsc70 and SGTA was mostly preserved . These data indicate that during time points of entry where SV40 is undergoing membrane penetration , SGTA and Hsc70 are being released from the B14-B12 complex at the ER membrane . This phenomenon held true in CV-1 cells transfected to express FLAG-tagged GFP or FLAG-tagged SGTA ( Figure 5C ) ; when cells uninfected or infected for 16 h were lysed and immunoprecipitated using anti-FLAG agarose beads , stable interactions between transfected SGTA and endogenous B14 as well as B12 remained disrupted at this later time point . While membrane penetration is observed to begin between 6 and 8 h . p . i . , this process likely continues at later times due to the asynchronous nature of SV40 entry [7] , [51] . Our time point analyses suggested that SV40 penetration across the ER membrane is required to release SGTA from the B14-B12 complex . To further test this hypothesis , we utilized a mutant SV40 lacking VP2 ( i . e . ΔVP2 SV40 ) . This mutant virus becomes internalized but fails to penetrate the ER membrane to reach the cytosol [29] . When compared to WT SV40 , ΔVP2 SV40 did not disrupt the interaction of SGTA and B14-B12 ( Figure 5D ) . Together , our findings indicate that the initiation of ER membrane penetration by SV40 is required for release of SGTA from the B14-B12 complex . We assessed whether this dissociation event was general for ER bound cytosolic chaperones or specific to SGTA . The cytosolic p97 chaperone is well documented to link to ERAD complexes at the ER membrane and facilitate retro-translocation of ERAD substrates [41] , [52] , [53] . In cells infected with SV40 , p97's interaction with the ERAD membrane component Derlin-1 was unchanged ( Figure 5E ) . This result is consistent with the recent finding that p97 is not required for SV40 entry and infection [29] . SV40 infection also did not disrupt endogenous B14-B12 interaction ( Figure 5E ) . We conclude SV40 specifically triggers SGTA to be released from the B14-B12 complex without globally disrupting the connection of cytosolic factors to the ER membrane . An additional explanation to account for the release of SGTA from B14-B12 is that SGTA disengages from B14-B12 in order to engage SV40 as it becomes exposed in the cytosol . To test this possibility , CV-1 cells were transfected to express FLAG-tagged SGTA and infected for 8 h followed by cross-linking and immunoprecipitation with anti-FLAG agarose beads . Indeed , VP1 was detected in the immunoprecipitate of infected cells ( Figure 5F ) . By contrast , VP1 was not detected in infected cells treated with BFA , which blocks ER arrival and subsequent transport to the cytosol ( Figure 5F ) . Thus , SGTA binds to SV40 upon cytosolic arrival . We next investigated whether this SV40-SGTA interaction required the Hsc70-dependent localization of SGTA to the B14-B12 complex . When cells transfected with FLAG-tagged GFP , WT SGTA or the Hsc70-binding defective SGTA mutant ( K160E/R164E ) were infected and subjected to immunoprecipitation , VP1 was observed to co-precipitate only with WT SGTA and not GFP or K160E/R164E SGTA ( Figure 5G ) . This result suggests that localization of SGTA to the B14-B12 complex via Hsc70 is required for SGTA to engage the virus during entry . Hsc70 has been reported to associate with polyomaviruses in vitro and in cells during both entry and de novo capsid assembly [54] , [55] . To assess whether SGTA binds to SV40 directly or requires Hsc70 , we tested whether the SGTA-SV40 interaction could be recapitulated in vitro using purified components . FLAG-tagged GFP , WT SGTA and K160E/R164E SGTA were purified from transfected 293T cells . WT but not GFP or K160E/R164E SGTA copurified Hsc70 ( Figure 5H ) . These purified components were incubated with purified SV40 pretreated with DTT and EGTA to partially mimic conformational changes that occur in the ER [23] , [24] . When GFP or SGTA proteins were immunoprecipitated from these reactions , VP1 was detected only in reactions containing SGTA ( Figure 5I ) . WT SGTA pulled down moderately more VP1 when compared to K160E/R164E SGTA , likely due to the presence of Hsc70 in the WT SGTA preparation . Nonetheless , the observation that purified K160E/R164E SGTA lacking copurified Hsc70 binds to VP1 demonstrates that SGTA can directly interact with SV40 . Our finding that SV40 liberates SGTA and Hsc70 from the B14-B12 complex suggests that the virus may restructure B14-B12 within the ER membrane . To assess whether SV40 imparts any reorganization of B14-B12 , we stained endogenous proteins in fixed cells uninfected or infected with SV40 for analysis by immunofluorescence microscopy . In uninfected cells , both B14 and B12 colocalized extensively with the ER membrane protein BAP31 ( Figure 6A and 6B , top panels ) , as expected . Strikingly , in infected cells , a significant portion of B14 and B12 was observed to concentrate into discrete foci within the ER ( Figure 6A and 6B , lower panels ) . Some cells were observed to contain a single focus , while others contained several . Notably , B14 and B12 foci colocalized with VP1 ( Figure 6C and 6D ) , suggesting that SV40 is responsible for foci formation . As reported previously , BAP31 also reorganizes into foci during SV40 infection [29] . We found that virus-induced BAP31 foci colocalized with the B14 and B12 foci ( Figure 6A and 6B ) . These results indicate that , while B14-B12 and BAP31 were identified independently to be critical for SV40 membrane penetration , they are likely to be unified in facilitating this entry step rather than functioning in distinct parallel pathways . The localization of Sec61α , a major translocon component reported to associate with BAP31 [56] , remained diffuse and unchanged by addition of SV40 ( Figure S2A , top panels ) . Similarly , Hrd1 staining revealed it did not form foci upon SV40 infection ( Figure S2A , middle panels ) . Although Hrd1 is a central component of ERAD machinery [53] , [57] , it is dispensable for SV40 entry and infection [29] . Consistent with our observation that SGTA disengages from B14-B12 during entry , we did not detect any SV40-induced enrichment of SGTA in the B14-B12-BAP31 foci ( Figure S2A , bottom panels ) . These data indicate that SV40 induces the reorganization of only a specific subset of ER membrane proteins into a discrete region . We observed that knockdown of SGTA did not inhibit foci formation ( Figure S2B ) , suggesting SGTA does not act prior to foci formation . We further characterized SV40-induced foci formation by performing time course experiments where the presence or absence of this structure was monitored during infection by staining with specific antibodies after fixation ( Figure 6E ) . When cells were infected for 2 h , foci were not detected . After 8 and 12 h of SV40 infection , approximately 25–70% of cells contained B12 , B14 , and BAP31 foci . By 16 h , nearly every cell was positive for B12 , B14 , and BAP31 foci . BFA strongly inhibited foci formation . Moreover , we used live cell imaging to monitor foci formation in cells transiently transfected to express GFP-tagged B14 ( GFP-B14 ) . GFP-B14 was observed to gradually reorganize into clear foci at approximately 5–6 h after the addition of SV40 ( Figure 6F ) . We conclude that SV40-induced reorganization of B14-B12 and other membrane components is dependent on ER arrival of the virus , occurs approximately 5–6 h . p . i . , and is largely irreversible . The timing of SV40's cytosolic arrival [40] correlates strongly with our detection of foci in the ER membrane , therefore we postulate that these virus-induced structures may represent ER exit/cytosol entry sites for the virus . We sought to identify the molecular determinants within B14 necessary for SV40-induced B14 foci formation . When full length FLAG-tagged B14 was transfected into CV-1 cells and stained with FLAG antibodies , the transfected protein colocalized extensively with BAP31 in a diffuse manner ( Figure 7A , left panels ) . When cells were infected , the ectopically expressed FLAG-B14 formed foci similar to the behavior of endogenous B14 ( Figure 7A , right panels ) , with variable number of foci per cell observed . Comparable to endogenous B14 foci ( Figure 6A ) , the ectopically expressed B14 foci also colocalized extensively with endogenous BAP31 foci . As transfected WT B14 forms foci in response to SV40 , we then asked whether transfected mutant forms of B14 are capable of forming foci during SV40 infection . Transfected H136Q B14 , which does not interact with Hsc70 or SGTA , formed foci identical to WT B14 ( Figure 7B , first and second row ) . Truncating most of B14's lumenal domain ( 97 amino acids ) to generate Δ lumenal B14 strongly restricted its ability to form foci , despite this mutant colocalizing normally with BAP31 in the absence of SV40 ( Figure 7B , third row; quantified in Figure 7C ) . In addition to the normal ER expression pattern , this mutant also precipitated endogenous B12 and SGTA ( Figure S3 ) , demonstrating the overall integrity of this variant . By contrast , a severe truncation of B14's cytosolic residues ( i . e . Δ cytosol B14 ) did not restrict foci formation , acting similar to the full length WT B14 ( Figure 7B , fourth row ) . Expression of the FLAG-B14 variants had minimal influence on the ability of endogenous BAP31 to form foci and did not perturb infection , likely due to the presence of endogenous B14 and sub-optimal transfection efficiency of CV-1 cells ( Figure 7D ) . We conclude that the lumenal domain of B14 is required for its SV40-triggered reorganization into foci and potentially acts as a sensor for the virus directly . We assessed the molecular requirements of B14 in promoting SV40 infection by performing rescue experiments . HeLa cells were chosen for these experiments because of their high transfection efficiency . DNA plasmids were first transfected for expression from an empty vector , WT B14 , H136Q B14 or Δ lumenal B14 . Subsequently siRNA transfections were carried out using a siRNA oligo targeting the 3′ UTR sequence of B14 to ensure depletion of only endogenous B14 . After 48 h of siRNA treatment , cells were incubated with SV40 and successful infection measured 48 h later by immunoblotting for TAg . Infection of cells expressing empty vector as a control resulted in a complete loss of TAg expression upon B14 knockdown compared to scrambled siRNA ( Figure 8A ) . This phenotype was reversed significantly in cells expressing WT B14 . By contrast , H136Q B14 and Δ lumenal B14 , while expressed at appropriate levels , failed to rescue infection in cells depleted of endogenous B14 ( Figure 8A; quantified in 8B ) . These data support a model whereby B14's ability to engage SGTA-Hsc70 is necessary for SV40 infection . The inability of Δ lumenal B14 to form foci and promote infection suggests that foci may be functionally important for SV40 as well .
To cause infection , nonenveloped viruses must penetrate a biological membrane to gain access into the target cell . While host cues priming these viruses for membrane penetration are well-characterized , cytosolic factors co-opted to complete this membrane penetration event remain unknown . In this study , we pinpoint SGTA as a cytosolic chaperone that promotes membrane penetration of SV40 and likely other polyomaviruses . A model depicting SGTA-dependent SV40 ER membrane penetration is presented in Figure 9 . Our initial finding revealed that a pool of SGTA binds to the cytosolic surface of the ER membrane by engaging two transmembrane J-proteins called B14 and B12 ( Figure 9A ) , essential factors for SV40 ER-to-cytosol transport and infection [33] . Interaction between SGTA and the B14-B12 complex requires Hsc70 , consistent with previous reports demonstrating that SGTA interacts with Hsc70 [44] . The interaction of SGTA with the B14-B12 complex prompted us to ask whether it might serve to dislocate SV40 into the cytosol from the ER membrane . Indeed , our functional studies demonstrated that SGTA downregulation disrupts SV40 and BKPyV infection by blocking virus ER-to-cytosol transport , with subsequent binding experiments establishing a direct SGTA-SV40 physical interaction . These results suggest that SGTA engages the virus on the cytosolic surface of the ER membrane to mobilize it into the cytosol . The use of SGTA to liberate SV40 into the cytosol from the ER membrane resolves a previous enigma in the mechanism of ER membrane penetration by SV40 . Structural alterations occurring within the ER render the virus hydrophobic , enabling it to bind to and integrate into the ER membrane [27] , [29] . Despite these remodeling events , SV40 remains a large and intact particle when it penetrates the ER membrane [40] . How this large and hydrophobic viral particle avoids aggregation in the aqueous cytosolic environment is unclear . Particle aggregation would clearly prevent SV40 from successfully reaching the nucleus to cause infection . One possibility entails cytosolic chaperones positioned at the ER membrane binding to and extracting the hydrophobic virus into the cytosol , and by use of this interaction , protecting the hydrophobic viral surfaces . Our results support this scenario by implicating SGTA , a cytosolic chaperone that binds to hydrophobic proteins [42] , in mobilizing the virus into the cytosol and concomitantly protecting the viral hydrophobic surfaces . Although SGTA has been recently linked to the ER-to-cytosol transport process known as ERAD [42] , its apparent role in this instance is to assist the Bag6 complex in capturing ERAD substrates in the cytosol and preventing their aggregation prior to proteasomal degradation [42] , [46] . By contrast , a separate study found that SGTA antagonizes Bag6 function during protein quality control in the cytosol [45] . Regardless of their relationship , SGTA appears to promote SV40 infection independently of Bag6 . Our data reveals that SGTA likely serves additional undocumented roles in ER biology , possibly in cooperation with B14-B12 . Intriguingly , we observed that SGTA and Hsc70 disengage from the B14-B12 complex , an event that coincides with the drastic reorganization of the B14-B12 complex into foci on the ER membrane ( Figure 9B ) . The energy source driving release of the virus-SGTA-Hsc70 complex into the cytosol is not known . It is possible that virus-induced B14-B12 foci formation imparts a conformational change that weakens the affinity between B14-B12 and SGTA-Hsc70 . While SGTA itself does not harbor ATPase activity , it can modulate the ATPase activity of Hsc70 [43] . Thus , if SV40 binding to SGTA promotes SGTA to drive Hsc70 preferentially to the ADP-bound state , ADP-bound Hsc70 would have lower affinity for its cognate J-proteins B14 and B12 [58] , [59] . This postulated scenario could explain how SV40 triggers release of SGTA-Hsc70 from B14-B12 . Future experiments will clarify the precise mechanism by which SV40 induces SGTA and Hsc70 to disengage from the B14-B12 complex . Although the precise physiological significance of this observation is not entirely clear , discharge of the SGTA-Hsc70 complex bound with virus into the cytosol from the ER membrane ( Figure 9C ) is conceptually consistent with the requirement of SV40 to reach the cytosol in preparation for nuclear import . As Hsc70 proteins are observed to disassemble murine polyomavirus in vitro [55] , polyomaviruses entering a host-cell could in principle co-opt both SGTA and Hsc70 activities to couple the cytosol release and viral disassembly reactions , with the latter likely being necessary for subsequent nuclear import . The B14-B12 foci were positive for VP1 , similar to a previous report that found VP1 colocalizes with BAP31 foci [29] . As ER membrane penetration is an inefficient process [40] , only a small fraction of virus in the foci are expected to be released into the cytosol , explaining why VP1 accumulates in the foci at later time points . These foci unlikely represent nonspecific aggregation structures as no dramatic changes in the solubility of the B14 and B12 membrane proteins were observed during foci formation . Moreover , B14-B12 foci colocalize with BAP31 , another ER membrane factor involved in SV40 ER membrane penetration [29] , but not with other membrane proteins dispensable for SV40 infection such as a core ERAD component Hrd1 . Thus foci consist of specific ER factors that conduct SV40 across the ER membrane but not ER components irrelevant to this process . Intriguingly , there is a temporal correlation between foci formation and virus arrival in the cytosol . Foci formation can be readily identified prior to 8 h . p . i . and SV40 cytosol arrival occurs between approximately 6–8 h . p . i . [29] , [40] . Collectively , these observations raise the possibility that the B14-B12 foci may represent cytosol entry sites that function to increase the transient recruitment of SGTA-Hsc70 precisely at the membrane penetration site where SV40 can be efficiently extracted into the cytosol . It remains possible that smaller foci structures , which are not clearly detected by microscopy , also exist to mediate viral transport . This would explain why a stable SGTA-B14 interaction appears fully disrupted at time points where not all cells have visible foci . A direct interaction between SV40 and B14-B12 or BAP31 has yet to be demonstrated , possibly due to a weak affinity of these factors to the particle within the membrane . This weak physical interaction could also explain the necessity of foci formation , which would facilitate multivalent interactions among several membrane components functioning to promote membrane penetration . At present , the relationship between B14-B12 and BAP31 remains obscure . Although these membrane proteins form foci in response to SV40 entry , no obvious physical interaction between B14-B12 and BAP31 can be isolated . Moreover , while BAP31 is thought to recognize membrane integrated SV40 with exposed VP2 via charge-pairing [29] , both B14 and B12 lack charged residues within their transmembrane domains and thus recognize SV40 differently . In this regard , by evaluating different B14 mutants , we found that the lumenal portion of B14 is required for its SV40-induced reorganization and its ability to promote infection . This region of B14 , which lacks any clear protein-protein interaction domains , could therefore act as a sensor for engaging SV40 complementarily with BAP31 . An outstanding question is whether a bona fide protein-conducting channel exists to accommodate SV40 transport across the ER membrane . If this is the case , it is unlikely that a native channel can support the transport of such a large viral particle whose diameter is approximately 45–50 nm [40] . Instead , the hydrophobic virus might initially integrate into the membrane and subsequently recruit and oligomerize ER membrane proteins such as B14 , B12 , and BAP31 . The oligomerized structure ( i . e . foci ) would surround the viral particle as an intermediate until SGTA-Hsc70 extracts it into the cytosol . In conclusion , this study identifies a novel cytosolic chaperone complex that completes ER membrane penetration of a nonenveloped virus , and utilizes dramatic rearrangement of ER membrane elements in the process . Viral entry , replication , and assembly are defining steps during the infection course . While diverse viruses are known to rearrange the ER membrane to facilitate viral replication and assembly [60] , essentially nothing is known regarding how viruses reorganize the ER membrane during entry . The possibility that SV40 and other polyomavirus family members might reorganize components of the ER membrane to fashion its own entry site would demonstrate that viruses have the capacity to restructure the ER membrane to accommodate the early events of infection .
Polyclonal DnaJB14 , DnaJB12 , ERdj5 , SGTA , Hrd1 were purchased from Proteintech Group ( Chicago , IL ) . An additional rabbit polyclonal against SGTA was provided by Yihong Ye ( NIH ) . Monoclonal BAP31 and polyclonal Hsc70 and Bag6 antibodies were purchased from Pierce ( Rockford , IL ) . Polyclonal Derlin-1 and Sec61α antibodies were provided by Tom Rapoport ( Harvard University ) . Normal rabbit and mouse IgG , polyclonal Hsp90 , PDI , and monoclonal SV40 Large T antigen antibodies were purchased from Santa Cruz Biotechnology ( Santa Cruz , CA ) . Rabbit anti-VP1 antibody was a gift from Harumi Kasamatsu ( UCLA ) . Monoclonal VP1 antibody was provided by Walter Scott ( University of Miami ) . Monoclonal p97 was purchased from RDI/Fitzgerald ( Concord , MA ) . Polyclonal BiP and rat anti-Hsc70 antibodies were purchased from Abcam ( Cambridge , MA ) . Polyclonal calnexin antibodies were purchased from Stressgen . Polyclonal ERp29 was a gift from Souren Mkrtchian ( Karolinska Institutet ) . Dulbecco's modified Eagle's medium ( DMEM ) , Opti-MEM , 0 . 25% trypsin-EDTA were purchased from Invitrogen ( Carlsbad , CA ) . Fetal Clone III ( FC ) was from HyClone ( Logan , UT ) . Complete-mini EDTA-free protease inhibitor cocktail tablets were purchased from Roche . Micro Bio-Spin P-30 Tris chromatography columns were purchased from Bio-Rad . Dithiothreitol ( DTT ) , Dithiobis succinimidylpropionate ( DSP ) , N-ethylmaleimide ( NEM ) and anti-FLAG M2 agarose beads were purchased from Sigma ( St Louis , MO ) . WT SV40 and ΔVP2 SV40 were prepared using OptiPrep gradient system as described previously [40] . Control siRNA ( labeled as scrambled ) is the All Star Negative purchased from Qiagen ( Valencia , CA ) . Custom siRNA sequences were generated and purchased from Dharmacon ( Pittsburgh , PA ) or Invitrogen . B14 siRNA: 5′ GGUUCCUGAAAUCUUGGACUGUUUA 3′ 5′ UAAACAGUCCAAGAUUUCAGGAACC 3′ SGTA siRNA #1: 5′ ACAAGAAGCGCCUGGCCUAUU 3′ 5′ UAGGCCAGGCGCUUCUUGUUU 3′ SGTA siRNA #2: 5′ CAGCCUACAGCAAACUCGGCAACUA 3′ 5′ UAGUUGCCGAGUUUGCUGUAGGCUG 3′ SGTA siRNA #3: 5′ CCAACCUCAAGAUAGCGGAGCUGAA 3′ 5′ UUCAGCUCCGCUAUCUUGAGGUUGG 3′ Bag6 siRNA #1: 5′ GCUUGGAGGUGUUGGUGAAUU3′ 5′ UUCACCAACACCUCCAAGCUU3′ Bag6 siRNA #2: 5′ GAUAAGAAGCUUCAGGAAUUU 3′ 5′ AUUCCUGAAGCUUCUUAUCUU 3′ Using Lipofectamine RNAiMAX ( Invitrogen ) , 25 nM ( B14 or Bag6 siRNAs ) or 12 . 5 nM ( SGTA siRNAs ) of control or custom siRNAs were reverse transfected into HeLa or CV-1 cells . Infection or biochemical assays were carried out 24 or 48 h . p . i . . A plasmid expressing human WT B14 was a gift from Daniel DaMaio ( Yale ) and cloned with an N-terminal FLAG tag into pcDNA3 . 1 minus ( Invitrogen ) . Site-directed mutagenesis was performed on His 136 to yield a H136Q mutant . The Δlumenal FLAG-B14 contains residues 1-282 and Δcytosol contains residues 226-379 , and were generated using standard cloning methods . A plasmid expressing SGTA-myc/FLAG is from Origene ( Rockville , MD ) and double point mutants were derived similarly by site-directed mutagenesis . DNA was transfected into plated CV-1 cells ( 50–80% confluent ) using FuGENE HD ( Promega ) and allowed to express for 18–24 h prior to experimentation . Interactions detected between endogenous B14 and B12 utilized lysates derived from CV-1 or HeLa cells with buffer containing 50 mM Tris pH 7 . 4 , 150 mM NaCl , 1 mM EDTA and protease inhibitors with 1% deoxyBigCHAP ( Calbiochem , Billerica , MA ) . Cleared lysates were incubated with antibodies overnight at 4°C with rotation . Antibodies were captured using protein A agarose beads and washed with lysis buffer containing 0 . 1% deoxyBigCHAP . SDS sample buffer was used for elution at 95°C . Interactions between endogenous SGTA and B14-B12 were detected similarly but with a different lysis buffer . 293T or CV-1 cells were lysed by resuspension and brief vortexing in 0 . 1% digitonin , 50 mM Tris , 150 mM NaCl , 1 mM EDTA and protease inhibitors followed by incubation on ice for 15 minutes . Cleared lysates were incubated with indicated antibodies , captured with protein A agarose beads and washed extensively with buffer lacking detergent . Immunoprecipitation of transfected SGTA-myc/FLAG was performed as above but with a buffer containing 0 . 2% digitionin and incubation with anti-FLAG M2 agarose beads . In 6 cm plates , 3×105 CV-1 cells were transfected to express FLAG-tagged GFP or SGTA proteins for at least 18 h . SV40 particles ( 15 μg ) were added to cells after synchronizing the entry at 4°C . Approximately 8 h . p . i . , cells were harvested and cross-linked using 2 mM DSP at room temperature for 30 mins . After quenching , cells were lysed in 1% Triton X-100 , 50 mM Tris pH 7 . 5 , 150 mM NaCl , 1 mM EDTA , 20 mM NEM and protease inhibitors . Cleared lysates were immunoprecipitated with anti-FLAG agarose beads overnight and washed 3 times with lysis buffer . Bound material was analyzed by immunoblotting using rabbit anti-VP1 antibody . As previously described [25] using primers against human XBP1: 5′ CGCGGATCCGAATGTGAGGCCAGTGG 3′ and 5′ GGGGCTTGGTATATATGTGG 3′ . HeLa cells were lysed in 1% Triton , 30 mM Tris pH 8 , 150 mM NaCl , and 4 mM MgCl2 . Cleared lysates were separated using lysis buffer and a Bio-Sil SEC 250 column ( Bio-Rad , Hercules , CA ) . Forty fractions of 0 . 5 mL were collected and fractions 10-22 were analyzed by immunoblotting . For purification of FLAG-tagged GFP , WT and mutant SGTA , 293T cells were transfected to express proteins for 48 h . Cells were lysed in buffer containing 1% Triton X-100 , 50 mM Tris , 150 mM NaCl , 1 mM EDTA and protease inhibitors . Cleared lysates were incubated with anti-FLAG agarose beads and bound proteins washed extensively with lysis buffer . Proteins were eluted with FLAG peptide overnight and concentrated using centrifugal filters that also removes residual FLAG peptides . Optiprep purified SV40 was treated with 3 mM DTT and 10 mM EGTA for 45 min at 37°C to mimic ER induced conformational changes . A spin column was used to exchange the buffer with PBS and to remove DTT and EGTA . Binding reactions were carried out in 50 μL PBS containing 250 ng of SV40 ( pretreated with DTT and EGTA as described above ) with or without 1 μg of purified protein . Reactions incubated for 1 h at 25°C followed by the addition of 0 . 25 mM DSP at 4°C for 30 min to stabilize transient interactions . After quenching with excess Tris , immunoprecipitation with anti-FLAG agarose beads was performed and bound material analyzed by immunoblotting . CV-1 cells were grown on 12 mm coverslips in 6 or 24-well plate for 24 h . Cells were treated with SV40 for the indicated time and then washed in PBS followed by fixation with 1% formaldehyde at room temperature . Cells were permeabilized with 0 . 2% Triton X-100 and blocked with 5% milk and 0 . 2% Tween . Primary antibodies were incubated for 1 h at room temperature , followed by fluorescent conjugated secondary antibodies for 30 min at room temperature . Coverslips were mounted with ProLong Gold ( Invitrogen ) . Images were taken using an inverted epifluorescence microscope ( Nikon Eclipse TE2000-E ) equipped with 60× and 100× 1 . 40 NA objective and a Photometrics CoolSnap HQ camera . For over-expression studies , cells were transfected with the desired plasmid with FuGene ( Promega ) at least 24 h prior to imaging . For live-cell imaging , cells are seeded on 35 mm glass bottom tissue culture dishes ( Greiner Bio-one , Germany ) . Imaging of the cells was performed from 2 to 20 h . p . i . using the microscope and objective mentioned above . The entire set-up was controlled by MetaMorph software ( Molecular devices ) and ImageJ software ( NIH ) was used for image processing , analysis , and assembly . CV-1 cells transfected with siRNA for 24 h were incubated with SV40 at 37°C ( MOI ≈0 . 5 or 5 ) . At 20–24 h . p . i . , cells were harvested for analysis by immunoblot or fixed and stained using antibodies against SV40 TAg as described previously [40] . For each infection experiment , at least 500 cells were counted in each condition . Approximately 30–50% or 5–10% of cells were positive for TAg in the control conditions when challenged at MOI ≈5 or 0 . 5 , respectively . Purified BKPyV and pAb416 against BK large T antigen were provided by Michael Imperiale ( University of Michigan ) . CV-1 cells transfected with siRNA for 24 h were infected at M . O . I . ≈0 . 5 . Cells were harvested after 40 h . p . i . , and immunoblot performed for analysis of TAg expression . Performed as in [40] . SV40 was added ( M . O . I ≈5 ) to CV-1 cells grown in 6 cm plates ( ∼80% confluent ) . Where indicated , BFA ( Epicenter , Madison , WI ) was added to the media at 2 . 5 μg/mL . After fractionation , 40% of the supernatant fraction was compared alongside 10% of the pellet fraction for VP1 immunoblot analyses , as only a portion of SV40 reaches the cytosol . Fractionation markers were analyzed with equivalent amounts of supernatant and pellet . ImageJ software was used for quantification of VP1 band intensities . CTA1 transport was monitored in CV-1 cells as described in [40] . Cells were intoxicated with CT at 10 nM for 90 mins . ER arrival assays were also performed as described previously [40] . Flp-In T-REx 293 cells ( Invitrogen ) were transfected with a pcDNA5/FRT/TO plasmid expressing WT B14-3xFLAG along with pOG44 plasmid expressing a Flp recombinase . Selection was carried out with media containing hygromycin and blasticidin over several weeks . Expression of B14-3xFLAG to near endogenous levels was induced with 5 ng/mL of tetracycline provided to the media for 16 h . Three confluent 15 cm plates were collected in PBS and lysed for 30 mins on ice in 2 . 5 mL of buffer containing 0 . 1% digitonin , 50 mM Tris pH 7 . 4 , 150 mM NaCl , 1 mM EDTA and protease inhibitor . Lysate was cleared with centrifugation at 20 , 000g for 15 min . Lysate divided in half was incubated for 2 h at 4°C with 30 μL of anti-FLAG M2 agarose beads that was pre-incubated with or without 3xFLAG peptide ( 100 μL , 0 . 25 mg/mL ) . Agarose beads were washed extensively with buffer lacking detergent . Bound proteins were eluted overnight at 4°C with 3xFLAG peptide ( 200 μL , 0 . 25 mg/mL ) . Three subsequent elutions were performed for 1 h each , pooled and concentrated using centrifugal filters ( Amicon Ultra 0 . 5 mL 3K membrane ) . SDS sample buffer was added and heated for 30 min at 37°C followed by SDS-PAGE and silver staining or immunoblotting . Bands excised from a silver stained gel were analyzed by mass spectrometry at Taplin Biological Mass Spectrometry Facility ( Harvard Medical School ) . In 6 cm plates , 4×105 HeLa cells were transfected with pcDNA3 . 1 ( empty vector ) or B14 constructs for at least 18 h . Then 2×105 cells were reverse transfected with either scrambled or B14 siRNA into wells of a 6 well plate . Cells were infected with SV40 48 h post-siRNA transfection and harvested for immunoblot after another 48 h . Quantitative data is presented as the mean of at least three independent experiments with standard deviation . Paired two-tailed Student's t-tests were used to acquire p-values . VP1 P03087-1 , VP2/3 P03093-1/2 , B14 Q8TBM8 , B12 Q9NXW2 , Hsc70 Q96I56 SGTA O43765 , BAP31 P51572 , Bag6 P46379 , TAg P03070 Hrd1 Q86TM6 , ERp29 P30040 , ERdj5 Q8IXB1 , Calnexin P27824 , Ubl4a P11441 , Trc35 Q7L5D6 , GEMIN5 Q8TEQ6 , GEMIN4 P57678 , KIF11 P52732 , DDX20 Q9UHI6 , EIF3H O15372 , SMN2 Q16637
|
The nonenveloped simian virus 40 ( SV40 ) is a model member of the Polyomaviridae family of viruses containing several related species that cause diseases in immunocompromised individuals . As with other nonenveloped viruses , the membrane penetration step during SV40 entry is mechanistically obscure . Productive SV40 infection requires trafficking of the viral particle to the endoplasmic reticulum ( ER ) from where it penetrates the ER membrane to reach the cytosol; further transport of the virus into the nucleus causes infection . How SV40 crosses the ER membrane is an enigmatic step . Here , we identify a cytosolic chaperone protein that physically engages SV40 and facilitates virus ER-to-cytosol transport . This factor called SGTA is hijacked specifically at the site of membrane penetration due to its recruitment by ER membrane proteins B14 and B12 previously implicated in supporting virus infection . Additionally , we observe that B14 and B12 reorganize during SV40 entry into discrete foci on the ER membrane . These virus-induced structures likely represent exit sites for the viral particles and could serve to transiently recruit high concentrations of SGTA to complete membrane penetration . Our data reveal that a cytosolic chaperone can play a direct role in membrane penetration of a nonenveloped virus .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"biology",
"and",
"life",
"sciences",
"microbiology",
"virology"
] |
2014
|
A Cytosolic Chaperone Complexes with Dynamic Membrane J-Proteins and Mobilizes a Nonenveloped Virus out of the Endoplasmic Reticulum
|
Leishmania ( Viannia ) parasites present particular challenges , as human and murine immune responses to infection are distinct from other Leishmania species , indicating a unique interaction with the host . Further , vaccination studies utilizing small animal models indicate that modalities and antigens that prevent infection by other Leishmania species are generally not protective . Using a newly developed mouse model of chronic L . ( Viannia ) panamensis infection and the heterologous DNA prime – modified vaccinia virus Ankara ( MVA ) boost vaccination modality , we examined whether the conserved vaccine candidate antigen tryparedoxin peroxidase ( TRYP ) could provide protection against infection/disease . Heterologous prime – boost ( DNA/MVA ) vaccination utilizing TRYP antigen can provide protection against disease caused by L . ( V . ) panamensis . However , protection is dependent on modulating the innate immune response using the TLR1/2 agonist Pam3CSK4 during DNA priming . Prime-boost vaccination using DNA alone fails to protect . Prior to infection protectively vaccinated mice exhibit augmented CD4 and CD8 IFNγ and memory responses as well as decreased IL-10 and IL-13 responses . IL-13 and IL-10 have been shown to be independently critical for disease in this model . CD8 T cells have an essential role in mediating host defense , as CD8 depletion reversed protection in the vaccinated mice; vaccinated mice depleted of CD4 T cells remained protected . Hence , vaccine-induced protection is dependent upon TLR1/2 activation instructing the generation of antigen specific CD8 cells and restricting IL-13 and IL-10 responses . Given the general effectiveness of prime-boost vaccination , the recalcitrance of Leishmania ( Viannia ) to vaccine approaches effective against other species of Leishmania is again evident . However , prime-boost vaccination modality can with modulation induce protective responses , indicating that the delivery system is critical . Moreover , these results suggest that CD8 T cells should be targeted for the development of a vaccine against infection caused by Leishmania ( Viannia ) parasites . Further , TLR1/2 modulation may be useful in vaccines where CD8 T cell responses are critical .
Traditionally , vaccination against cutaneous leishmaniasis ( CL ) has involved leishmanization ( inoculation of live Leishmania ) , which has been practiced throughout the Middle East and was employed in government sponsored vaccination programs both in Israel and Russia . However , safety and standardization issues discouraged further use of live vaccination [1] , [2] . Subsequently , killed Leishmania promastigotes have been examined with some but limited efficacy in clinical trials [3] , [4] . Consequently , leishmaniasis vaccine efforts have focused on the use of live attenuated vaccines [5] , [6] and also defined molecular vaccines and delivery systems [7] . An optimal vaccine against cutaneous leishmaniasis would consistently provide protection against the various disease-causing species . However , studies indicate that distinct Leishmania species elicit different responses in their hosts , suggesting that a uniform approach might be challenging . Although a Th1-like response is considered to lead to disease resolution , the mechanisms contributing to protection across the species are not well characterized/understood . In particular , the Leishmania ( Viannia ) subgenus is phylogenetically divergent from the Leishmania ( Leishmania ) subgenus [8] , [9] , [10] . Members of the L . ( Viannia ) subgenus can generate a hyperinflammatory response that fails to resolve [11] , [12] , [13] , . L . ( V ) . panamensis elicits a mixed Th1/Th2 and non-resolving hyperinflammatory response to infection in humans [17] , [18] . Consistent with this , vaccine studies attempting to demonstrate immunological protection against L . ( Viannia ) parasites [19] , [20] , [21] using a murine model , have met with limited success . Salay et al . [19] tested four different highly conserved leishmanial antigens ( DNA or recombinant protein ) along with adjuvants that have protected against infection with other species causing CL ( L . mexicana , L . amazonensis , L . major ) . However , strong antigen-specific IFNγ production by immunized mice failed to translate into protection against L . ( Viannia ) braziliensis infection . As a result this study suggested investigation of alternate immunization strategies to protect against L . ( Viannia ) parasites . Similarly , antigens demonstrated to protect against visceral leishmaniasis [21] failed to protect against L . ( Viannia ) braziliensis . Recently , partial protection was [20] demonstrated against L . ( V . ) braziliensis by utilizing an attenuated centrin-deficient L . donovani strain . Taken as a whole these studies might suggest that defined antigens may not provide protection against L . ( Viannia ) . However , vaccine delivery systems are critical to determining the elicited immune response and therefore can determine protection provided for an antigen . In particular , as the mechanisms involved in disease resolution for L . ( Viannia ) are not well understood , further investigation of delivery systems/antigens is warranted and may ultimately provide insight into immune mechanisms leading to healing . Hence we explored other immunization methods to induce protection against L . ( V . ) panamensis , using a newly developed murine model for chronic disease [22] . Herein we report for the first time that heterologous prime ( DNA ) -boost ( modified vaccinia virus Ankara = MVA ) modality using the single antigen tryparedoxin peroxidase ( TRYP ) and including the TLR1/2 agonist N-palmitoyl-S-[2 , 3-bis ( palmitoyloxy ) - ( 2RS ) -propyl]-[R]-cysteinyl-[S]-seryl-[S]-lysyl-[S]-lysyl-[S]-lysyl-[S]-lysine ( Pam3CSK4 ) as adjuvant during DNA priming is effective in achieving protection against L . ( V . ) panamensis . Markedly , prime boost immunization in the absence of Pam3CSK4 did not elicit protection thereby implicating a strategic role for Pam3CSK4 in achieving protection . Pam3CSK4 appears to direct heightened CD4 and CD8 T memory cell responses and reduced levels of IL-10 and IL-13 , which ultimately results in significant protection against L . ( V . ) panamensis . Furthermore CD8 cells , but interestingly not CD4 T cells , are crucial in mediating the protection induced , indicating that CD8 T cell responses may be critical for vaccine development against L . ( Viannia ) parasites .
L . ( V . ) panamensis was grown and cultured into infective stage parasites as described previously [22] . Briefly L . ( V . ) panamensis was grown in Schneider's Medium supplemented with 20% heat inactivated FCS and 17 . 5 µg/mL gentamycin ( GIBCO BRL ) . Promastigotes were grown at 22°C . Live late stationary phase ( 15–21 days in culture ) promastigotes were harvested for infection using a step percoll gradient ( Sigma Chemical Co . ) in PBS containing 20 mM EDTA . Washed parasites ( 5×104 ) were used to infect mice in the top of the right hind foot . Female BALB/c mice ( 5 to 6 weeks old ) were purchased from the NCI . All mice were housed in Yale University School of Medicine facilities , which are American Association for Accreditation of Laboratory Animal Care ( AAALAC ) accredited and USDA registered animal facilities . The experiments were approved by Yale University Committee on Use and Care of Animals ( Assurance number A3230-01 ) . TRYP and p36 ( LACK ) genes were cloned into pVAX ( Invitrogen , CA ) and pCI-neo ( Promega , WI ) vectors respectively . Plasmids were purified using Qiagen Endofree Plasmid Giga kit ( Qiagen , CA ) . Empty plasmid was used for the controls . Plasmid preparations were tested for endotoxin by Limulus Amebocyte Lysate test ( Lonza , MD ) ; less than 0 . 1 ng LPS per 100 µg of plasmid was present in preparations employed for vaccination . The p36 and TRYP recombinant proteins were expressed using a histidine-tag construct that was cloned into pRSET A vector kindly provided by Dr . Larraga ( Centro de Investigaciones Biológicas , Spain ) and pET-15b vector , respectively . Recombinant protein was purified using PrepEase Histidine-Tagged Protein Purification kit ( USB , OH ) and endotoxin was removed as described [23] . Coomassie blue staining of SDS-PAGE analysis of recombinant antigen was used to determine protein purity . Vaccinia virus Ankara ( MVA ) expressing TRYP and LACK were prepared as previously described [24] , [25] . For adjuvant evaluation , mice ( 3/group ) received two intra dermal injections of p36 DNA ( 100 ug in 100 ul ) per vaccination with or without adjuvants ( α-GalCer ( 1 ug ) , LPS ( 10 ug ) , CpG ( 50 ug ) , Pam3CSK4 ( 10 ug ) , MALP-2 ( 0 . 5 ug ) ) . After an interval of 3 weeks , mice were boosted using the same DNA-adjuvant combination . Splenocytes from the vaccinated mice were evaluated by in vitro cytokine production 4 weeks after the final immunization . This experiment was done twice and 3 mice per group were sufficient to achieve statistical significance and evaluation of data . In the case of DNA-vaccinia virus ( MVA ) prime-boost vaccination , 2 weeks after the priming immunization with TRYP ( 100 ug/100 ul ) ± Pam3CSK4 , mice ( 8 to 10/group ) were boosted intraperitoneally with 3×106 PFU per mouse of MVA-TRYP . This was followed by infection with 5×104 late stationary phase L . ( V . ) panamensis promastigotes 6 weeks after the MVA boost . Lesion development was monitored by measuring the thickness of the infected and uninfected feet using a dial gauge caliper ( Starrett Thickness Gauge ) . Parasite burdens in the infected foot and draining lymph node ( DLN ) were determined by limiting dilution analysis as described previously [26] , [27] . As parasite burden changes in the infected foot and associated DLN were comparable comparative in initial experiments , parasite burdens were determined only in the infected feet of the immunodepleted mice . For depletion of CD4 or CD8 cells , immunized mice were injected intraperitoneally with 100 µg of anti-CD4 ( GK1 . 5 ) or anti-CD8 ( 53-5-43 ) antibody ( eBioscience , CA ) at -3 and -1 day before infection . Flow cytometry indicated that more than 95% of the target cell population was depleted . Pre-challenge immunoassays were carried out 12 weeks after the MVA boost in TRYP immunized mice . DLN cells and splenocytes were plated at 5×106/ml in RPMI ( 10% fetal bovine serum , 2 mM L-glutamine , 100 units/ml penicillin , 100 µg/ml streptomycin , and 50 µM β-mercaptoethanol ) in 96 or 24 well plates . The cells were then stimulated with recomfbinant TRYP ( 5 µg/ml ) , recombinant LACK/p36 ( 5 µg/ml ) , soluble leishmania antigen ( SLA; equivalent to 5×106 parasites/ml ) or left unstimulated for 72 hours . Supernatants were collected and analyzed for IFNγ , IL-10 , and IL-13 using paired antibodies from BD Biosciences ( CA ) and R&D systems ( MN ) . For flow cytometry , brefeldin A ( BD Biosciences , CA ) at 1 ug/ml was added to stimulated splenocytes during the last 4 hours , cells were surface stained with T cell markers , fixed with 2% paraformaldehyde and permeabilized with 0 . 05% saponin followed by intracellular staining . Isotype control antibodies were IgG1-PE-Cy7 and IgG1-PE . Forward and side scatter were used to determine lymphocytes followed by gating on CD4+ or CD8+ cells . Integrated mean fluorescence intensity ( iMFI ) was calculated by the following formula: iMFI = MFI x frequency [28] . Data were acquired using an LSRll ( BD Biosciences , CA ) and analyzed using FlowJo ( Treestar Inc . , Oregon ) . For proliferation analyses to evaluate memory responses , splenic lymphocytes were labeled with 5 µM carboxyfluorescein succinimidyl ester ( CFSE ) at 3 months/12 weeks after the final MVA boost; cells were then placed in 96 well plates at 5×106/ml in RPMI ( 10% fetal bovine serum , L-glutamine , penicillin/streptomycin , and β-mercaptoethanol ) , and stimulated with recombinant TRYP ( 5 µg/ml ) for 3 days . Following surface staining with CD4 and CD8 antibodies , FACS analysis was done as described above to measure proliferation by dilution of CFSE dye . Unlabeled cells and unstimulated CFSE-labeled cells were used as controls . Antibodies ( CD4-Pacific blue , CD8-APC , IFNγ-PE-Cy7 , IL-13-PE , IgG1-PE-Cy7 , IgG1-PE , affinity purified CD4 ( L3T4 ) , and CD8 ( Ly-2 ) ) were purchased from BD Biosciences ( CA ) and eBiosciences ( CA ) . Pam3CSK4 , CpG ( ODN1826 ) , ultra pure E . coli lipopolysaccharide ( LPS ) , and MALP-2 were purchased from Invivogen Inc ( CA ) . α-galactosyl-ceramide was obtained from Biomol International ( PA ) . All experiments were approved Yale University Committee on Use and Care of Animals ( Assurance number A3230-01 ) . Student's t test was used to determine p values indicating statistical differences for all experiments and p<0 . 05 was considered statistically significant .
Initially , we sought to examine the effects of adjuvants on the immune response induced during DNA vaccination , as it is known that priming is critical [29] , , to the overall response induced in heterologous prime-boost vaccination . Compounds known to activate NK-T cells ( alpha-galactosyl ceramide ( α-Galcer ) ) [33] , as well as TLR ligands TLR9 ( CpG [34] ) , TLR2/6 ( MALP-2 [35] ) , TLR4 ( LPS [36] ) , and TLR2/1 ( Pam3CSK4 [37] ) were chosen as potential adjuvants for DNA vaccination , using the Leishmania homologue of receptors for activated C-kinases ( LACK ) /p36 antigen . The adjuvants ( Figure S1A ) were evaluated on the basis of their ability to enhance the production of antigen specific IFNγ relative to IL-10 in DNA immunized mice , as better protection against cutaneous leishmaniasis has been associated with a higher ratio of IFNγ to IL-10 [24] and disease exacerbation of L . ( Viannia ) has been related to IL-10 levels [38] . High IL-10 production was observed for MALP-2 and LPS . The IFNγ/IL-10 ranking was Pam3CSK4≥ α-GalCer>CpG>>LPS>>>MALP-2 . Consequently , amongst the adjuvants , Pam3CSK4 and α-GalCer appeared to induce a potentially useful immune response . Based on these results , Pam3CSK4 and α-GalCer were selected to determine if a combination of Pam3CSK4 and/or α-GalCer , which have distinct cellular targets , could act synergistically to further enhance the immunogenicity of DNA-p36 . In these experiments ( Figures S1B and S1C ) , IFNγ , IL-13 , and IL-10 responses to p36 were determined . Although a considerable increase in the IFNγ to IL-10 ratio was seen in response to immunization with DNA-p36+ α-GalCer +Pam3CSK4 , α-GalCer alone induced increased levels of IL-13 ( Figure S1C ) . As IL-13 has been shown to play a critical role in determining progression of non-healing pathogenesis of L . ( V . ) panamensis [22] , Pam3CSK4 was selected as a potential adjuvant for use in a prime ( DNA ) boost ( vaccinia ) immunization against murine L . ( V . ) panamensis infection . To further evaluate the potential of Pam3CSK4 in DNA priming , heterologous prime-boost immunization ( Figure 1 ) using the TRYP antigen was examined . Mice were immunized intradermally with DNA-TRYP or DNA-TRYP together with Pam3CSK4 . Control mice received empty vector DNA . Two weeks later , groups of the immunized mice were boosted with vaccinia virus expressing the TRYP antigen ( MVA-TRYP ) or with control vaccinia virus ( MVA ) . Twelve weeks after the final immunization , the immune responses of splenocytes from immunized and non-immunized mice were analyzed . Mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP produced increased levels of IFNγ and granzyme B in response to both TRYP and SLA when compared to mice immunized with DNA- TRYP+MVA-TRYP or DNA-TRYP ( Figure 1 ) . Furthermore , mice receiving Pam3CSK4 during priming also produced significantly lower levels of antigen specific IL-13 and IL-10 when compared to mice immunized with DNA-TRYP+MVA-TRYP . Thus , DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice generated higher amounts of IFNγ as well as reduced levels of IL-13 and IL-10 in response to either TRYP or SLA antigen when compared to mice immunized with DNA-TRYP alone . Hence the activation of TLR1/2 ( Pam3CSK4 ) during DNA priming ( as found for DNA vaccination alone ) appears to promote the down-regulation of IL-13 and IL-10 responses and concurrent up-regulation of Th1 cytokines in heterologous prime-boost vaccination , leading to higher IFNγ to IL-13 and IFNγ to IL-10 ratios . To further evaluate the effects of Pam3CSK4 on the development of specific long-term CD4 and CD8 T cell memory , the proliferative responses to TRYP antigen were examined 12 weeks after the MVA-TRYP boost . Splenic lymphocytes from immunized mice were labeled with CFSE and then stimulated for 3 days with recombinant TRYP protein . Increased proliferation of both CD4 and CD8 cells ( Figure 2A ) was observed in DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice when compared to the other vaccinated groups . Overall , the responses observed ( 10–15% proliferating cells ) were comparable to other studies examining long-term memory [39] , [40] , [41] . The kinetics indicated a heightened response ( CD4 and CD8 T cells ) after 3 days of stimulation for the cells from mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP . As the number of proliferating cells is expected to be in proportion to memory populations , these results clearly indicate increased levels of both CD4 and CD8 memory cells as a consequence of vaccination using Pam3CSK4 . In particular , a higher increase in antigen specific CD8 cell proliferation ( 5 . 4-fold ) was observed in mice receiving DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP in comparison to CD4 cells ( 1 . 5-fold ) . These data suggest that although both CD4 and CD8 memory populations expand as a result of TLR1/2 ligation , a selective effect on CD8 T cell populations occurs . Overall , these data point to a more rapid and robust response and higher levels of memory populations ( CD4+ and CD8+ ) in mice receiving Pam3CSK4 in comparison to those immunized with DNA-TRYP alone . The immunization with DNA-TRYP ( Pam3CSK4 ) + MVA-TRYP resulted in a polarized Th1 immune response 12 weeks after the final booster dose ( Figure 1 ) . Given that expansion of both CD4 and CD8 memory cells ( Figure 2A ) was observed , it was of interest to investigate the precise cellular components of this immune response . FACS analysis of cells stimulated with recombinant TRYP was carried out at twelve weeks after the final immunization ( Figure 2B ) . These results , consistent with ELISA results , indicated that the frequency of CD4 and CD8 cells producing IFNγ were increased in DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice in comparison to the DNA-TRYP+MVA-TRYP immunized mice . Notably a significantly lower level of CD4 and CD8 T cells producing IL-13 was found . Overall , the increased frequency of IFNγ producing CD4 and CD8 cells in DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice in comparison to the DNA-TRYP+MVA-TRYP immunized mice or control groups were statistically different ( Figure 2C ) . Overall , these results suggest that TLR1/2 activation drives the development of Th1/TC1-like responding T ( CD4 and CD8 ) cells . Consequently , the ligation of TLR9 ( by bacterial CpG sequences ) together with TLR1/2 during priming appears to preferentially enhance the generation of TRYP-specific memory T cells , producing IFNγ but also significantly less IL-13 . Given the fact that mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP exhibit enhanced levels of memory T cells together with an overall reduction in IL-13 and increased IFNγ production in comparison to DNA-TRYP+MVA-TRYP or DNA-TRYP vaccinated mice , we asked whether these responses might be useful in directing protection against infection . Mice were vaccinated as indicated above . Control mice were immunized with control plasmid and control MVA . Six weeks after the final immunization , all mice were infected with 5×104 L . ( V ) . panamensis promastigotes and lesion development was monitored . As shown in Figure 3A , mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP exhibited significantly smaller lesions when compared to control mice ( control plasmid and MVA ) and all other vaccine groups . Furthermore , significantly lower parasite burden levels were found at the both the site of infection and DLN of the DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP vaccinated mice , when compared to control immunized mice ( 574-fold ) or mice immunized with DNA-TRYP ( 267-fold ) or TRYP+MVA-TRYP ( 314-fold ) ( Figure 3B ) . Consistent with previously reported results [19] , DNA-TRYP immunization alone failed to protect against L . ( V ) . panamensis infection . Notably , heterologous prime-boost vaccination alone also does not induce protection as seen from lesion size measurement and parasite load at the site of infection . However , parasite numbers in DLN in DNA-TRYP+MVA-TRYP immunized mice are significantly lower ( 11-fold ) than that of control mice . Therefore Pam3CSK4 plays a critical role in achieving protection against murine L . ( V ) . panamensis infection using heterologous prime-boost vaccination . To evaluate mechanisms underlying the protection selectively induced by DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunization , the immune responses in DLN cells from the L . ( V ) . panamensis infected mice ( control and immunized ) were examined . In the control group of mice , as found for chronic infection with L . ( V . ) panamensis a mixed cytokine response was observed ( IFN , IL-13 , IL-10 ) . This demonstrated an ongoing inflammatory-anti-inflammatory immune response concomitant with parasite persistence . In general , all TRYP vaccinated groups of mice produced lower levels of cytokines than the control group of mice ( vector ) . Although a reduction in IFNγ , as well as in IL-13 and IL-10 was observed in the DLNs of L . ( V ) . panamensis infected mice immunized with DNA-TRYP ( with or without Pam3CSK4 ) +MVA-TRYP in comparison to control mice ( Table 1 ) , the predominant effect was on the levels of IL-13 and IL-10 . Interestingly the levels of IFNγ observed for vaccine groups boosted with MVA-TRYP ( DNA-TRYP and DNA-TRYP ( Pam3CSK4 ) ) were comparable . However , reductions in both IL-13 and IL-10 occurred for the mice vaccinated with DNA-TRYP ( ±Pam3CSK4 ) +MVA-TRYP in comparison to those receiving DNA-TRYP alone , with the principal decrease being in IL-10 for the mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP . Although the infected vaccinated mice produced lower levels of cytokines than the control mice , it is notable that the relative levels of the cytokines differ between the various groups , with the highest IFNγ/IL-10 or IFNγ/IL-13 ratios observed for the DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP group of mice . The down-regulation of IL-10 as well as IL-13 is consistent with the memory responses observed prior to infectious challenge ( Figures 1 and 2 ) and suggests that lower levels of these cytokines are critical to parasite containment . These results are consistent with the roles of these cytokines in pathogenesis [22] . Further , these findings are similar to vaccine studies of L . major utilizing MVA vaccination [24] where IFNγ/IL-10 was found to be predictive of protection . The cytokine responses clearly changed with the mode of vaccination; however , it was unclear what the role of specific T cell populations might be in this process . The IFNγ responses and memory populations of both CD4 and CD8 T cells appeared to increase with TLR1/2 activation ( Figure 2 ) . To examine the specific contribution of effector CD8 and CD4 T cells to protection , vaccinated mice ( DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP ) were immunodepleted immediately prior to infection with L . ( V . ) panamensis . As seen in Figure 4A , CD8 T cell depletion significantly reversed protection and lesion development . Intriguingly , although lesion development was still somewhat restrained in the CD8 T cell depleted group in comparison to the vector control group , the parasite burdens in these two groups were comparable ( Figure 4B ) , indicating no control on parasite growth occurred in the absence of CD8 T cells . The importance of CD8 T cells to protection is consistent with the heightened in vitro memory CD8 T cell expansion in response to antigen observed ( Figure 2 ) of the mice primed with DNA-TRYP ( Pam3CSK4 ) in comparison to mice receiving DNA-TRYP alone . Interestingly based upon lesion development , the DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice depleted of CD4 T cells are significantly resistant to infection . These data are confirmed by parasite burden levels at the site of infection ( Figure 4B ) , which indicate significantly lower levels of parasites in the vaccinated mice deleted of CD4 T cells when compared to DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice . Overall , it appears that CD8 T cell effectors in the absence of CD4 T cells can provide significant protection against L . ( V . ) panamensis infection . Furthermore , CD4 T cell effector populations do not appear to significantly contribute to protection in the vaccinated mice . Analysis of cytokine production by DLN cells of the various groups of infected vaccinated mice indicated that the groups of protected mice ( DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP and DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP–CD4 depleted ) produced significantly lower levels of IFNγ , IL-13 and IL-10 ( Figure 5 ) , when compared to infected control mice . Mice immunized with DNA-TRYP+MVA-TRYP and notably mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP and depleted of CD8 cells had elevated levels of all three cytokines . Interestingly DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice depleted of CD4 cells produced lower levels of IL-13 and IL-10 when compared to DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP immunized mice . However , no difference in IFNγ levels was observed between these two groups . Reduced levels of IL-10 and IL-13 have been found to lead to control of L . ( V . ) panamensis infection [22] . Consequently , it appears that healing and resolution of infection is dependent upon effector CD8 T cells; overall activation of CD8 T cells results in lower levels of IL-13 and IL-10 as well as IFNγ produced in response to infection . Interestingly , the depletion of CD4 T cells leads to a further reduction of both IL-10 and IL-13 , whereas no change in the IFNγ response occurs . Together , these data demonstrate the essential role of CD8 cells in mediating protection induced by DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP against L . ( V . ) panamensis infection . Moreover , these results suggest that CD8 T cells may perform this function through the regulation of IL-13 and IL-10 production . However , further experiments are required to determine this point .
The Leishmania ( Viannia ) subgenus is phylogenetically divergent from other Leishmania [8] , [9] , [10] . Reflected in this is the fact that infection by members of this subgenus generates a hyperinflammatory and mixed Th1/Th2 response ( and high levels of IFNγ ) that can fail to resolve [11] , [12] , [13] , [14] , [15] . Furthermore vaccination against infection in the murine model has been challenging , as approaches utilizing conserved antigens previously shown to induce substantial protection against other Leishmania sp . have failed to provide protection [19] , [21] implying that novel immunization strategies are required . Recent studies [20] indicate that immunization with replication defective Leishmania can provide partial protection against L . ( V . ) braziliensis infection; but the protective mechanisms involved were not explored . Our results in part are consistent with these studies , as neither DNA vaccination alone nor heterologous prime-boost vaccination ( DNA+MVA ) ( using the conserved TRYP antigen ) leads to significant protection against L . ( V . ) panamensis infection/disease . Nonetheless , the prime-boost modality has been shown to provide protection against cutaneous leishmaniasis and even the more recalcitrant visceral leishmaniasis , when DNA vaccination alone failed [42] , [43] . In this report we demonstrate that significant protection against chronic L . ( V ) . panamensis infection can be achieved using a heterologous prime-boost ( DNA -MVA ) modality using aTLR2/1 ligand ( Pam3CSK4 ) as the adjuvant and a single defined antigen , TRYP . As expected in response to heterologous prime-boost vaccination , groups of mice boosted with MVA-TRYP produced high levels of IFNγ . However mice immunized with DNA-TRYP ( Pam3CSK4 ) +MVA-TRYP produced higher levels of IFNγ as well as significantly reduced levels of IL-13 and IL-10 when compared to mice immunized with DNA-TRYP+MVA-TRYP . This effect of Pam3CSK4 is consistent with initial experiments using the Pam3CSK4 adjuvant for DNA vaccination alone , thereby indicating the striking capacity of Pam3CSK4 during priming to down-regulate IL-13 and IL-10 responses , leading to a Th1 biased immune response . Interestingly Pam3CSK4 has been shown to induce a mixed cytokine response ( IL-12 , IL-10 , TNFα ) in mouse bone marrow derived dendritic cells [44] . However evidence indicates that during allergic inflammation , that enhanced Th1 responses are observed [45] in response to Pam3CSK4 stimulation . Consequently , the tissue site and context ( presence of other immunomodulators ) can impact the effect of an adjuvant on the developing immune response . Within the dermal compartment ( site of DNA vaccination ) skin resident DCs [46] , [47] are the probable target population of intradermally delivered DNA-TRYP and Pam3CSK4 [48] and TLR9 ( bacterial DNA-CpG ) /TLR2 activation . Costimulation ( TLR , NOD or TCR of NK-T cells [32] , [49] , [50] ) of dendritic , macrophage , and early responding cells may further cooperate and selectively drive/amplify specific responses [44] , [51] . DNA immunization and activation of TLR9 results in strong IL-12 production , leading to a predominant Th1 immune response [52] . Interestingly , both synergy and co-operation between TLR2 and TLR9 have been observed in response to infection [51] , [53] , [54] resulting in heightened Th1-like responses . Joint TLR9 and TLR2 engagement has also been shown to result in the production of MCP-1 and synergistic production of RANTES [55] , which would lead to increased recruitment of macrophages/monocytes , T cells , and dendritic cells [56] . This increased cellular recruitment could potentially lead to enhanced T cell expansion , as observed herein for both CD4+ and CD8+ T cells . Further , TLR2 receptors are present on T cells and therefore TLR2 activation could also modulate the developing adaptive immune response . TLR2 activation of Tregs has been shown to mitigate ( at least temporally ) their suppressive quality [57] , [58] , [59] , which could lead to increased proliferation and expansion of antigen specific T effector cells . CD8 T cell activation has been related to both healing and pathogenesis in leishmaniasis ( reviewed in [60] ) . In the human immune response to Leishmania ( Viannia ) infection , studies implicate CD8 T cells in disease resolution [61] , [62] as well as pathology [63] , [64] . The variation in the observed effects found for CD8+ T cells may reflect the functional heterogeneity of these cells [65] , [66] , [67] . CD8 T cells have been shown to exert a curative role in murine models of leishmaniasis , which has been attributed to the production of IFNγ [60] , [68] , [69] , [70] as well as a potential role for perforin ( CTL function ) [71] . Further , other mediators produced by CD8 T cells ( granzymes , chemokines , cytokines ) may also contribute to the host defense . Although murine studies have unambiguously demonstrated that CD8 cells participate in vaccine-induced protection against infection [32] , [68] , [72] caused by other Leishmania species , the contributions to host defense against Leishmania ( Viannia ) have not been previously determined . While both CD4 and CD8 T cell responses ( memory and IFNγ ) were increased in response to vaccination using TLR1/2 , protection was largely due to the CD8 T effector cell response , as protection was reversed in mice depleted of CD8 cells , but not upon CD4 cell depletion . These results differ from heterologous prime-boost vaccination studies utilizing NK-T activation during DNA priming [32] , where CD4 T responses appeared responsible for anti-leishmanial response [32] . The overall reason that CD8 T cell were the primary effectors of protection is not clear and may be due to the antigen and/or adjuvant utilized for these studies as well as the specific modes of action/effector mechanisms of the CD8 T cells elicited . It is of interest that TryP has been found in the leishmanial exosome [73] , [74] . Consequently , it may be that the preferential mode of action found by CD8 T ( in spite of a clear CD4 T cell response ) is biased by the release and subcellular localization of this antigen during infection . However , further work is required to determine this point and the potential of exosomal antigens as CD8 vaccine candidates . Alternately , other studies employing Pam3CSK4 as an adjuvant clearly demonstrate its ability to enhance CD8 responses [75] , [76] , [77] . Our findings are consistent with these observations . Long-term memory ( 12 weeks after the vaccinia boost ) development in immunized mice was heightened in mice receiving Pam3CSK4 , with a marked expansion of the CD8 T cell population . Although the precise mechanism by which Pam3CSK4 augments the CD8 T cell response is not completely understood , Pam3CSK4 has been shown to enhance dendritic cell cross presentation to CD8 T cells [78] . Further , reports show that Pam3CSK4 engages TLR2 on CD8 cells [79] , prolonging their survival and increasing proliferation that may contribute to increased frequency of antigen-specific CD8 cells . Furthermore , TLR2 engagement on DC subsets has been reported to lead to enhanced trafficking of these cells to the draining lymph nodes [80] , [81] , which would further contribute to the development of enhanced CD8 response . We have recently shown a role for IL-13 and IL-10 in determining disease progression in the murine model of L . ( V ) . panamensis [22] infection . Results presented here are consistent with this finding and further demonstrate a correlation between parasite load and levels of these cytokines . Unexpectedly , depletion of effector CD4 T cells did not appear to influence resistance . CD4 cells appear to be a source of IL-13 and IL-10 in the vaccinated mice , as observed in FACS analyses and the fact that CD4 cellular depletion results in further reduction of IL-13 and IL-10 . In contrast , CD8 T effector cell depletion reversed protection and led to the production of higher levels of both IL-10 and IL-13 . Overall , the consequence of CD8 cell activation was to modulate protection by limiting IL-10 and IL-13 . CD8 T cells have been shown to modulate Th2 CD4 T cell function both through direct T-T interaction [82] or the indirect consequence of CD8 activation on other cell populations [83] , [84] . Similarly CD8 T cell have been observed in the case of L . major [69] to direct increase levels of CD4-IFNγ production . Our results are consistent with the possibility that CD8 T cells are involved in the immune “deviation” of CD4 T cells and that this is involved in the development of protection against infection . However , further work is required to determine this point . An understanding of the mechanisms by which CD8 T cells promote host defense and mediate protection is of obvious interest for vaccine development against L . ( Viannia ) . This is the first report of successful protection against L . ( V . ) panamensis using a single antigen . By utilizing a TLR2 agonist ( Pam3CSK4 ) in a heterologous prime boost immunization method we have demonstrated protection against L . ( V . ) panamensis in a murine model . This protection was achieved specifically through the expansion of antigen-specific effector CD8 T cells . These findings suggest that the modulation of TLR1/2 signaling may dramatically improve the efficacy of DNA-based vaccine modalities , especially where CD8 T cell activation is critical , thereby contributing to effective and affordable anti parasitic vaccines .
|
Leishmania ( Viannia ) are the predominant agents of leishmaniasis in Latin America . Given the fact that leishmaniasis is a zoonosis , eradication is unlikely; a vaccine could provide effective prevention of disease . However , these parasites present a challenge and we do not fully understand what elements of the host immune defense prevent disease . We examined the ability of vaccination to protect against L . ( Viannia ) infection using the highly immunogenic heterologous prime-boost ( DNA-modified vaccinia virus ) modality and a single Leishmania antigen ( TRYP ) . Although this mode of vaccination can induce protection against other leishmaniases ( cutaneous , visceral ) , no protection was observed against L . ( V . ) panamensis . However , we found that if the vaccination was modified and the innate immune response was activated through Toll-like receptor1/2 ( TLR1/2 ) during the DNA priming , vaccinated mice were protected . Protection was dependent on CD8 T cells . Vaccinated mice had higher CD8 T cell responses and decreased levels of cytokines known to promote infection . Given the long-term persistence of CD8 T cell memory , these findings are encouraging for vaccine development . Further , these results suggest that modulation of TLR1/2 signaling could improve the efficacy of DNA-based vaccines , especially where CD8 T cell activation is critical , thereby contributing to effective and affordable anti parasitic vaccines .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"immunity",
"biology",
"microbiology",
"parasitology",
"pathogenesis"
] |
2011
|
TLR1/2 Activation during Heterologous Prime-Boost Vaccination (DNA-MVA) Enhances CD8+ T Cell Responses Providing Protection against Leishmania (Viannia)
|
Leptospirosis is a bacterial zoonotic disease of worldwide importance , though relatively neglected in many African countries including sub Saharan Africa that is among areas with high burden of this disease . The disease is often mistaken for other febrile illnesses such as dengue , malaria , rickettsioses and enteric fever . Leptospirosis is an occupational disease likely to affect people working in environments prone to infestation with rodents which are the primary reservoir hosts of this disease . Some of the populations at risk include: sugarcane plantation workers , wetland farmers , fishermen and abattoir workers . In this study we investigated the prevalence of antibodies against Leptospira among sugarcane plantation and factory workers , fishing communities as well as among rodents and shrews in domestic and peridomestic environments within the study areas . The study was conducted in Kagera region , northwestern Tanzania and it involved sugarcane plantation workers ( cutters and weeders ) , sugar factory workers and the fishing community at Kagera Sugar Company in Missenyi district and Musira island in Lake Victoria , Kagera , respectively . Blood was collected from consenting human adults , and from rodents and shrews ( insectivores ) captured live using Sherman traps . Serological detection of leptospiral antibodies in blood serum was carried out by the microscopic agglutination test ( MAT ) . A total of 455 participants were recruited from the sugarcane plantation ( n = 401 ) and fishing community ( n = 54 ) while 31 rodents and shrews were captured . The overall prevalence of antibodies against Leptospira in human was 15 . 8% . Sugarcane cutters had higher seroprevalence than other sugar factory workers . Prevalent antibodies against Leptospira serovars in humans were against serovars Lora ( 6 . 8% ) , Sokoine ( 5 . 3% ) , Pomona ( 2 . 4% ) , Hebdomadis ( 1 . 1% ) and Kenya ( 0 . 2% ) . Detected leptospiral serovars in reservoir hosts were Sokoine ( 12 . 5% ) and Grippotyphosa ( 4 . 2% ) . Serovar Sokoine was detected both in humans and small mammals . Leptospirosis is a public health threat affecting populations at risk , such as sugarcane plantation workers and fishing communities . Public awareness targeting risk occupational groups is much needed for mitigation of leptospirosis in the study areas and other vulnerable populations in Tanzania and elsewhere .
Leptospirosis is a public health concern especially in the tropical and subtropical countries where the environment is optimal for survival of pathogenic leptospires [1] . The annual morbidity and mortality caused by leptospirosis worldwide is estimated to be 14 . 7 cases per 100 , 000 population [2] . Globally , Oceania region has the highest disease burden ( 150 . 6 cases/100 , 000 population ) , South east Asia ( 55 . 5 ) , Caribbean ( 50 . 6 ) and East Sub Saharan Africa ( 25 . 6 ) [2 , 3] . In Tanzania the annual incidence is 75–102 cases per 100 , 000 population [4] . Rodents are considered major reservoirs of Leptospira [5] and other wild animals and birds found in wetland areas may also carry and spread leptospires into the surroundings [6] . The disease is associated with certain occupational activities such as rice and sugarcane farming , fishing and fish farming , livestock keeping , handling animal products and water sports [7 , 8] . Males are most affected than females contributing to 80% of the total burden [3] . Humans can be infected through contact with urine or other materials from infected animals or contaminated water and soil [9] . In Tanzania , leptospirosis has been reported in patients with non-malaria fevers [10 , 11] and in animals including rodents and domestic animals [12–15] . Antibodies against Leptospira have been demonstrated also in freshwater fishes [6] in Tanzania suggesting potential risk to fishermen and people undertaking irrigation activities such as , rice farming and sugarcane plantation . Studies on leptospirosis in these at risk populations are lacking , hence the burden of leptospirosis in fishing communities and sugarcane plantations is not known . Sugarcane plantation and rice farming are important agricultural sectors in Tanzania , which engage permanent and seasonal workers from different parts of the country . Understanding the burden of leptospirosis in these occupational groups could provide baseline information needed for informing policy , especially because the disease is neglected and rarely considered for diagnosis in the health system [16] . In this study we investigated the serological prevalence of leptospirosis in selected risk populations of sugarcane plantation workers and fishing in northwestern Tanzania Also , we identified potential Leptospira serovars circulating in the region , which would serve as important antigens for diagnostic purposes .
This study was conducted in Kagera region northwestern Tanzania at Musira island ( S 01° 19 . 914’ , E 031° 49 . 772’ with elevation of 1120 meter above sea level , and at Kagera sugar company ( S 01° 12 . 807 , E 031° 16 . 510’ ) with elevation of 1157 meter above sea level . Kagera region receives bi-modal rainfall pattern ranging between 900–2 , 000 mm per annum , temperatures range between 20°C and 28°C . Kagera region is located along Lake Victoria hence fishing is among major socio-economic activity apart from large scale sugarcane plantation . Kagera Sugar Company is one of the biggest sugarcane plantations in the country . The two study sites ( fishing community and Musira and sugarcane plantation community at Kagera sugar company are 76 . 2 km apart ( Fig 1 ) . The ethical clearance for conducting this study was obtained from the Medical Research Coordinating Committee of the National Institute for Medical Research ( NIMR ) , Certificate No . NIMR/HQ/R . 8a/Vol . IX/2453 , as well as from the Kilimanjaro Christian Medical University College , Research Ethics and Review Committee ( CRERC ) , Moshi Tanzania ( Ref . No . 993 ) . Permission was also sought from local authorities in the study area .
A total of 455 participants were sampled of which 401 ( 132 females and 269 males ) were from sugarcane plantation and 54 ( 16 females and 38 males were from the Musira fishing island in Lake Victoria . The demographic profile of human participants was as shown in Table 1 . The majority of rodents were Rattus spp . ( 55% ) trapped indoors . Other rodents trapped included forest species ( Lophuromys spp . ) captured in the bushes near the sugarcane plantation and Arvicanthis spp . found in fallow land near sugarcane fields ( Table 2 ) . Prevalence of human leptospirosis in the two study populations of sugarcane plantation workers and fishing community was 15 . 8% . Fifty eight of the 317 ( 18 . 3% ) sugarcane cutters were seropositive compared to 8 out of 54 ( 14 . 8% ) of the fishing community subjects . Two of the 13 ( 15 . 4% ) hospitalized patients were seropositive while other participants including office cleaners , petty traders and security guards contributed to 7 . 0% seropositivity ( Table 1 ) . Antibodies against tested Leptospira were relatively lower in rodents than in humans . The highest titre ( 1:2560 ) was observed in two individuals against serovar Pomona ( Fig 2 ) . Prevalence of antibodies against Leptospira among different occupational groups , populations , gender and age groups were compared to determine whether certain groups were at more risk than others . The prevalence of antibodies against Leptospira between male and female participants was 6 . 4% , which was not statistically significant ( p = 0 . 0800 , 95% CI = -0 . 8658 to 12 . 6811 , χ2 = 3 . 065 , df = 1 ) . The prevalence of antibodies against Leptospira in participants in age group of 18–37 year and 38–57 year old differed by 2 . 2% which was not statistically significant ( p = 0 . 6206 , 95% CI = -5 . 5629 to 12 . 5150 , χ2 = 0 . 245 , df = 1 ) . There was significant difference in percentage of positive individuals ( 40% ) between participants in age group of 18–37 year and ≥58 year old ( 40 . 0% ) ( p = 0 . 0003 , 95% CI = 13 . 1776 to 64 . 4124 , χ2 = 12 . 898 , df = 1 ) . Age group of 38–57 yrs and >58 year old also showed significant difference in percentage of positives ( 37 . 8% ) that was also statistically significant ( p = 0 . 0044 , 95% CI = 9 . 5238 to 62 . 8984 , χ2 = 8 . 105 , df = 1 ) . The prevalence of antileptospiral antibodies between fishing community and sugarcane cutters was 3 . 6% that is not statistically significant ( p = 0 . 5241 , 95% CI = -8 . 8082 to 12 . 0938 , χ2 = 0 . 406 , df = 1 ) . Comparison of positive rate found in fishing community and unexposed group ( others ) showed 9 . 3% , which was not statistically significant ( p = 0 . 0777 , 95% CI = -1 . 2424 to 21 . 5463 , χ2 = 3 . 111 , df = 1 ) . The difference in positive rate between sugarcane cutters and unexposed group ( others ) was 12 . 9% , which was statistically significant ( p = 0 . 0068 , 95% CI = 4 . 1963 to 18 . 6242 , χ2 = 7 . 329 , df = 1 ) . Hospitalized participants and unexposed group ( others ) showed a difference of 9 . 9% that was not statistically significant ( p = 0 . 1999 , 95% CI = -3 . 6319 to 36 . 9580 , χ2 = 1 . 643 , df = 1 ) . Hospitalized participants and hospital staff also showed a difference of 1 . 1% that was not significant ( p = 0 . 9490 , 95% CI = -37 . 5454 to 30 . 4008 , χ2 = 0 . 004 , df = 1 ) .
This study shows high prevalence of antibodies against Leptospira in humans involved in sugar production and fishing in the Kagera region , northwestern Tanzania . Leptospirosis in rodents and shrews captured in the areas is also reported . Findings suggests that sugarcane plantation workers especially sugarcane cutters and fishing communities are potentially at risk . A prevalence of 15 . 8% was found in sugarcane plantation workers , with cane cutters having the higher prevalence of 18 . 4% , followed by other plantation workers and hospitalized patients . Prevalence of anti-leptospiral antibodies was also high ( 14 . 8% ) in fishermen and other individuals living on the Musira island , which is a fishing island . This suggests that fishing communities can get leptospirosis following contact with water contaminated with urine of the reservoir hosts . The prevalence of human leptospirosis in sugarcane plantation workers reported in this study ( 18 . 4% ) is lower than that reported in sugarcane plantation workers in central America ( 59% ) [19] but higher than the 0 . 7% prevalence reported from Trinidad and Tobago [23] . Prevalence of antibodies against Leptospira among different occupational groups , populations , gender and age groups showed variations suggesting that individuals belonging to certain groups and occupation groups have different levels of risk of contracting leptospirosis . For example , while there was no significant difference in the prevalence of leptospiral antibodies between male and females despite that the study had more males than females due to the nature of the occupation of the study populations , there was a significant difference in prevalence of antibodies against Leptospira found in participants in two age groups of 18–37; 38–57 year old versus participants with age above 58 year old . Findings show that participants with over 58 year old have significantly higher proportion of antibodies against Leptospira than those with age below 58 year old ( i . e . 18–37; 38–57 year old ) . This could be probably associated with potential prolonged exposure to risk environment such as sugarcane cutting for many years than newer entrants . The fishing community and sugarcane plantations appear to have similar risk levels since the prevalence of antibodies in these two populations was not statistically significant . However , comparison of fishing community and sugarcane cutters considered risk populations with unexposed groups consisting of participants engaged with less risk activities such as office work , security and petty traders show that fishing community does not differ to the unexposed group while sugarcane cutters show more risk than unexposed group . This can be explained with fact that fishing community included the general population of the fishing island including school pupils and other residents likely to have various levels of risk of contracting leptospirosis while sugarcane cutters consisted a uniform group of individuals engaged with same activity of cutting sugar hence likely to have same level of risk higher than the general population . The prevalent antibodies against Leptospira serovars found in humans were against Leptospira interrogans serovar Lora ( 6 . 8% ) , L . kirschneri serovar Sokoine ( 5 . 3% ) and slightly Leptospira interrogans serovar Pomona ( 2 . 4% ) . Leptospira interrogans serovar Hebdomadis and L . borgpetersenii serovar Kenya were least found with prevalence of 1 . 1% and 0 . 2% , respectively . Leptospira kirschneri serovar Sokoine and L . kirschneri serovar Grippotyphosa were frequently found in both humans and animals as previously reported [8 , 15] in agro-pastoralists communities living in Katavi-Rukwa ecosystem [8] indicating a wider distribution of leptospirosis in Tanzania . These findings shows that the roof rat ( Rattus spp . ) is an important reservoir of leptospirosis in Kagera region as demonstrated by high positivity rate among the house rats collected in different localities in the study areas . Comparison of positive rates found in the roof rats and an insectivore showed no statistically significant difference due to small sample size of rodents and shrews collected . A larger sample size estimated for this study was not achieved due to seasonal variations in rodent populations hence suggesting further sampling to enhance robust determination of the major reservoir of Leptospira in this region . Antibodies against L . serovar Kenya , Lora , Pomona and Hebdomadis were not detected in rodents nor insectivores . The rats were seropositive against L . kirschneri serovars Sokoine and L . kirschneri serovar Grippotyphosa . Rodents had lower antibody titres ( 1:20–1:40 ) than humans in which higher titres up to 1:2560 were determined by MAT which is the gold standard test for leptospirosis diagnosis [9 , 24] . High antibody titres against Leptospira serovars detected in humans suggest the existence of recent infections . The predominant circulating Leptospira serovars which antibodies against was detected in humans , namely Leptospira interrogans serovar Lora , L . kirschneri serovar Sokoine , L . interrogans serovar Pomona , L . interrogans serovar Hebdomadis and L . borgpetersenii serovar Kenya have been previously reported in humans , rodents and domestic animals [10 , 15 , 25] . Leptospira kirschneri serovar Sokoine was mainly found in both humans and animals in Tanzania whereas L . interrogans serovar Grippotyphosa was mainly detected in the reservoir hosts . Leptospira interrogans serovar Lora was not detected in rodents , indicating potential diversity of sources of human infection . It is known that certain Leptospira serovars demonstrate host-specificity and might be absent in certain rodent species [15] . Further investigations are needed to establish the source or reservoir hosts of serovar Lora in the study areas and to determine whether the plantation workers who also come from outside Kagera had leptospirosis exposure prior to their recruitment at the sugarcane plantation . This could be achieved by including leptospirosis screening during general health examinations performed before recruiting cane cutters . The observed high prevalence of leptospirosis in the fishing community corroborate previous report of high seropositivity/leptospiral antibodies in freshwater fishes and thus potential risk of leptospirosis in fishing communities and in people working in the fishing industry [15 , 19 , 23] . It is recommended that leptospirosis control should include rodent management , and public awareness . Furthermore , leptospirosis screening should be introduced in risk occupational groups in Tanzania and elsewhere where the disease is neglected [16] . Detection of leptospiral antibodies in hospitalized patients during this study indicates further the importance of considering leptospirosis among febrile illnesses that are non-malarial . The prevalence of 15 . 4% of leptospirosis in hospitalized patients corroborates previous reports from northern Tanzania and Morogoro among hospitalized patients with febrile illness [10 , 11 , 26] . This further emphasizes the need to include leptospirosis in differential diagnosis of febrile illnesses . Further surveillance studies are needed to isolate and characterize the disease causative Leptospira serovars beyond serological surveillance . These should include cross agglutination absorption test , and molecular typing [25] . The major limitations of this study were failure to isolate the causal agent , which would have enabled its characterization . Similarly , future studies should include larger populations of potential reservoirs . Leptospirosis is a public health threat in sugarcane plantation workers and the fishing communities . Preventive measures are needed to mitigate risks of leptospirosis . These should include rodent control , public awareness and screening for leptospirosis in individuals with non-malarial fevers [16] and vulnerable occupational groups such as sugarcane cutters . Leptospira serovars Lora , Sokoine , Pomona , Hebdomadis , Kenya and Grippotyphosa should be included as antigens for broader leptospirosis screening in humans and animals from this region .
|
Leptospirosis is caused by a spirochete bacterium of the genus Leptospira . The disease is worldwide distributed , although highly neglected in some parts of the developing world . This study investigated the prevalence of antibodies against Leptospira in sugarcane plantation workers , a fishing community and rodents and shrews in the Kagera region , northwestern Tanzania . Seventy two of the 455 ( 15 . 8% ) screened participants were seropositive to leptospirosis . The seroprevalence was higher among sugarcane cutters ( 18 . 4% ) than other plantation workers , and 15 . 4% of hospitalized patients in the plantation hospital were seropositive . Prevalence of antibodies against Leptospira in the fishing community was 14 . 8% . Antibodies against Leptospira serovars detected in humans with their respective proportions , in brackets , were Lora ( 6 . 8% ) , Sokoine ( 5 . 3% ) , Pomona ( 2 . 4% ) , Hebdomadis ( 1 . 1% ) and Kenya ( 0 . 2% ) . In the small mammals , the most detected antibodies were against Leptospira serovar Sokoine ( 12 . 5% ) and serovar Grippotyphosa ( 4 . 2% ) . These results show that leptospirosis is public health risk requiring attention of the health system as well as the agricultural sector for its management .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"leptospira",
"pathology",
"and",
"laboratory",
"medicine",
"shrews",
"pathogens",
"sociology",
"tropical",
"diseases",
"microbiology",
"vertebrates",
"social",
"sciences",
"geographical",
"locations",
"animals",
"mammals",
"social",
"systems",
"bacterial",
"diseases",
"tanzania",
"crops",
"neglected",
"tropical",
"diseases",
"plants",
"bacteria",
"bacterial",
"pathogens",
"africa",
"sugarcane",
"infectious",
"diseases",
"grasses",
"zoonoses",
"crop",
"science",
"medical",
"microbiology",
"microbial",
"pathogens",
"leptospirosis",
"agriculture",
"people",
"and",
"places",
"rodents",
"eukaryota",
"biology",
"and",
"life",
"sciences",
"leptospira",
"interrogans",
"amniotes",
"organisms"
] |
2019
|
Leptospirosis in sugarcane plantation and fishing communities in Kagera northwestern Tanzania
|
Nutation is an oscillatory movement that plants display during their development . Despite its ubiquity among plants movements , the relation between the observed movement and the underlying biological mechanisms remains unclear . Here we show that the kinematics of the full organ in 3D give a simple picture of plant nutation , where the orientation of the curvature along the main axis of the organ aligns with the direction of maximal differential growth . Within this framework we reexamine the validity of widely used experimental measurements of the apical tip as markers of growth dynamics . We show that though this relation is correct under certain conditions , it does not generally hold , and is not sufficient to uncover the specific role of each mechanism . As an example we re-interpret previously measured experimental observations using our model .
During their development , plant organs display a large range of movements . These movements may be broadly divided into two classes; tropisms and nastic movements . Tropisms are the reorientation towards an external stimulus , e . g . light or gravity [1 , 2] . Nastic movements account for endogenous , autonomous movements and are not directed towards an external stimulus . Nutation , often called circumnutation , is a particular class of nastic movements , in which the plant organ successively bends in different directions , resulting in an apparent oscillatory swinging motion . Despite its ubiquity among plant movements nutation has not been studied as extensively as tropisms , and the mechanism responsible for this movement , as well as its regulation , remain unclear ( see [3–5] for a review ) . Current theories or concepts of nutational mechanisms generally fall into two categories [6] . The first suggests the influence of external drivers such as gravity or light , where the movement stems from an overshoot during the straightening of the plant in response to the direction of gravity ( or light ) . The second assumes an endogenous driver such as an oscillator , suggested by Darwin [7] , possibly related to the growth process itself [3 , 8] . Studies have shown that though gravitropism may influence and modify the observed movements , the two processes exist independently [3 , 9 , 10] , consistent with symmetry arguments which indicate that gravitropism alone cannot induce movement outside of the plane defined by the main axis of the plant and the direction of gravity [11 , 12] . Together with the fact that nutation is observed in the absence of light , this suggests that external cues cannot drive nutation . Interestingly , Brown [3 , 5] postulated that since nutation does not present any significant evolutionary benefit , it may be the consequence of some fundamental mechanism in the growth process . Observations of rice coleoptile mutations ( lazy ) that grow normally yet do not exhibit nutation [10] , suggest that growth alone may not be sufficient to generate nutation [3 , 8] . Therefore based on the present literature , the strongest hypothesis remains a growth-driven endogenous oscillator . We note that curvature of an elongated organ in three dimensional space can result from two different growth mechanisms , namely bending and torsion . Bending can result from the differential growth of the opposite sides of an organ [13 , 14] , i . e an initially straight organ will bend towards the direction of minimal growth . Studies have mainly focused on the case where the organ is curved in the same ( vertical ) plane as that of the differential growth , restricting movement to that plane only . However the plane of curvature should change when it is not in the same plane as that of the differential growth , producing movement in the horizontal ( apical ) plane . It is instructive to note that lines drawn on the surface of the organ , parallel to its main axis , will remain parallel to the main axis during its growth regardless of the direction of curvature , as shown in Fig 1 . A . The second mechanism , torsion , is responsible for the movement of twining plants [15 , 16] , and is due to the helical arrangement of cells around the main axis of the plant , possibly due to the torsional arrangement of cellulose [17] . In this case parallel lines drawn on the surface of the organ will take a helical form around the organ during its growth , as shown in Fig 1 , e . g . the cotyledon on top of a hypocotyl will rotate . However this process can lead to a 3D curved organ only if the organ is already initially curved , and furthermore results in a helical form ( see Fig 1 . C ) . Moreover , observations of torsion in nutating plant organs have been found to be too slow to account for the observed nutation [18 , 19] . These observations hint that the dominant growth mechanism underlying nutation is differential growth under the action of an internal oscillator [19] . This internal oscillator could then be related to the auxin dynamics or the sensitivity of the membrane to auxin . Indeed a relation has been found between oscillations in ion fluxes and nutation [20 , 21] . Moreover there are some reports of relationships between nutation and biological rhythms [22 , 23] , demonstrating genetically that the circadian clock controls nutation speed [24] . Together , these results suggest that genetically regulated rhythmical membrane transport processes are central to plant nutation , and may play the role of an internal oscillator [5] . In this study we consider nutation as a growth-driven process , in line with previous work on tropisms and differential growth [11–14] where mechanical effects such as buckling and instabilities are disregarded . We then focus on the relation between internal oscillatory growth patterns and the observed movement . Attempts have already been made to develop a mathematical model of nutation , but the full three-dimensional geometry of the organ has been neglected , resulting in an incomplete kinematic description [25] . The existing models account only for the kinematics of the apical tip , and this has been shown to be insufficient to understand the underlying mechanisms , since geometrical and local effects are neglected [1 , 13] . A similar problem exists in the experimental measurement of nutation , where the full dynamics of plants in three-dimensional space and time are rarely taken into account . It is common , instead , to track the projection of the apical part of the organ in the plane orthogonal to the gravitational field ( defined here as the horizontal plane Pa ) [18 , 26] . Such measurements carried out on different species and organs [7 , 26] exhibit disorganized patterns ( zig-zag shaped ) , and organized patterns ( for instance , elliptical patterns , and the limits of these patterns , e . g . a circle or a line ) [8 , 18 , 22 , 27 , 28] . The interpretation of these measurements remains unclear , since the relation between the differential growth pattern and the kinematics in space and time is not clearly defined . Here we will state a growth-driven parsimonious model couched within a three-dimensional geometrical framework , accounting for observed classes of movement patterns . The analysis is done both analytically and through numerical simulations . The model’s limitations are also discussed . Lastly , the model is applied to existing experimental observations , and the relevance of apical ( horizontal ) measurements is discussed . The details of all calculations are given in the appendix . In addition , an interactive simulator is available online [29] . Predefined solutions are accessible through the numerical key of the keyboard and are referenced throughout the manuscript .
The geometric framework used to describe the kinematics of tropisms [11 , 12 , 14] is only sufficient to describe growing elongated organs in a single plane and is therefore inadequate here . Unlike the movements observed for example in gravitropism where the curve is constrained to a unique plane , in nutation the organ is curved along different planes in 3D space ( Fig 1 . A ) . We start by introducing a few assumptions and definitions which will be essential for the construction of our three dimensional model ( Fig 1 ) . The organ is assumed to be cylindrical with a constant radius R along the organ . It is assumed that no shear growth is observed , so the cross section remains in plane . The organ is described by the curvilinear abscissa s along the median line . Each point at the surface of the organ is then defined via cylindrical coordinates ( s , ϕ ) where s is its position along the abscissa , and ϕ is the angle of the point on the cross section , relative to an arbitrarily chosen direction . This description is depicted in Fig 2 . A and in S1 Video . In order to fully describe the curvature of an organ curved in an arbitrary direction in space , it is first necessary to define two vectors: t , the tangent to the median line , and c , the normal ( perpendicular ) to the median line , as shown in Fig 2 . B . The orientation of the latter in the cross section , ψc ( t ) , is in the same plane as the principal direction of curvature ( see Fig 3 ) . This means that for each element of the curve , the curvature is maximal in the plane defined by the vectors t and c ( see Fig 2 . B ) . Fig 2 . C shows a cross section of the shoot , by definition in the plane orthogonal to t , defining the orientation ψc of the principal direction of curvature c .
In what follows we re-examine existing experimental observations in the context of our model . We analyze different classes of movements recorded in the horizontal plane , examining possible underlying mechanisms . We first consider the simulated apical trajectories of the most common observed patterns , the circle and the ellipse . Fig 6 presents the underlying form of the variation of the principal direction of growth d ψ g ( t ) d t and its differential growth Δ ( ψ g ( t ) , t ) E ˙ ( t ) , as imposed by the model via eqs 20 and 21 . In the case when the apical tip draws a circle in the horizontal plane , it follows that d ψ g ( t ) d t and Δ ( ψ g ( t ) , t ) E ˙ ( t ) are constant in time ( shown in Fig 6A ) . In the case of an ellipse , there are two possible mechanisms: ( i ) a periodic d ψ g ( t ) d t with maxima at some ψ g 0 and ψ g 0 + π ( Fig 6B ) , meaning that the direction of differential growth changes faster on opposite sides of the organ , giving less time for the organ to curve out in those directions , resulting in a smaller radius at those ends . ( ii ) a periodic Δ ( ψ g ( t ) , t ) E ˙ ( t ) ( Fig 6C ) with maxima at some ψ g 0 + π / 2 and ψ g 0 + π / 2 , meaning that the differential growth is larger at opposite sides of the organ , resulting in a greater curvature and therefore also larger radius at those ends ( simulations giving rise to these patterns are given in S3 , S4 and S5 Videos ) . For the sake of intuition , Fig 6 presents limit cases where either d ψ g ( t ) d t or Δ ( ψ g ( t ) , t ) E ˙ ( t ) are periodic while the other is constant , however in reality both may be periodic , and their relative magnitude and shift in phase will dictate the final apical pattern . We also note that taking the average value of d ψ g ( t ) d t yields the time it takes for the direction of differential growth to make a full rotation of the organ , i . e . T r = 2 π / ⟨ d ψ g ( t ) d t ⟩ . ( 22 ) If d ψ g ( t ) d t itself exhibits periodicity , we expect the time between two maxima to coincide with this period , since consecutive maxima are expected to be an angle of π apart . Moreover , we note that d ψ g ( t ) d t and Δ ( ψ g ( t ) , t ) E ˙ ( t ) are plotted here as a function of ψg , assuming the organ does not exhibit torsion , i . e . the rotation of the cotyledon on top of the organ’s movement , in which case the behavior would be shifted leading to erroneous conclusions . On the other hand a periodic or constant behavior would still be observed when plotting these values as a function of time . Lastly , as the sign of Δ ( ψ g ( t ) , t ) E ˙ ( t ) does not change , one cannot distinguish between variation of the median elongation , which is supposed to be positive , and variation of the differential growth term . Let us now consider the effect of elongation on the observed patterns . In the case where the whole organ is growing , L < Lgz , the curvature does not change as the organ grows , but the increasing length of the organ results in a spiral ( Fig 7 . A ) . The time to make a full turn remains unchanged , as Eq 22 is independent of the length of the organ ( see [29]—key 4 ) . In the case where L > Lgz , the pattern remains circular , however the part of the organ outside of the growth zone is fixed in a helical pattern , and the organ is curved in different directions . The observed circular pattern will exhibit a drift . If this helix is small , CLgz ≪ R/Lgz , it may not be noticeable experimentally ( see [29]—key 5 ) . No experimental account of this kind of helical pattern has been reported , suggesting a strong regulation of the curvature . Following the case of gravitropism [14] , proprioception is a good candidate for curvature regulation . A proprioceptive term can easily be added to Eq 13: Δ ‖ ( s , t ) = - γ C + Δ ψ g ( s , t ) sin ( ψ g ( s , t ) - ψ c ( s , t ) ) ( 23 ) The results obtained for a circular pattern are then slightly modified ( Fig 7 . B ) . Here the curvature of the organ is reduced by proprioception , which tends to straighten the organ [11] . The observed pattern is shifted around the base of the organ , in order to reduce the maximal curvature reached by the organ . The apical tip now converges towards a single stable orbit centered around the base that is fully independent of the initial conditions ( see [29]—key 3 ) . Due to the evidence that proprioception prevents fixed curvature in the case of gravitropism [14] , it is reasonable to postulate that such regulation is also sufficient in the case of nutation . Some experimental observations have shown the existence of epi- and hypo-trochoid patterns ( spirograph pattern ) [22] . These patterns provide an interesting case where the validity of hypothesis H1 ( which assumes no local effects ) is put in question . Mathematically , a trochoid is constructed as a sum of linear oscillators . If two segments of the organ of length L1 and L2 possess different temporal behaviors of the orientation of differential growth ψg1 ( t ) = ω1t and ψg2 ( t ) = ω2t , a trochoid will then be observed in the horizontal plane , as shown in Fig 8 , and from the simulations presented in S6 , S7 and S8 Videos . Applying Eq 20 to the apical curves cannot discern between the two separate oscillators , and will therefore result in an effective ψg ( t ) . Furthermore , the sign of the effective d ψ g ( t ) d t is dominated by the faster oscillator . We now analyze an existing dataset of apical movements of 8 Arabidopsis thaliana plants published by Stolarz et al . [18] ( see Fig 9A , 9B , 9C and 9F for examples of measured apical patterns . Most of the observed patterns are elliptical . We apply eqs 20 and 21 on all 8 measurements of the apical tip in the horizontal plane ( xa , ya ) , extracting ψg ( t ) ( shown in Fig 9D ) and Δ ( ψ g ( t ) , t ) E ˙ ( t ) . After ∼20 hours most plants exhibit a linear behavior , equivalent to a constant time derivative d ψ g ( t ) d t . Averaging over time and over the different plants results in 〈 d ψ g ( t ) d t 〉 = 4 . 6 × 10 - 4 ± 7 × 10 - 5 s - 1 , and substituting this in Eq 22 gives the time taken for a full rotation of the differential growth direction , Tr = 230 ± 40 min . At this point we focus on a single curve ( plotted in dark blue ) , where no torsion has been observed . Examining dψgdt for closely ( plotted in Fig 9E ) we identify oscillations . As mentioned earlier , the model predicts the maxima to be Tr/2 apart , representing opposite points along the apical curve . Indeed we find that the average time between every other maximum is To = 215 ± 25 min , in agreement with Tr found from the average value of dψgdt . Moreover , since this plant does not exhibit torsion , one can plot d ψ g ( t ) d t and Δ ( ψ g ( t ) , t ) E ˙ ( t ) as a function of ψg ( t ) ( shown in Fig 9G and 9H ) . As predicted from the model , we find the minima of d ψ g ( t ) d t and the maxima of Δ ( ψ g ( t ) , t ) E ˙ ( t ) situated at ψg = ±π/2 , representing the farthest sides of the ellipse ( on the right and left ) .
A detailed analysis has now been carried out of the kinematics of differential growth outside of the plane of curvature , and its implications on plant movement . This shows how a classical measurement , here the position of the tip in the horizontal plane , is insufficient to provide a clear picture of the relation between observed movements and the underlying growth mechanisms . Furthermore this study shows that the kinematics of the out of plane curvature can be described as a simple extension of the kinematics relating curvature in the plane and differential growth [14] . By projecting the differential growth on the planes parallel and perpendicular to the plane of curvature , only one supplementary equation is necessary to describe the full kinematics . This equation relates the orientation of the curvature and the growth in the perpendicular plane . The amplitude of the curvature is modified by the difference in growth rate between the two sides of the organ in the plane of curvature . The orientation of curvature in space is then modified by the differential growth in the perpendicular plane . Only three parameters are necessary to account for the full movement: i . the elongation rate along the median line E ˙ ( s , t ) sets the time scale of the movement , ii . the principal direction of differential growth ψg ( s , t ) , the direction in which the differential growth is maximal and iii . the distribution of the differential growth in this direction Δ ( ψg ( s , t ) ) ( s , t ) . The description of this geometrical framework has been neglected to date , and is a central step to unravel the relation between differential growth and nutation . In both the case of gravitropism [14] and nutation , the destabilizing effects of growth on the movement are regulated by proprioception . The autonomous capacity of plants to control and regulate their own shape is reinforced as a central element of postural control . During gravitropic movements , it has been shown experimentally that effects due to growth could be neglected due to the strong influence of proprioception [14] . It is then expected that this regulation is sufficient to avoid the effects due to growth during nutation . The position of the apical tip in the the horizontal plane , perpendicular to gravity , has been central to the study of kinematics . The relevance of this measure has never been clearly discussed and the underlying hypotheses have remained hidden . In particular the relation between the movement of the apical tip and the dynamics of the differential growth , the motor of the movement , is difficult to extract because the full shape of the plants remains unknown . A simple set of hypotheses needs to be properly stated to constrain the relation between shapes and movements . The whole organ is now considered as a block that undergoes the same variation all along the organ . Despite the simplification of the problem , this has proved useful to unravel the underlying dynamics of the differential growth , retaining the general observed behavior . As plants tend to align their curvature orientation ψc with the principal direction of growth ψg , the pattern observed in the horizontal plane can remain a marker of growth . Common observed patterns , like the circle or the ellipse , are then directly related to different oscillating patterns of differential growth . Furthermore , simple input such as an oscillation of the principal direction of growth , can produce robust , stable stereotypical patterns independently of the initial conditions . Minimal regulation of the movement is necessary to achieve commonly observed patterns like the circle or the ellipse . If measurements in the horizontal plane are useful to understand the kinematics of nutation , they are limited in their scope of analysis . Future studies allowing the measurements of proper 3D kinematics should provide a better understanding of the dynamics of differential growth , and give the exact validity of the measurements performed in the horizontal plane . In conclusion , we have presented a mathematical description of the kinematics of plant nutation based on the interplay between geometry and differential growth . This framework allows a full 3D analysis of complex observed kinematics , shedding light on the underling mechanism , while revisiting the interpretation of common horizontal measurements of the plant tip .
|
In his writings , Darwin considered nutation , the revolving movement of the apical tip of plants , as the most widespread plant movement . In spite of its ubiquity , plant nutation has not received as much attention as other plant movements , and its underlying mechanism remains unclear . A better understanding of this presumably growth-driven process is bound to shed light on basic growth processes in plants . In the work presented here we redefine the problem by describing the kinematics in three dimensions , as opposed to the typical description restricted to the horizontal plane . Within this framework we reveal a simple picture of the underlying dynamics , where the orientation of curvature follows the orientation of maximal differential growth . This parsimonious model recovers the major classes of nutation patterns , as shown both analytically and numerically . We then discuss the limitations of classical measurements where only the movement of the apical tip is tracked , suggesting more adequate measurements .
|
[
"Abstract",
"Introduction",
"Models",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"ellipses",
"classical",
"mechanics",
"plant",
"growth",
"and",
"development",
"phototropism",
"social",
"sciences",
"plant",
"physiology",
"geometry",
"neuroscience",
"circadian",
"oscillators",
"developmental",
"biology",
"plant",
"science",
"mathematics",
"organism",
"development",
"chronobiology",
"kinematics",
"musculoskeletal",
"system",
"tropism",
"plant",
"tropisms",
"physics",
"biochemistry",
"psychology",
"proprioception",
"anatomy",
"curvature",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"sensory",
"perception",
"gravitropism"
] |
2016
|
The Kinematics of Plant Nutation Reveals a Simple Relation between Curvature and the Orientation of Differential Growth
|
Substrate permissiveness has long been regarded as the raw materials for the evolution of new enzymatic functions . In land plants , hydroxycinnamoyltransferase ( HCT ) is an essential enzyme of the phenylpropanoid metabolism . Although essential enzymes are normally associated with high substrate specificity , HCT can utilize a variety of non-native substrates . To examine the structural and dynamic basis of substrate permissiveness in this enzyme , we report the crystal structure of HCT from Selaginella moellendorffii and molecular dynamics ( MD ) simulations performed on five orthologous HCTs from several major lineages of land plants . Through altogether 17-μs MD simulations , we demonstrate the prevalent swing motion of an arginine handle on a submicrosecond timescale across all five HCTs , which plays a key role in native substrate recognition by these intrinsically promiscuous enzymes . Our simulations further reveal how a non-native substrate of HCT engages a binding site different from that of the native substrate and diffuses to reach the catalytic center and its co-substrate . By numerically solving the Smoluchowski equation , we show that the presence of such an alternative binding site , even when it is distant from the catalytic center , always increases the reaction rate of a given substrate . However , this increase is only significant for enzyme-substrate reactions heavily influenced by diffusion . In these cases , binding non-native substrates ‘off-center’ provides an effective rationale to develop substrate permissiveness while maintaining the native functions of promiscuous enzymes .
How enzymes evolve to acquire novel functions has attracted numerous studies on the subject of enzyme promiscuity , which can be subcategorized as substrate permissiveness , mechanistic elasticity , and concomitant product diversity [1–4] . The ability to recruit a non-native substrate may lead to the development of new enzymatic functions , or neofunctionalization , through rounds of mutation and selection . This process is generally thought to occur due to the presence and drift of intrinsic substrate/product permissiveness in the progenitor enzyme without significantly affecting its native function [5] . In plants , enzymes involved in specialized metabolism likely evolve from their primary counterparts through exploiting ancestral promiscuity [6 , 7] . This makes plant specialized metabolism an ideal system to study the role of promiscuity in enzyme evolution . One interesting example is the hydroxycinnamoyl-CoA:shikimate hydroxycinnamoyltransferase ( HCT ) , which belongs to the BAHD acyltransferase family involved in the biosynthesis of a diversity of ester- and amide-containing natural products [8] . HCT produces p-coumaroylshikimate ( S1 Fig ) by transferring the p-coumaroyl group from the acyl donor p-coumaroyl-CoA to the acyl acceptor shikimate . It is an essential enzyme in the phenylpropanoid metabolism , conserved across all land plants [9] . Interestingly , unlike many other essential metabolic enzymes , HCT exhibits relatively low affinity toward its native substrate shikimate and can utilize a variety of non-native substrates . Previously , we crystallized the A . thaliana HCT ( AtHCT ) in the apo , p-coumaroyl-CoA-bound , and p-coumaroylshikimate-bound forms , as well as the C . blumei HCT ( CbHCT ) in complex with p-coumaroyl-CoA and the non-native acyl acceptor substrate 3-hydroxyacetophenone ( 3-HAP , see S1 Fig ) [10] . Comparative analysis of these structures and multiple copies of 100-ns molecular dynamics ( MD ) simulations reveals that a conserved arginine acts as a ‘catalytic handle’—the residue adopts a primarily external conformation in the apo state of the enzyme , and swings to an internal conformation in the presence of the native substrate shikimate . In contrast , binding of the neutral , non-native substrate 3-HAP is unable to elicit such a response . This difference in active-site dynamics helps the promiscuous HCT to maintain the competitiveness of its native reaction over alternative non-native reactions . However , our previous MD simulations did not capture the transition between the external and internal states of the arginine handle , leaving the timescale of the transition and its mechanistic relevance to the enzymatic function an open question . It is also unknown whether such active site dynamics are a universal feature among HCTs . Furthermore , it remains a mystery how the non-native substrate 3-HAP , which binds at a site over 8 Å away from the catalytic center , engages its co-substrate p-coumaroyl-CoA as well as the catalytic residues of the HCT active site . Here , in order to address the above questions , we report microsecond-long simulations performed on the specialized machine Anton [11] of HCTs from five land plants , namely , AtHCT , CbHCT , Coffea canephora HCT ( CcHCT ) , Sorghum bicolor HCT ( SbHCT ) , as well as the newly crystallized HCT from Selaginella moellendorffii ( SmHCT ) . Through collectively 17-μs simulations , we demonstrate the prevalent , sub-microsecond swing motion of the arginine handle across all five HCTs , and reveal how the non-native substrate 3-HAP engages its co-substrate and catalytic residues of the enzyme . Finally , by solving the Smoluchowski equation first for HCT and then for a generic enzyme model , we quantify the impact of off-center binding on 3-HAP reaction rate and examine in general how such binding facilitates non-native reactions in promiscuous enzymes .
We crystallized the HCT ortholog from the lycophyte Selaginella moellendorffii and solved its apo structure at 2 . 9-Å resolution . The SmHCT crystal had a space group of P212121 , and contained two molecules in the asymmetric unit . Several surface loops of the protein ( residues 47-54 , 216-237 , and 259-262 ) have low electron density support and are omitted from the model . Similar to other HCT orthologs , the structure of SmHCT consists of two pseudo-symmetric domains ( Fig 1a ) : one comprised of residues 1-179 and 392-411 , the other of 241-391 and 412-451 , and a long intervening loop consisting of residues 180-240 . The SmHCT active site is located at the interface between its two domains with a cavity volume of 1372 . 2±10 . 4 Å3 . Inside the active site , the universally conserved catalytic His157 acts as a general base to deprotonate the 5-hydroxyl of shikimate , priming it for the subsequent nucleophilic attack on p-coumaroyl-CoA . Apart from His157 , two other key residues involved in substrate binding and catalysis , Thr385 and Trp387 , are also structurally conserved in SmHCT compared to other available HCT structures ( Fig 1b ) . However , there is a clear structural difference in the arginine handle , Arg372 , which resides on one of the three short loops ( L1 ) surrounding the entrance of the active site and coordinates the shikimate carboxylate with two salt bridges throughout the catalytic cycle . A lack of electron density for its side chain suggests that this residue is extremely flexible in SmHCT , occupying many different conformations in the apo state . This is in clear contrast to other crystallized HCT orthologs , which have defined density for the arginine handle . Since there is density to support the protein backbone in this region , Arg372 was modeled based on the location of its Cα atom and highest-probability side chain rotamer . We note that despite the ambiguity in its side chain conformation , the backbone of Arg372 already indicates that the residue protrudes away from the active site much more severely in SmHCT than in other known HCT orthologs ( Fig 1b ) . In order to systematically examine their active-site dynamics , we performed microsecond-long simulations for five orthologous HCTs . Each HCT is simulated in both apo and holo states , with the latter containing the protein in complex with shikimate and p-coumaroyl-CoA ( S2 Fig ) . For HCTs without holo crystal structures , their corresponding apo structures were used , with p-coumaroyl-CoA and shikimate manually introduced into the active site . To fully relax residues around the newly added substrates , a series of simulated annealing ( SA ) simulations were performed before the holo state Anton simulations were launched ( see SI for details ) . These SA simulations can be considered as enforcing a rapid transition from the apo to holo state , while the subsequent Anton runs are used to collect statistics for the holo state . Consistent with our previous work [10] , the Anton trajectories reveal the distinct conformations sampled by the arginine handle in the apo vs . holo HCTs: Fig 2a depicts the largest cluster from the clustering analysis of the arginine handle in each simulation , while Fig 2b shows the residue’s occupancy map generated on a 3D grid and averaged over each Anton trajectory . Together , they indicate that compared with its apo state , the arginine handle is not only more internally oriented but also considerably less flexible across all five holo HCTs . Adopting primarily an internal conformation , the holo state arginine handle anchors the native substrate shikimate at the catalytic center , explaining its essential role in the acyl transfer reaction [12 , 13] . We should add that although overall the arginine handle formed stable salt bridges with shikimate , our microsecond-long trajectories recorded the occasional loss of these interactions . Given this observation , we performed an extra copy of holo simulation for each HCT in order to improve sampling statistics . Among the ten holo HCT simulations listed in S2 Table , the loss of Arginine handle-shikimate interaction was observed in two trajectories: one in AtHCT and the other in CbHCT , with the salt bridges broken at t = 765 ns and t = 925 ns , respectively . In both cases , shikimate left the active site shortly afterwards , and the then free arginine handle began to swing externally , resembling its dynamics in the apo state . As described earlier , the crystal structure of apo SmHCT contains an arginine handle more flexible than that of other HCTs . Given that this residue is indispensable for catalysis [12 , 13] , it must be capable of a sufficiently rapid conformational switch upon shikimate binding , i . e . , retracting its protruded backbone and swinging its side chain , regardless of its initial position , into the internal conformation shown in Fig 2a . While this conformational switch can be enforced by SA simulations mentioned above , these simulations cannot provide any information on its timescale . Such a switch was also absent in the 1-μs apo SmHCT simulation , during which the arginine handle was found to be extremely flexible , resulting in a large 1% occupancy isosurface enclosing the space visited by its side chain and a small 50% occupancy isosurface containing primarily its backbone ( Fig 2b ) . In light of the above results , we launched a 1-μs simulation of SmHCT with a free shikimate initially placed in bulk water . During this simulation , the arginine handle was found to spontaneously retrieve into the active site , a process partly mediated by a salt bridge with the highly conserved Glu206 ( Fig 3a ) . Although shikimate did not successfully enter the active site and the arginine handle eventually swang back out , this simulation revealed the potential pathway and timescale of the residue in its switch from the apo to the holo state . Indeed , in a separate , 0 . 7-μs SmHCT simulation with p-coumaroyl-CoA and shikimate manually placed into the active site ( without simulated annealing ) , the initially outward facing arginine handle underwent a similar swing-in motion at t ≈ 150 ns ( S3 Fig ) . Remarkably , a transient swing-out motion , essentially opposite to the swing-in motion described above , was recorded in apo AtHCT ( Fig 3b ) . Although the arginine handle swung back in shortly after , its maximum root mean square deviation reached 13 . 2 Å , demonstrating the large scale of the transient conformational change . The above results indicate that the arginine handle in HCT is capable of considerably more dramatic swing motions than those revealed by our previous crystal structures and short MD trajectories [10] . Indeed , the total volume of space visited by the residue ranged from 1367 Å3 ( CcHCT ) to 3415 Å3 ( SmHCT ) in the apo simulations ( S3 Table ) . Furthermore , the swing motion of the arginine handle readily occurred on the submicrosecond timescale—as shown in Fig 3c , we characterized the internal and external states of the residue with a swing angle θ . In general , the transition between the two conformational states was found to take at least hundreds of nanoseconds , explaining its absence in our previous , 100-ns trajectories [10] . Although our limited number of samples precludes an accurate measure of the transition rate , averaging across all HCTs yields an estimate of ∼1 . 7 μs−1 . Given that the turnover rate of HCT is about 5 orders of magnitude slower , the transition of the arginine handle between its internal and external states is clearly fast enough to be mechanistically relevant for the acyl transfer reaction catalyzed by the enzyme . HCTs are known for their considerable substrate permissiveness [10 , 14] . In terms of acyl acceptor , they can process ligands with significant structural differences from the native substrate shikimate . In our previous work , we crystallized CbHCT in complex with its non-native acyl acceptor substrate , 3-HAP . As shown in Fig 4a , the ligand is coordinated by Arg350 , Thr298 and Tyr274 , as well as a number of non-polar residues in this non-productive pose . However , with its hydroxyl located over 8 Å away from p-coumaroyl-CoA and the catalytic His153 , it is unclear how the acyl transfer reaction of this non-native substrate would proceed . To explore this process , we performed a 1-μs simulation of CbHCT in complex with p-coumaroyl-CoA and 3-HAP . The crystal structure and the largest two clusters of 3-HAP from the simulation trajectory are shown in Fig 4a–4c . Together , they indicate that the ligand manifests considerable flexibility at its binding site and forms a limited number of specific interactions with the protein: apart from hydrophobic interactions with Phe355 , 3-HAP may form hydrogen bonds with the side chains of Arg350 and Thr298 , or , the backbone of Tyr274 after a nearly 180° rotation . This level of flexibility and the lack of more specific interactions hints that binding of 3-HAP is not particularly strong . Indeed , this non-native substrate was found to wander around in the enzyme lumen during the 1-μs simulation and considerably alter both its orientation and center-of-mass ( COM ) position , a behavior in clear contrast to that of shikimate ( S4 Fig ) . At t = 543 ns , 3-HAP wandered into a nearly productive pose , with its hydroxyl positioned within 3 . 7 Å and 3 . 6 Å of p-coumaroyl-CoA and His153 , respectively ( Fig 4d ) . The molecule was also stabilized by a hydrogen bond with the arginine handle , which lasted approximately 190 ns . This fleeting , nearly productive pose explains the reactivity of 3-HAP , i . e . , despite initially bound in a non-productive pose , the non-native substrate can escape from its relatively weak binding site to explore the interior of the enzyme until it reaches its acyl transfer partner p-coumaroyl-CoA and catalytic residues from HCT . Notably , the binding site of 3-HAP coincides with one of the sites ( site 1 ) predicted by FTMAP ( Fig 4e ) . FTMAP [15 , 16] distributes 16 types of small organic probe molecules on the surface of a macromolecule to map out its binding ‘hot spots’ . Regions that bind multiple probe clusters are identified as consensus sites , i . e . , the hot spots [15] . We overlapped the consensus sites of all apo HCT crystal structures and identified three major sites across the five HCTs ( with the exception of CcHCT , which did not have site 1 ) . A common site 2 was found to overlap with the binding site of the pantothenate moiety of p-coumaroyl-CoA . Interestingly , a third site ( site 3 ) was found near the arginine handle and was briefly visited by 3-HAP during the 1-μs Anton simulation . In addition to the strictly conserved Arg350 , residues surrounding site 3 include Val30 , Pro32 , and Asn294 , all of which are highly conserved across HCT orthologs [10] . It is worth mentioning that the formation of this site depends on the conformation of the arginine handle: in the crystal structures and Anton trajectories of holo HCTs , site 3 disappears due to the more internally oriented arginine handle . Given the binding dynamics described above , an intriguing question arises: how big an effect on the reaction rate , if any , can we expect from the presence of an off-center binding site ? Qualitatively , the presence of a binding site , even an off-center one , should increase the probability of 3-HAP visiting the enzyme active site , thereby , facilitating its reaction; however , a long residence time at this site risks trapping the molecule at a location far away from the catalytic residues , thereby , interfering with its reaction . Assuming that substrates do not interact with each other , a previous study on channel transport suggests that the former effect should be dominating [17] . Multiple studies on protein-ligand binding [18–21] , many of which focused on the electrostatic interactions between charged ligands and an enzyme , also point at a facilitating role of an attractive potential . Nonetheless , as these studies are not tailored for promiscuous enzymes , they do not provide a quantitative answer to the question raised above . Here , we quantify the influence of an off-center binding site on the reaction rate of a neutral , non-native substrate in both HCT and a generic , cylinder-shaped enzyme model . Using the equilibrated structure of CbHCT in complex with p-coumaroyl-CoA , we first created the molecular mesh of the enzyme ( Fig 5a ) and then solved the steady-state Smoluchowski equation ( SSSE ) characterizing the diffusion of the substrate . The reaction rate ( k ) was determined by integrating the flux of the substrate over the surface of the catalytic center , which had a radiation boundary condition [22–28] . With an intrinsic reactivity parameter α representing how ‘good’ the enzyme was , all other chemical details were hidden from our calculation . The relative change in the reaction rate ( Δk ) was obtained by solving the SSSE first without and then with the 3-HAP binding site , which was modeled as a 1 . 7-Å-radius sphere ( Fig 4 ) based on analysis of the Anton trajectory ( see SI for details ) . Its affinity for 3-HAP was varied from -5 to -1 kcal/mol , i . e . , from relatively strong to extremely weak , to examine how binding strength at this site affected Δk . Our calculations indicate that the presence of the off-center binding site always increases the reaction rate of 3-HAP . In essence , this enhancement is analogous to the channel transport case reported by Bauer and Nadler [17] , i . e . , an increased probability of the substrate visiting the catalytic center , brought by the presence of an off-center site , outweighs the prolonged first passage time caused by the ‘trapping’ effect of this site . The magnitude of such rate enhancement , as further demonstrated below in the cylindrical enzyme model , depends on the size and location of the off-center site , the geometry of the enzyme lumen as well as the reactivity of the enzyme . For 3-HAP , Δk could only reach 0 . 5% even if we assumed that its reaction was diffusion-limited ( α = ∞ ) . This result can be attributed to a rather large separation between the 3-HAP binding site and the catalytic center as well as the tunnel-shaped HCT lumen . In reality , as the acyl transfer reaction between 3-HAP and p-coumaroyl-CoA is far from the diffusion limit [10] , Δk brought by the off-center binding site should be negligible . Despite the small Δk observed above , it is of interest to systematically examine various physico-chemical properties controlling the impact of an off-center binding site . Thus , we went on to study a generic , cylindrical enzyme model , the volume of which was chosen to match a typical enzyme cavity volume of 1000 Å3 [29] . The height ( Hin ) of the cylinder and the radius of its base ( rin ) were varied , while its volume was kept approximately constant . For a given pair of Hin and rin , we scanned the affinity ( ΔGb ) and size ( rb ) of the off-center site as well as its distance to the catalytic center ( d ) . Setting ΔGb to zero yields the base reaction rate of the enzyme ( k0 ) , which is on the order of 108 M−1s−1 when α is infinite , i . e . , the reaction is diffusion-limited [30] . As shown in Fig 5c , Δk depends strongly on the reactivity of the enzyme . Only when the reaction is limited or heavily influenced by diffusion ( k0 > 107 M−1s−1 ) , can Δk become significant . In these cases , the effect of the off-center site closely depends on its distance to the catalytic center—the smaller d is , the larger Δk becomes ( Fig 5e ) . Additionally , the geometry of the enzyme lumen matters: as the cylindrical model elongates , Δk decreases ( S5a Fig ) . In the limiting case , our cylinder should approach the 1D model of Berezhkovskii et al . [31] . The decrease in Δk observed here is analogous to a decreasing contribution from the internal domain of the 1D model , i . e . , as the enzyme lumen narrows , finding its entrance from the outside becomes too slow and the presence of a binding site within the active site no longer significantly accelerates the rate at which the substrate reaches the catalytic center . For this reason , off-center binding is less likely to be important in a tunnel-shaped enzyme compared with a basin-shaped one with a wide opening to its active site . Most interestingly , with other metrics held constant , the size of the off-center binding site , rather than its strength , dominated Δk . For instance , in the 3D plot shown in Fig 5e , strengthening the site’s affinity from -2 to -4 kcal/mol with α = ∞ , d = 3 Å and rb = 1 . 8 Å only brought Δk from 9% to 10% . In contrast , doubling the volume of the binding site yielded Δk = 18% even with ΔGb staying at -2 kcal/mol . Furthermore , splitting the binding site into multiple sites while keeping the same total volume does not affect Δk significantly ( Fig 6 ) . In general , we found that with other metrics held constant , Δk scaled linearly with the volume of an off-center site , whereas its relation with ΔGb was approximately an exponential one ( S5 Fig ) . Thus , rather than enhancing the affinity of an off-center binding site , developing larger and/or additional sites , even ones with only weak affinity , appears to be a more effective strategy for a promiscuous enzyme to speed up the reaction of its non-native substrates .
Combining X-ray crystallography and microsecond-long MD simulations , we investigated plant HCTs’ substrate permissiveness by examining their active-site dynamics in the apo state and upon the binding of their native as well as non-native acyl acceptor substrates . A prevalent swing motion of the arginine handle was observed across all HCTs , the timescale of which was found to be on the order of sub-microsecond , i . e . , sufficiently fast to facilitate the recognition of HCT substrates in vivo . Apart from stabilizing the native substrate shikimate in its reactive pose , the arginine handle formed a transient hydrogen bond with the non-native , neutral 3-HAP as the latter adopted a nearly productive pose . The conformation of this residue also dictated the formation of an additional binding hot spot ( site 3 ) briefly visited by 3-HAP , suggesting that it may enable HCTs to provide multiple binding sites for their non-native substrates . Utilizing a flexible arginine to recognize different substrates is a known strategy to sow the seeds for promiscuity . One prominent example is the amine transaminase , where an arginine can flip away to create a cavity to accommodate even substrates lacking a carboxylate group [32] . Given the remarkable flexibility of the arginine handle in HCTs , it appears that a similar strategy has been adopted by these enzymes—with the residue sweeping across the active site on a sub-microsecond timescale , HCTs can readily accommodate its native and various non-native substrates of different sizes , with or without a carboxylate group . Among the five HCTs studied here , clear differences exist in the extent of the arginine handle’s flexibility . For instance , SmHCT presents an externally protruded backbone of Arg372 in the apo state , a distinguishing feature compared to the remaining HCTs . Considering the total volume of space visited by the residue during Anton simulations ( S3 Table ) , we found that CcHCT and SmHCT appeared to be at the two ends of the flexibility spectrum , with the arginine handle being only moderately flexible in the former and exhibiting large-scale swing motion in the latter . Such differences are not entirely unexpected: While HCTs share a common feature of substrate permissiveness , each has its unique promiscuity profile . For instance , the activity of AtHCT towards the non-native acyl acceptor naringenin is impaired relative to that of CbHCT and SmHCT ( S6 Fig ) . While the origin of their differential promiscuity profiles remains to be fully understood , given the central role of the arginine handle in substrate recognition , its different conformational flexibility among various HCTs likely plays a part . The non-native substrate 3-HAP binds HCT at a site approximately 8 Å away from the catalytic center . This off-center site was identified as a hot spot ( site 1 ) by FTMAP , suggesting that it could also serve as a binding site for other non-native acyl acceptor substrates . Our simulations revealed that this site had a relatively weak affinity for 3-HAP , allowing the molecule to leave and explore the interior of the enzyme , eventually achieving a nearly productive pose during a 1-μs simulation . This pose is only stabilized by a hydrogen bond between the carbonyl oxygen of 3-HAP and the arginine handle , which lasted approximately 190 ns , i . e . , considerably weaker than the salt bridge formed between the latter residue and the native substrate shikimate . This difference reflects the competitiveness of the native substrate over the non-native one , affirming the findings of our previous study [10] . The binding dynamics of 3-HAP raises the intriguing question regarding the role of an off-center binding site . Intuitively , one may expect ( correctly ) that binding at an off-center site serve to increase the presence of the substrate within the enzyme lumen , which should lead to its enhanced probability of visiting the catalytic center . However , it is nontrivial to determine whether such an enhancement will always speed up the corresponding reaction , since an off-center site also risks trapping the substrate at a location away from the catalytic center , resulting in a prolonged first passage time [17] . Furthermore , exactly how big an effect ( if any ) can be expected from an off-center site cannot be deduced from qualitative arguments . By solving the steady-state Smoluchowski equation first for HCT and then for a generic enzyme model , we show that the reaction rate of a given substrate is always enhanced by the presence of an off-center binding site , i . e . , Δk > 0; however , its large distance to the catalytic center as well as the relatively narrow entrance of the HCT lumen dictates that Δk brought by the 3-HAP binding site is capped at ∼0 . 5% . The actual Δk is likely much smaller given that the reaction of 3-HAP is far from the diffusion limit . In general , only when a reaction is limited or heavily influenced by diffusion ( k0 > 107 M−1s−1 ) , can Δk become significant . Otherwise , with diffusion faster than catalysis , on average a substrate will always be around when the reaction occurs , with or without the presence of an off-center binding site . For reactions heavily influenced by diffusion , a significant rate enhancement can be achieved by off-center binding . Notably , even a relatively weak binding site ( -2 to -3 kcal/mol ) can already produce a non-negligible Δk . Indeed , our calculations indicate that Δk depends more strongly on the size of the site rather than its strength . Furthermore , splitting the binding site while keeping the same total volume does not significantly affect the resulting Δk . These data suggest that having multiple , weak binding sites can be highly desirable for a non-native substrate . From an evolutionary perspective , it may in fact be less challenging to develop these sites than a single , strong binding site , since the latter tends to require highly specific interactions . Overall , while stabilizing the transition states remains the key to enabling non-native reactions , providing off-center binding sites may constitute a low-barrier mechanism to facilitate substrate permissiveness by certain enzymes . If favorable , such activities can be refined through rounds of mutation , duplication , and selection to yield enzymes with novel functions and unique molecules in the network of specialized metabolism .
The Protein Data Bank accession number of SmHCT crystal structure is 6DD2 . X-ray diffraction intensities were indexed and integrated with iMosflm [33] and scaled with Scala under CCP4 [34 , 35] . The phase was determined with molecular replacement using Phaser under Phenix [36] . Coot was used for manual map inspection and model rebuilding [37] . Anton simulations performed in this work are listed in S2 Table . Prior to the Anton runs , all systems were subjected to a 20 ns equilibration performed with NAMD 2 . 10 [38] . The missing loops in CcHCT , SbHCT and SmHCT were modeled using Modeller [39] . CHARMM36 [40] and CHARMM General force field [41] were employed for all simulations . Parameters of the CoA moiety from ref [42] were used , while parameters of shikimate and the p-coumaroyl moiety were obtained from the CGenFF program [43 , 44] and further validated using Gaussian [45] and the Force Field Toolkit [46] plugin of VMD [47] . Clustering analysis was performed with GROMACS [48] and structural alignment was done with the Multiseq [49] plugin of VMD . SSSE calculations were performed using Mathematica ( version 11 . 2 ) [50] , with the molecular meshes of CbHCT and the cylindrical enzyme model created using PyMOL [51] and Blender [52] . See SI for more details .
|
Examples abound of enzymes that can process substrates other than their native ones . However , the structural and dynamic basis of this promiscuity remains to be fully understood . In this work , we examine HCT , an intrinsically promiscuous acyltransferase with conserved function in all land plants . We uncover the sub-microsecond swing motion of a key arginine residue facilitating the recognition of both native and non-native substrates of HCT . We also quantify the impact of an off-center binding site on the non-native reaction rate . Although our calculations were inspired by HCT , the results apply in general , i . e . , for enzymes heavily influenced by diffusion , binding non-native substrates ‘off-center’ , even with rather weak affinity , can accelerate non-native reactions to appreciable levels .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"crystal",
"structure",
"chemical",
"compounds",
"molecular",
"dynamics",
"enzymes",
"condensed",
"matter",
"physics",
"enzymology",
"organic",
"compounds",
"enzyme",
"metabolism",
"basic",
"amino",
"acids",
"amino",
"acids",
"crystallography",
"enzyme",
"chemistry",
"crystallization",
"solid",
"state",
"physics",
"proteins",
"chemistry",
"salt",
"bridges",
"physics",
"chemical",
"precipitation",
"biochemistry",
"enzyme",
"structure",
"arginine",
"organic",
"chemistry",
"electrochemistry",
"biology",
"and",
"life",
"sciences",
"chemical",
"reactions",
"physical",
"sciences",
"computational",
"chemistry"
] |
2018
|
Structural and dynamic basis of substrate permissiveness in hydroxycinnamoyltransferase (HCT)
|
Bacteria move towards favourable and away from toxic environments by changing their swimming pattern . This response is regulated by the chemotaxis signalling pathway , which has an important feature: it uses feedback to ‘reset’ ( adapt ) the bacterial sensing ability , which allows the bacteria to sense a range of background environmental changes . The role of this feedback has been studied extensively in the simple chemotaxis pathway of Escherichia coli . However it has been recently found that the majority of bacteria have multiple chemotaxis homologues of the E . coli proteins , resulting in more complex pathways . In this paper we investigate the configuration and role of feedback in Rhodobacter sphaeroides , a bacterium containing multiple homologues of the chemotaxis proteins found in E . coli . Multiple proteins could produce different possible feedback configurations , each having different chemotactic performance qualities and levels of robustness to variations and uncertainties in biological parameters and to intracellular noise . We develop four models corresponding to different feedback configurations . Using a series of carefully designed experiments we discriminate between these models and invalidate three of them . When these models are examined in terms of robustness to noise and parametric uncertainties , we find that the non-invalidated model is superior to the others . Moreover , it has a ‘cascade control’ feedback architecture which is used extensively in engineering to improve system performance , including robustness . Given that the majority of bacteria are known to have multiple chemotaxis pathways , in this paper we show that some feedback architectures allow them to have better performance than others . In particular , cascade control may be an important feature in achieving robust functionality in more complex signalling pathways and in improving their performance .
Living organisms respond to changes in their internal and external environment in order to survive . The sensing , signalling and response mechanisms often consist of complicated pathways the dynamical behaviour of which is often difficult to understand without mathematical models [1] . Considering the structure and dynamics of these signalling pathways as integrated dynamical systems can help us understand how the pathway architecture and parameter values result in the performance and robustness in the response dynamics [2] . One extensively studied sensory pathway is bacterial chemotaxis . This pathway controls changes in bacterial motion in response to environmental stimuli , biasing movement towards regions of higher concentration of beneficial or lower concentration of toxic chemicals . The chemotaxis signalling pathway in the bacterium Escherichia coli is a simple network with one feedback loop [3] which has been extensively studied and used as a paradigm for the mechanism of chemotaxis signalling networks [4] . In E . coli , chemical ligands bind to methyl-accepting chemotaxis protein ( MCP ) receptors that span the cell membrane and alter the activity of a cytoplasmic histidine kinase called CheA . When attractant ligands stimulate the chemotaxis pathway by binding to MCP , there is a decrease in the autophosphorylation rate of CheA; conversely , repellent binding or lack of attractant binding increase CheA autophosphorylation activity . CheA , when phosphorylated , can transfer the phosphoryl group to two possible response regulators: CheY and CheB . CheY-P ( where ‘-P’ denotes phosphorylation ) interacts with FliM in the multiple E . coli flagellar motors resulting in a change in the direction of rotation of the motor . At the same time , a negative feedback loop allows the system to sense temporal gradients and react to a wide ligand concentration range: the MCP receptors , which are constantly methylated by the action of a methyltransferase CheR , are de-methylated by CheB-P . This negative feedback loop restores the CheA autophosphorylation rate and the flagellar activity to the pre-stimulus equilibrium state [5] , [6] . Describing this pathway mathematically as a dynamical system can be facilitated by using tools from control theory . For example , it has been shown that the adaptation mechanism in the E . coli model [7] , [8] is a particular example of integral control , a feedback system design principle used in control engineering to ensure the elimination of offset errors between a system's desired and actual signals , irrespective of the levels of other signals [9] . Many species have chemotaxis pathways that are much more complicated than that of E . coli [10] , [11] , either containing chemotaxis proteins not found in E . coli , e . g . in the case of Bacillus subtilis [12]; or containing multiple homologues of the proteins found in the E . coli pathway , as in the case of Rhodobacter sphaeroides [11] , [13] . Furthermore , in R . sphaeroides there are two receptor clusters containing sensory proteins which localize to different parts of the cell , one located at the cell pole and the other in the cytoplasm [14] . Although the purpose of the two clusters is unclear , in vitro phosphotransfer experiments [15] , [16] show that the CheA homologues located at the two clusters can phosphotransfer to different CheY and CheB homologues: at the cell pole CheA2-P phosphotransfers to CheY3 , CheY4 , CheY6 , CheB1 and CheB2 , while at the cytoplasm CheA3A4-P phosphotransfers to CheY6 and CheB2 . The two methylesterase proteins , CheB1 and CheB2 , which are homologues of CheB in E . coli , are responsible for the adaptation mechanism in R . sphaeroides [13] , [17] . Past localization studies have shown that CheB1 and CheB2 are found diffuse throughout the cytoplasm [14] . This is different to E . coli where the CheB protein is localized at the cell pole , and could potentially mean that the two proteins de-methylate either receptor cluster [14] . As a system featuring an adaptation mechanism similar to that in E . coli , but with multiple homologues of the E . coli chemotaxis proteins , it is useful to examine the R . sphaeroides chemotaxis pathway from a control engineering perspective . In this way , we can suggest structures for the R . sphaeroides chemotaxis pathway that integrate the control mechanisms thought to be responsible for adaptation in E . coli along with the possible feedback architectures that arise from the dual sensory modules present in R . sphaeroides . The relative evolutionary advantages of the different architectures can then be compared from both control engineering and biological points of view . The fact that there are two endogenous ‘measurements’ available to the feedback mechanism ( CheB1-P and CheB2-P ) which can be used to regulate two signals ( CheA2 and CheA3A4 ) makes the whole chemotaxis feedback pathway a multi-input , multi-output control system ( as opposed to possessing only one CheB and one CheA as in the E . coli models [7] , [18] ) . This introduces extra degrees of freedom in the feedback control mechanism of the system and , thus , the potential for better regulation . However , the different conceivable connectivity configurations between the two CheB-P proteins and the two receptor clusters actually correspond to different feedback control architectures , each with different properties . Some of these configurations , as will be demonstrated , could allow the bacterium to integrate information from both internal and external sources and to function more efficiently , e . g . , by varying how strongly it reacts to external attractants depending on its internal state . At the same time , the additional receptor cluster not found in E . coli has the potential of introducing extra sources of performance degradation such as noise ( both intrinsic and extrinsic ) and variations in quantities internal to the cell such as protein copy numbers and phosphorylation rates: the feedback signalling pathway may be required to remedy this , and in this regard , some of these feedback architectures perform better than others . One of the different pathway configurations that is possible in this system has similarities to a feedback architecture commonly found in engineering control systems termed cascade control [19] , which is usually employed when the process to be controlled can be split into a slow ‘primary’ sub-process ( in Figure 1 ) and a faster , secondary sub-process ( in Figure 1 ) . Without the internal feedback shown dashed in Figure 1 the primary module maintains a set-point for the secondary module to follow and the output of the secondary module is fed back to the primary . A cascade control design places an additional feedback loop around the fast secondary process ( shown dashed ) . This has been known to improve system performance in several ways: it reduces the sensitivity of the output of the secondary module to changes in the parameters ( thus improving robustness ) , it attenuates the effects of disturbance signals , it makes the step response of the control system to inputs and disturbances less oscillatory and , since the secondary process is relatively fast , the effects of unwanted disturbances are corrected before they affect the system output . Including this additional internal feedback also allows the control system designer more flexibility in increasing the feedback gain to achieve higher bandwidth and faster system responses without losing stability . In fact , cascade control is employed as a design principle in several engineering systems such as aircraft pitch control and industrial heat exchangers ( see Text S1 for further details ) . In our previous work [20] , we used a model invalidation technique to arrive at a possible pathway architecture that allows the R . sphaeroides chemotaxis system to convey , via a signalling cascade , sensed changes in ligand concentration outside the cell to the flagellar motor . In that model , proteins CheY3-P and CheY4-P act together to promote autophosphorylation of CheA3A4 ( schematically illustrated in Figure 2 ( A ) ) whilst CheY6-P binds with the FliM rotor switch to increase the frequency of motor switching ( and hence reduce the motor rotation frequency ) . This stimulation of CheA3A4 need not be a direct interaction [20] . In this paper , we assume that the chemotaxis pathway has the same forward signalling pathway of [20] and then suggest four plausible interconnection structures for the feedback pathway between the two CheB-P proteins and the two receptor clusters . Following this , we present the results of experiments that are used to invalidate all but one of these structures . We then discuss the results of in silico experiments that highlight the differences in chemotactic performance between the different models with particular focus on the robustness of chemotaxis to parametric variations in the chemotaxis pathway and noise [21] , [22] . Using analytical techniques from control theory , we demonstrate that the model not invalidated by our experiments is structurally similar to the cascade control architecture , and we use the structural properties of this interconnection , which are commonly used to reduce the effects of uncertainty and disturbances in various engineering applications , to explain the robustness features of the suggested model .
Given the structure of the forward path of the chemotaxis pathway from [20] , illustrated in Figure 2 ( A ) , and given the rates previously measured in [15] , [16] for the phosphotransfer reactions also shown in Figure 2 ( A ) , we constructed a generic ordinary differential equation model of the R . sphaeroides chemotaxis pathway , detailed in Materials and Methods . With this forward signalling pathway , the model makes the following assumptions: One effect of a sensed increase in ligand concentration is a decrease in the flagellar switching frequency due to decreased amounts of CheY6-P binding with FliM . Figure 3 shows the result of a simulation of the signalling pathway that demonstrates the fall in the concentration of CheY6-P in response to a step decrease in the number of active receptors at the polar or at the cytoplasmic clusters . The reaction rates of the phosphotransfer network are such that a change in the number of active receptors at the cytoplasmic cluster causes a faster fall in CheY6-P concentration than does a similar change in the number of active receptors at the polar cluster . Qualitatively , the adaptation mechanism in the generic ODE model presented in Materials and Methods functions as follows: CheB1-P and CheB2-P are assumed to de-methylate active receptors , and the phosphotransfer network responds to a sensed increase in ligand concentration by reducing the concentration of CheB1-P , CheB2-P , CheY3-P , CheY4-P and CheY6-P . This results in a reduction in the de-methylation rate of active receptors in the two receptor clusters , and also results in a decrease in the flagellar stopping frequency ( which corresponds to an increase in the flagellar rotation rate ) . The constant methylation of inactive receptors by CheR2 and CheR3 then causes the number of methylated receptors , and , it is assumed , of active receptors , to increase . Thus , the number of active receptors is eventually restored to its pre-stimulus equilibrium level . In turn , the phosphotransfer network then restores the amount of CheY6-P , and hence the flagellar switching frequency , back to its original level . According to the model of the forward signalling pathway , the proteins CheB1-P and CheB2-P therefore act as feedback signals that restore the chemotaxis pathway to its original state . However , the exact connectivity between CheB1-P/CheB2-P and the two receptor clusters is unknown . To determine the most likely interconnection structure and to provide a rationale of how such a structure may be advantageous in terms of chemotactic performance , we created four variants of the generic ODE model with the forward pathway , each having a different interconnection structure between the proteins CheB1-P/CheB2-P and the two receptor clusters ( Figure 2 ( B ) ) . All models were able to produce wild type response data and behaved as expected for the response data generated with gene deletions available at the time . The unknown parameters in the models ( ) were fitted to wild type data for each model . The significance of these parameters is as follows: For notational convenience , it is useful to group the CheB1-P/CheB2-P feedback gains into a feedback matrix . The four CheB1-P and CheB2-P feedback connectivities ( and their associated ) for which models were constructed are as follows: After constructing these four models , we carried out experiments to differentiate between them , by finding the optimal initial conditions of the cells in the assay so as to maximize the difference between the outputs of the different models [20] , [24] . The conditions searched were limited to what could be implemented experimentally and included deletions , over-expression of proteins and combinations of these . To confirm these conditions allow for invalidation , simulations were run of the four models I–IV testing the possible initial conditions and inputs . The simulations showed that the initial conditions that allow for the best model invalidation were the deletion of CheR3 and , in a separate experiment , the deletion of CheB1 ( Figure 4 ) . The experiments were then implemented in R . sphaeroides , subjecting a population of cells to a step increase in ligand concentration ( propionate ) and then measuring the resulting flagellar activity through a tethered cell assay ( Figure 4 ) . Experimentally the deletion of either CheB1 or CheR3 resulted in cells with a rotation frequency of −8 Hz that showed no noticeable response to the addition or removal of ligand . In the simulations , only Models I and III displayed this behaviour upon deletion of CheR3 ( Figure 4 , top row ) and only Model III displayed this behaviour upon deletion of CheB1 ( Figure 4 , bottom row ) . Models I , II and IV were thus invalidated and only Model III was able to replicate the experimental data . As a test of this model invalidation , a further experiment wherein CheB2 was deleted was performed . The result of this experiment and the outputs of the four models under the CheB2 deletion ( overlaid ) are shown in Figure 5 . Models I and III were once again able to replicate the deletion data whilst Models II and IV produced outputs that differed from the experimental outcome . The experiments described above demonstrated that the proposed Models I , II and IV are invalid , being unable to explain experimental data . To compare the four models further , in silico experiments were performed on the data-fitted Models I–IV that compared how the different feedback configurations affect chemotactic performance in terms of the sensitivity of the flagellar stopping frequency in response to variations in the values of the models' biochemical parameters and in response to noise . Following these results , we use linear models with structures that represent the different connectivities of Models I–IV to analyze these structures' relative sensitivities to parametric variations and noise . The performance of the different chemotaxis models was compared by simulating the efficiency of each model in ascending an attractant gradient , as illustrated in Figure 6 ( left ) . For each chemotaxis model , Figure 6 shows the average distance travelled up the attractant gradient by ten bacteria during a simulation lasting 80 seconds . As shown in Figure 6 ( right ) , the chemotactic performances of the different models according to this measure were nearly identical ( see Materials and Methods for more details ) . The bacterium's environment is typically composed of regions of high and low chemoattractant or chemorepellant concentrations . Additionally , the bacterium will sense small , fast fluctuations in the detected level of ligand due to molecular noise . To test how sensitive the chemotaxis Models I–IV are to such ligand fluctuations , an in silico experiment was performed on each model in which the ligand concentration sensed by the polar cluster , L , was modelled as the noisy signal L = max ( 0 , 1+η ) , where η is a white noise signal with a zero-mean , unit variance Gaussian distribution . The resulting rotation frequencies were then recorded and are shown in Figure 7 . As can be seen in Figure 7 , ligand level fluctuations sensed at the polar cluster of receptors resulted in larger variance of the rotation frequency in Models I , II and IV than in Model III . The sensitivity of the chemotaxis Models I–IV to ligand inputs was then tested in two in silico experiments which were performed on each model and in which the flagellar rotation frequency was recorded in response to sinusoidal variations in the ligand signals ( the latter of which corresponds to ligand inputs acting on the cytoplasmic cluster ) . As can be seen in Figure 8 , ligand level fluctuations sensed at the polar cluster of receptors resulted in larger changes in the rotation frequency in Models II and IV than in I and III . When the ligand concentration variations were sensed at the cytoplasmic cluster the result was a greater variation in the rotation frequency in Models I and III than in the other two models . Once more , these simulations suggest that CheB1-P de-methylating the cytoplasmic cluster differentiates the performance of Models II and IV from Models I and III . To investigate the sensitivity of the models to parameter variations , we performed an in silico experiment in which , for each of the different chemotaxis models , the variation of the steady-state of the chemotaxis system was measured under randomly chosen values of the copy numbers of chemotaxis proteins ( see Materials and Methods ) . For each chemotaxis protein , the resulting coefficient of variation of the steady-state is shown in Figure 9 . Once more , there was a similarity in the sensitivity of each model to these parametric variations between Models I and III and between Models II and IV , with the latter pair showing slightly higher sensitivity to copy numbers of the chemotaxis protein CheY6 among others . In addition , Model III showed considerably lower sensitivity with respect to CheB1 copy numbers than the other models . Further insight to the differences in performance between the models can be obtained by analyzing the interconnection structure of these models using control theory . In particular , the way in which such feedback arrangements can affect the performance of control systems like the R . sphaeroides chemotaxis pathway can be studied by comparing the behaviour of different linear systems that are structurally similar to Models I–IV . The block diagram in Figure 10 depicts a system composed of two modules representing the polar and cytoplasmic clusters . The CheB1-P/CheB2-P outputs of the two modules exhibit exact adaptation through integral control in response to step changes in the input ligand concentration level , as in E . coli [8] . Depending on the values of feedback gains and ( which correspond to respectively in the chemotaxis models described above ) , the system can represent one of the four chemotaxis models: The gains in Figure 10 are such that , representing the fact that the cytoplasmic receptor cluster can , as a result of the measured reaction rates , relay a sensed ligand input signal to the flagellar motor faster than the polar receptors cluster ( see Figure 3 ) . For the examples we shall consider we set and . Gains and correspond to and in the chemotaxis model respectively . The frequency domain transfer function of the system in Figure 10 from the ligand inputs and to the output is then ( 1 ) where . This function is a frequency-domain map from signals and to the output Y , which corresponds to the flagellar rotation frequency . In the following , we shall use this frequency domain representation of the chemotaxis system to demonstrate how the feedback of linear systems with structures similar to the chemotaxis Models I–IV affects system performance . The Bode magnitude diagrams ( Materials and Methods ) in Figure 11 ( A ) illustrate the effect of increasing in reducing the sensitivity function of the system ( 1 ) over most excitation frequencies ( see the Discussion and Text S1 for a brief introduction to sensitivity functions ) . At the same time , Figure 11 ( B ) shows that strengthening the feedback , which corresponds to increasing the de-methylation of the cytoplasmic cluster by CheB2-P , decreases the sensitivity of the polar cluster over low frequencies . Figure 12 presents a Bode magnitude plot showing the gain of the linear system ( 1 ) to inputs and which represent sensed ligand at the polar and cytoplasmic receptor clusters respectively . The figures show that , similar to the simulations of Models I and III , the linear model with a gain ( similar in structure to Model I ) and ( similar in structure to Model III ) also shows a relatively low sensitivity to high frequency ( noisy ) inputs at the polar receptor cluster and a relatively high sensitivity to noise detected at the cytoplasmic receptor cluster .
R . sphaeroides has a more complex chemotaxis network than E . coli and the multiple receptor clusters and multiple feedback pathways mean that mutants will not always have an intuitive phenotype . For example the ΔcheB1 mutant does not have the loss of response phenotype one would expect from a direct comparison with the E . coli system . We can try to understand why ΔcheB1 has a steady state at −8 Hz by looking at the structure of the model we have been unable to invalidate , and the reason is as follows: CheB1 , CheB2 and CheY6 ( along with CheY3 and CheY4 ) each compete for phosphoryl groups from CheA2-P . CheB1 is present in relatively large copy numbers and CheB1-P has negligible degradation rate ( see Table 1 ) . When present , CheB1 ‘stores’ a large proportion of phosphoryl groups . When absent , the competition for phosphoryl groups from CheA2-P remains between CheB2 , CheY6 , CheY3 and CheY4 . The rate of phosphorylation of CheY6 by CheA2-P is relatively small , CheY6-P receiving most of its phosphorylation from the CheA3A4-P complex . Therefore deleting cheB1 shifts the equilibrium of the system so that a higher proportion of the phosphoryl groups from CheA2-P go to CheY3 , CheY4 or CheB2 . The increase in CheY3-P and CheY4-P results in a stronger negative feedback to the cytoplasmic cluster , and the steady-state amount of active receptors at the cytoplasmic cluster is therefore less in the case of ΔcheB1 . The consequence of this is that the main source of phosphorylation for CheY6-P , which is CheA3A4-P , is reduced , and hence the level of CheY6-P is reduced . The stopping frequency is consequently reduced . Therefore , rather than ΔcheB1 leading to a loss of response to stimulus , the result of this deletion is a shift in the steady state to a high rotation frequency . The performance measure of Figure 6 suggests that in ascending a ligand gradient under ideal conditions the four models behave almost identically , which may be expected as they all exhibit the same output profile under a step ligand addition . At the same time , simulations of the chemotaxis models showed a difference in robustness between Model III and the other models . From an evolutionary point of view , this may suggest that Model III may have advantages in terms of the robustness of chemotactic performance with respect to the other models . These differences in performance and their implications for chemotaxis are discussed next . It is desirable that the chemotactic performance of the bacterium is unaffected by changes such as noise in gene expression between the expression of CheOp2 and CheOp3 and therefore the ability to filter out any parametric variations from the pathway's output would be an advantageous feature . The pathway's primary output and the main determinant of chemotaxis performance is the flagellar rotation frequency , which , according to the four models presented , is directly controlled by CheY6 . It was shown that Models I and III ( the latter of which was not invalidated ) have a slightly lower sensitivity to variations in the copy number of CheY6 compared to Models II and IV ( Figure 9 ) . If Model III is indeed valid , such robustness could serve to better maintain the nominal steady state rotation frequency . Model III also has advantages with respect to Model I due to the CheB2-P feedback to the polar cluster . Strengthening this feedback to the polar cluster , which corresponds to increasing the de-methylation rate of the polar cluster by CheB2-P , is equivalent to increasing the gain in the linear system ( 1 ) – see Figure 10 . For the linear model ( 1 ) , this reduction in sensitivity is illustrated in the Bode sensitivity plot in Figure 11 ( A ) . From the point of view of control system design , this feedback is typically used to reduce the magnitude of the system's sensitivity function ( see Text S1 ) . This function is dependent on the frequency at which the system is excited and can be shown to be equal to the relative incremental change in the overall system's transfer function in response to an incremental change in the transfer function of the system's sub-modules and . If the sensitivity of the chemotaxis system is low , then the bacterium would be able to maintain its chemotactic response despite changes in the system's biological parameters . The Bode plots ( Materials and Methods ) in Figure 11 ( A ) illustrate the effect of increasing in reducing the sensitivity function of the system ( 1 ) over most excitation frequencies . This effect can observed in the chemotaxis models in Figure 9 and Figure 13 , where it is shown that strengthening the CheB2 feedback to the polar cluster reduces the sensitivity of the steady state rotation frequency to changes in the copy numbers of CheB1 and CheA2 ( see Materials and Methods ) . Simulation results in Figure 7 show that the switching frequency in Model III has a low sensitivity to noisy variations in ligand signals detected at the polar receptor cluster relative to the other models . Figure 8 shows the result of a further set of simulations of the four chemotaxis models in which the gain of each chemotaxis model in response to sinusoidal ligand variation detected at the two clusters is given as a function of ligand fluctuation frequency ( see Materials and Methods ) . The figure shows that the switching frequency in Models I and III has a relatively low gain with respect to varying ligand signals detected at the polar receptor cluster and a relatively high gain with respect to ligand variations detected at the cytoplasmic cluster . The Bode magnitude plots in Figure 12 show the frequency-dependent gain of the linear system ( 1 ) to sinusoidal ligand inputs in the case , which is structurally similar to Models I and III . These plots parallel the results of the frequency response magnitude plots of Figure 8 which , for Models I and III , show low gain in response to high frequency inputs at the polar receptor cluster and high gain in response to high frequency signals at the cytoplasmic receptor cluster . The rejection of high frequency inputs at the cell pole may be advantageous in that the flagellar switching rate is then only varied when the polar cluster senses a relatively significant ligand concentration gradient that is large in spatial extent , and remains relatively unchanged when the receptors are subject to rapid fluctuations in sensed ligand due , for example , to molecular noise at the receptor such as that simulated in Figure 7 . Although the chemotaxis model assumes that the cytoplasmic cluster input depends on the sensed ligand , it is unknown what the cytoplasmic cluster senses . In addition to the possibility that this input is a function of the sensed ligand concentration , this cluster may potentially also integrate information about the metabolic state of the cell . In this case , this signalling may well be important to chemotactic performance and the relatively high gain of Model III to inputs at the cytoplasmic cluster may suggest that this configuration would favour internal signals over external signals in terms of output . However , if chemotaxis is sensitive to such signals , it would be important that: ( i ) these signals are tightly controlled and relatively free of the influence of noise and ( ii ) the cytoplasmic cluster be insensitive to variations in its biological parameters , as sensitivity to such variations would diminish the system's ability to correctly respond to inputs to the cytoplasmic cluster . In Model III , the CheB2-P feedback loop around the cytoplasmic cluster could offer this reduction in the sensitivity function of this cluster to such parametric variations . This reduction in sensitivity to variations of cytoplasmic cluster parameters is illustrated in Figure 11 ( B ) using the linear model ( 1 ) of the chemotaxis system . The figure shows that increasing the feedback gain , which corresponds to the gain of the CheB2-P feedback to the cytoplasmic cluster in Model III , achieves a reduction in the sensitivity of the cytoplasmic cluster . In this way , the cytoplasmic cluster remains sensitive to its inputs , as shown by the large gain at high frequency in Figure 12 ( B ) , whilst its sensitivity to parametric variation is reduced due to the internal CheB2-P feedback . This effect can be observed in the chemotaxis models in Figure 14 , where it is shown that strengthening the CheB2 feedback to the cytoplasmic cluster reduces the sensitivity of the steady state rotation frequency to changes in the copy numbers of CheA3A4 and CheY6 ( see Materials and Methods ) . Figure 8 also shows that for Model I and III , high frequency variations in the ligand concentration sensed at the polar cluster are largely filtered out before causing flagellar switching . This may suggest that the relatively slow dynamics of the polar receptor cluster enable it to function as a low pass filter , preventing any high-frequency noisy variations in the sensed concentration of ligand from being signalled through to the flagellar motor . Figure 12 ( A ) illustrates this attenuation of high frequency polar cluster ligand inputs for the linear model ( 1 ) . When combined with the forward signalling pathway which was not invalidated previously [20] , Model III has a feedback structure that corresponds to a control scheme termed cascade control . This term is used to denote a modular system that includes two feedback loops , one nested within the other . The nested loop is used to regulate a sub-process of the system whilst the ‘external’ negative feedback loop from the system output to the input is used to regulate the entire system . The measured reaction rates of the two clusters [15] , [16] are also such that the cytoplasmic cluster is faster than the polar cluster in responding to inputs , which would be required for the chemotaxis pathway to function as a cascade controlled system [19] . This modularization of the chemotaxis system into fast and slow parts mirrors the division of the cascade controlled system in Figure 1 into the slow and fast subsystems and respectively . The cascade control architecture enables the slow ( primary ) subsystem to fix a set-point for the fast ( secondary ) system and for the feedback around the secondary system to quickly regulate the secondary output in response to disturbances and variations in the secondary process [19] . This difference in speed is represented by having , and in the linear model ( 1 ) . Model III also features both an ‘internal’ feedback loop nested within an ‘external’ one corresponding to the dashed and solid feedbacks in Figure 1 , respectively . These two feedbacks are manifested by the CheB2-P feedback that de-methylates the cytoplasmic and the polar clusters respectively . Interestingly this architecture mirrors the ability of the system to phosphotransfer , with the membrane cluster being able to phosphotransfer to and be de-methylated by both CheB proteins and the cytoplasmic cluster only phosphotransferring to CheB2 , the protein that is able to de-methylate it . It does however raise an interesting question . Whereas CheB in E . coli is localised to the polar signalling cluster , in R . sphaeroides both expressed CheB's are found to be delocalised . Yet , only one of the CheB proteins interacts with both signalling clusters . Thus the advantage of having delocalised CheB1 is unclear . We have shown that if the R . sphaeroides chemotaxis pathway has a cascade control architecture , this would enable robust chemotaxis in an uncertain , noisy environment , conferring a selective advantage . In E . coli , one feedback loop is used to achieve perfect adaptation and sensing of temporal gradients and because there is only one signalling cluster all signal integration occurs there . Unlike E . coli , the R . sphaeroides chemotaxis pathway with cascade control feedback provides the bacterium with two feedback loops , one embedded within the other , to adapt and to reduce its sensitivity to parameter variations and noise . The other advantage to this architecture is demonstrated by the simulations shown in Figure 12 , which illustrate that with this structure the system would be strongly sensitive to fast-changing inputs to the cytoplasmic cluster , perhaps from the metabolic state of the cell . Understanding how biological networks achieve robust functionality in the face of disturbances and noise in their internal and external environment is a key question in systems biology . Such networks can be seen as control engineering feedback systems and can be analyzed using system engineering tools in order to understand the advantages of particular internal connectivities over others . In line with this methodology , this paper first utilized a network discrimination approach [20] to construct a model of the feedback connectivity within the R . sphaeroides chemotaxis pathway , and then explained the robustness properties of that model by re-interpreting the theoretical advantages of its cascade control structure in a biological framework and comparing it to the other possible models . This suggests a mechanism by which the bacterium can achieve robust chemotactic performance despite biochemical parameter variations and noise . Given that many chemotactic systems have multiple homologues [10] it would appear that using more complex feedback architectures to improve performance may be common in chemotaxis and in other signalling pathways , raising the possibility that this methodology can be used to analyze a wide set of biological systems .
In the next three subsections , we present the three different modules of the chemotaxis signalling pathway: sensing , transduction and actuation . We assume the same underlying mechanisms for the polar ( MCP ) and the cytoplasmic ( Tlp ) receptors . The parameters of the Tlp cluster are labelled with a tilde superscript . We also make the same assumptions of our model as those in [20] , which are adopted from the E . coli chemotaxis literature [23] . With the notation defined in Table 2 , the model for the sensing mechanism is as follows: ( 2 ) We assume that the cytoplasmic receptor cluster senses extracellular ligand concentrations indirectly; for example , could be internalized attractants , a by-product of the internalization process or a metabolic response to it . For simplicity , we assume the following affine relationship between L and ( 3 ) We let ε = 1 ( µM ) −1 and . The remaining unknown parameters in this model are the dimensionless quantities , the feedback matrix ( which have units of ( µM s ) −1 ) and ( which have units of s−1 ) . The significance of these parameters was detailed in the Results section . We obtain the following values for these unknown parameters for the different models by fitting them to data from tethered cell assays: The difference between models I–IV lies in the structure of the CheB1-P , CheB2-P feedback . We assume that the structure of the phosphotransfer network is the same as that of the models presented previously in [20] , with the modification that when polar and cytoplasmic receptors are in their active state the respective auto-phosphorylation rates of CheA2 and CheA3 , and , are accelerated to and where and are the reaction constants of the auto-phosphorylation of CheA2 and CheA3 obtained from in vitro experiments in the absence of the influence of receptors , as given in Table 1 and in [20] . Biologically , it would be expected that the auto-phosphorylation rates and ( for the case where CheA2 and CheA3 are each in a fully active complex ) are higher than the rates and measured in vitro . We denote the flagellar stopping frequency by M . We assume some interaction which does not lead to a long lasting binding between CheY6-P and the FliM rotor switch . However , stopping frequency decreases at a constant rate in the absence of CheY6-P . This relationship between the CheY6-P and the stopping frequency effectively constitutes a low-pass filter that attenuates fast changes in CheY6 -P concentration . We model this behaviour by: ( 4 ) The output of the model is the flagellar rotation frequency observed in tethered cell assays . We use the following heuristic description to convert motor activity into R . sphaeroides body rotations ( given in rot/sec or Hertz ) : ( 5 ) We set which means that saturation occurs at −8 rot/sec . This value follows from experimental observations – even for major changes in attractant concentrations this value was almost never surpassed .
The measure of chemotactic performance used in the paper is the relative distance travelled by the bacterial cells up an attractant gradient . The medium in which the cells chemotax is assumed to be a two-dimensional plane having an x- and a y- dimension ( where distance along these two directions is unit-less ) , as illustrated in Figure 6 ( left ) . The ligand concentration L is assumed to vary as L = 100x ( for x>0 ) and L = 0 otherwise , remaining unchanged along the y direction . The simulation is initialized with the bacterial cells having a starting position of x = 0 and an initial orientation aligned with the positive x direction . At each switch , the bacterium is assumed to change its orientation by an angle ( measured in radians ) randomly selected from the zero-mean , unity-variance Gaussian distribution . The concentration of ligand at its position , is then input into the chemotaxis model described above . The output , the flagellar rotation frequency ( in Hz ) , is then translated to the size of the step the bacterium makes in the direction of its orientation .
To measure the effect of the variation of a particular parameter on the steady state flagellar rotation frequency , several values of the parameter of interest were randomly selected from a normal distribution with a mean given by the nominal value of the parameter for the given model and with a standard deviation given by half the nominal value of the parameter . A simulation of the model at steady state was then run and the resulting steady state rotation frequency was recorded for each of the randomly chosen parameter values . The coefficient of variation , given by the ratio of the standard deviation of the recorded steady state values to the nominal steady state value was then computed . This dimensionless quantity can be used to compare the dispersion of quantities with a non-zero mean . Sensitivity to a certain parameter value is therefore high when its corresponding coefficient of variation is high , as this would indicate a significant shift from the nominal output in response to a variation in parameter values .
To compare the different chemotaxis feedback structures in an analytical way , the linear system ( 1 ) was constructed . A rich theory exists to analyze and compare the properties of linear systems in the so-called frequency domain using their associated transfer functions [25] . Using such tools , it is possible to study the effects of excitation frequency on systems' gains and sensitivities as was done in this paper . As an example of how this method works , consider a linear dynamical system ( 6 ) where A , B and C are matrices of appropriate dimension , whose entries depend on the model parameters , and is a sinusoidal input with angular frequency and fixed amplitude r . System ( 6 ) is the so called state space representation of the model in the time domain . It is common in control systems engineering to investigate the behaviour of such a system's dependency on excitation frequency . This requires transforming the system to the frequency domain via the Laplace transform . We denote the Laplace transform of u and y by U ( s ) and Y ( s ) respectively , where s is a complex independent variable . Then , where G ( s ) is the transfer function in the frequency domain and is given by [25]: By evaluating this function for values of s on the imaginary axis ( by setting where j is ) we obtain a frequency domain relationship between the system's input and output . If the system is stable ( the eigenvalues of matrix A have negative real parts ) and is excited with a periodic input signal u of frequency , then after some transient behaviour the output y is given by a sinusoidal wave that is phase shifted and amplified with respect to u by amounts dependent on . The amplification factor is given in decibels bywhilst the phase shift is given by The Bode magnitude plot shows the variation of in decibels with frequency of excitation . The Bode phase plot shows the variation of in radians with frequency of excitation .
Model parameters were obtained by performing least squares fitting on previously obtained experimental data [15] , as described in [20] . These are listed in Table 1 . Protein concentrations were obtained via quantitative western blotting as described in [20] .
We tested the gain of each model to sinusoidally varying ligand input signals , applied separately at the polar and at the cytoplasmic clusters . In the first case we applied the constant ligand input , with to the cytoplasmic cluster whilst simultaneously applying to the polar cluster sinusoidally varying ligand signals given by , with frequency in the range 0 . 01 to 1 rads−1 . In the second case we applied the constant ligand input to the polar cluster whilst simultaneously applying to the cytoplasmic cluster sinusoidally varying ligand signals given by , with frequency in the range 0 . 01 to 1 rads−1 . The frequency response magnitude plots of Figure 8 show the magnitude of the fundamental frequency of the sinusoidal variation in the flagellar rotation frequency in response to these sinusoidal ligand input signals .
The strains used in this study are shown in Table 3 . R . sphaeroides strains were grown in succinate medium at under aerobic conditions with shaking . Where required , nalidixic acid was used at concentrations of 25 g ml−1 .
Tethered cell responses to propionate of the R . sphaeroides strains were characterized as described previously [20] . For each strain and wild type 4 slides were analyzed each containing 10 cells .
|
Bacteria move towards favourable environments by changing their swimming pattern . An important feature of this response , which is called bacterial chemotaxis , is that their sensing ability remains independent of the background environment in which they find themselves . This feature has been studied extensively in the bacterium E . coli , which has a simple chemotaxis decision mechanism . However , it has been recently found that most bacteria could potentially have a much more complicated decision mechanism for this response . In this paper , we look at the chemotaxis behaviour of one such bacterium , R . sphaeroides . We develop mathematical models of possible decision mechanisms and undertake an experimental procedure to investigate their validity . We find that only one of four such models can explain the chemotaxis response in R . sphaeroides . Compared to the other models , this model corresponds to a decision mechanism that provides the bacterium with improved swimming performance over the others . Moreover , this decision mechanism has been used extensively to improve performance in several engineering systems . We suggest that this mechanism may play an important role in improving chemotactic performance in other bacteria and in other signalling pathways .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Measuring",
"chemotactic",
"performance",
"Parametric",
"sensitivity",
"analysis",
"Linear",
"systems",
"analysis",
"techniques",
"Model",
"parameters",
"Response",
"to",
"noisy",
"ligand",
"input",
"Plasmids",
"and",
"strains",
"Tethered",
"cell",
"analysis"
] |
[
"microbiology/microbial",
"physiology",
"and",
"metabolism",
"biochemistry/cell",
"signaling",
"and",
"trafficking",
"structures",
"computational",
"biology/systems",
"biology",
"computational",
"biology/signaling",
"networks"
] |
2011
|
Feedback Control Architecture and the Bacterial Chemotaxis Network
|
It is widely assumed that active RNA polymerases track along their templates to produce a transcript . We test this using chromosome conformation capture and human genes switched on rapidly and synchronously by tumour necrosis factor alpha ( TNFα ) ; one is 221 kbp SAMD4A , which a polymerase takes more than 1 h to transcribe . Ten minutes after stimulation , the SAMD4A promoter comes together with other TNFα-responsive promoters . Subsequently , these contacts are lost as new downstream ones appear; contacts are invariably between sequences being transcribed . Super-resolution microscopy confirms that nascent transcripts ( detected by RNA fluorescence in situ hybridization ) co-localize at relevant times . Results are consistent with an alternative view of transcription: polymerases fixed in factories reel in their respective templates , so different parts of the templates transiently lie together .
It is widely assumed that an RNA polymerase transcribes by diffusing to a promoter , binding , and then tracking down the template as it makes its transcript [1] . Accumulating evidence , however , is consistent with an alternative: a promoter diffuses to a transcription factory where it binds to a transiently immobilized polymerase , which then reels in its template as it extrudes a transcript [2]–[6] . Here , we address the question: Are transcribing enzymes mobile or immobile ? Our strategy involves switching on transcription of two genes rapidly and synchronously using tumour necrosis factor alpha ( TNFα ) . This cytokine orchestrates the inflammatory response in human umbilical vein endothelial cells ( HUVECs ) by signalling through nuclear factor kappa B ( NF-κB ) to activate a sub-set of genes [7]–[8] . SAMD4A—a 221 kbp-long gene that encodes a regulator of this pathway—is amongst the first few to respond . Microarray analysis reveals that a synchronous wave of transcription initiates within 15 min , before sweeping down the gene ( at ∼3 kbp/min ) to reach the terminus ∼70 min later ( Figure S1 ) ; no transcripts from the non-coding strand are detected [9] . RNA FISH using intronic probes confirms that almost half the cells in the population respond; essentially no nascent RNA can be detected prior to stimulation , no transcription occurs from the antisense strand , and probes targeting successive introns only yield signal as the wave passes by [9] . TNFAIP2—a short 11 kbp gene that lies ∼50 Mbp away from SAMD4A on chromosome 14—encodes another regulator . It is switched on as rapidly and then repeatedly transcribed over the next 90 min . We use it as an external reference point ( or “anchor” ) and analyze the contacts it makes with different parts of SAMD4A using chromosome conformation capture ( 3C ) —a powerful tool for detecting proximity of two DNA sequences in 3D nuclear space [10]–[12] . If the conventional model for transcription applies , we would not expect the anchor to lie close to any part of SAMD4A either before or after adding TNFα , as it lies so far away on the chromosome ( Figure 1 , left ) . Even if polymerases on the two genes happened to lie together ( for whatever reason ) , tracking of one down the long gene should increase the distance between transcribed parts of the two genes . But if both genes were transcribed by polymerases transiently immobilized in one factory , the short gene—which would repeatedly attach to ( and detach from ) the factory as it initiates ( and terminates ) —should always lie close to just the part of SAMD4A being transcribed at that particular moment ( Figure 1 , right ) . Thus , as the polymerase reels in SAMD4A , introns 1 , 2 , 3 , etc . should successively be brought into the factory to lie transiently next to the anchor . Results using TNFAIP2 ( and other anchors ) are impossible to reconcile with the widely held assumption that polymerases track; rather they are consistent with active polymerases being immobilized in factories .
As our strategy requires one gene to be used as an anchor , we applied 3C and a variant of “associated chromosome trap” ( ACT ) [13]–[14] to search for genes that interacted with SAMD4A . A number were found , and we chose four that were detected in independent experiments and which were relatively short ( <60 kbp ) : TNFAIP2 , GCH1 , PTRF , and SLC6A5 ( Figure S2 ) . We initially verified that all five genes responded to TNFα by reverse-transcriptase PCR ( RT-PCR ) . No intronic RNA ( or only low levels in the case of PTRF ) copied from the five genes was detected before induction , but higher levels were seen within 10 min of TNFα treatment ( Figures 2A and S3F ) . Intronic RNA copied from further downstream in SAMD4A then appeared consistent with pioneering polymerases transcribing its 221 kbp at ∼3 kbp/min . Thus , RNA copied immediately downstream of the transcription start site ( tss ) appeared after 10 min , from ∼34 kbp into intron 1 after 30 min , from intron 3 after 60 min , and from the terminus after 85 min . In contrast , signal from each end of TNFAIP2 is seen by 10 min . This 11 kbp gene is so short , and synchrony sufficiently poor , that some polymerases in the population are initiating as others are terminating ( Figure 2A ) . GCH1 and SLC6A5—both genes of ∼60 kbp—present intermediate patterns; pioneering polymerases reach termini after ∼30 min , before a second ( reasonably synchronous ) transcription cycle begins ( Figures 2A and S3F ) . Such cycling has now been seen on various mammalian genes ( e . g . , [15] ) . Chromatin immunoprecipitation ( ChIP ) showed an enrichment of RNA polymerase II bound to the tss of all five genes within 10 min ( Figures 2B and S3G ) . It also showed that NF-κB bound to promoters within 10 min ( Figure S4 ) , as might be expected [16] . RNA fluorescence in situ hybridization ( FISH ) also shows that intronic RNA copied from the relevant parts of the genes is present at the appropriate times ( Figure S5 ) . Therefore , results obtained with four independent methods ( i . e . , microarrays , RT-PCR , ChIP , RNA FISH ) are in agreement and provide data on when polymerizing complexes are actively transcribing the sequences to be analyzed . These data are summarized in cartoons that accompany the results . Contacts between selected regions of SAMD4A and TNFAIP2 were monitored by 3C , where the presence of a band after 34 PCR cycles reflects a high contact frequency ( Figure 3 ) . Essentially no contacts are seen between the tss of TNFAIP2 ( the anchor ) and regions ∼25 kbp upstream or downstream of SAMD4A ( a , h ) at any time , or between the anchor and any region of SAMD4A ( b–g ) at 0 min—when no polymerases are engaged on either gene ( Figure 3B , cartoon ) . By 10 min ( when polymerases are first found on both genes; cartoon ) , contacts appear between the anchor and SAMD4A regions b , c ( Figure 3B ) . Such contacts are soon lost , as new ones appear further 3′ on SAMD4A; they seem to steadily “slide” down the long gene . Thus , by 30 min , contacts are with regions c and d , by 60 min with region e , and by 85 min with regions e , f , and g . ( The presence of more than one contact at certain times is consistent with imperfect synchrony amongst the ∼106 cells assayed . ) Treatment with DRB ( 5 , 6-dichloro-1-β-D-ribofuranosylbenzimidazole ) —a reagent that inhibits transcription and releases polymerases from the template ( Figure S6; [17]–[18] ) —reduces contacts ( Figure 3B , grey box ) . Similar changing contacts were seen using ( i ) real-time PCR to quantify selected interactions ( Figure S7 ) , ( ii ) the 3′ end of TNFAIP2 as an anchor ( Figure 3C , D; the gene is short enough for polymerases to be found at the same times on promoter and terminus in different cells in the population ) , and ( iii ) if HindIII replaced SacI as the restriction enzyme used for 3C ( Figure S8A , B ) . In every case , contacts are only seen at times when active polymerases are transcribing contacting sequences . Note that several genes lying within 50 Mbp on either side of SAMD4A do not interact with it ( e . g . , responsive NFKBIA , SAV1 , IRF9 , GPR68 , and PAPLN; non-responsive GMFB , YY1 , HIF1A , and C14orf2; and constitutive RCOR1; Figure S9A ) . As a whole , these results are inconsistent with the model involving tracking polymerases ( Figure 1 , left ) but are simply explained if the two contacting templates are transiently tethered to polymerases fixed in one factory ( Figure 1 , right ) . PTRF is a 21 kbp gene that lies on a different chromosome ( i . e . , 17 ) from SAMD4A ( on 14 ) . The pattern of interactions between the two is much the same as those seen between SAMD4A and TNFAIP2 ( Figure S3D , E ) , which is again consistent with the model involving fixed polymerases ( Figure 1 , right ) . A more complex pattern of changing contacts is seen between SAMD4A and a 60 kbp gene on chromosome 11 , SLC6A5 ( Figure 4 ) ; this pattern suggests that polymerases must be present on both contacting sequences . Thus , as before , no contacts are seen between the tss of SLC6A5 ( the anchor ) and regions upstream or downstream of SAMD4A ( a , h ) at any time , or between the anchor and any region of SAMD4A at 0 min—when no polymerases are engaged on either gene ( Figure 4B , cartoon ) . Again as before , contacts appear between the anchor and SAMD4A region c ( which includes the tss and the beginning of intron 1 ) after 10 min ( Figure 4B ) , when polymerases are first found on both . But after 30 min ( when contacts with region d were seen in Figure 3B ) , essentially no contacts are found ( Figure 4B ) . This is consistent with pioneering polymerases leaving the tss of the anchor so that they are now transcribing the 3′ end of this ∼60 kbp gene , as data in Figure 2 indicate . By 60 min ( when a second polymerase is just initiating on the tss of SLC6A5; Figure 2 ) , we see a strong ( second ) contact with the region on SAMD4A that its pioneering polymerase is now transcribing ( i . e . , e in Figure 4B ) . This interaction is DRB-sensitive ( Figure 4B , grey box ) , and so depends on continuing transcription . No prominent interactions are seen at 85 min ( Figure 4B ) even though we know SAMD4A is still being transcribed . Moreover , the contact seen with region f in Figure 3B is missing , presumably because the second polymerase on SLC6A5 has left the tss used as the anchor and is now transcribing the 3′ end ( Figure 2 ) . An almost identical pattern with analogous missing contacts is seen if HindIII replaces SacI during preparation of the 3C template ( Figure S8A , C ) . If the above explanation is correct , with contacts only being seen if active polymerases are present on both contacting partners , then use of the 3′ end of SLC6A5 as an anchor should change the pattern as follows . The two bands seen in Figure 4B should disappear ( as polymerases at the relevant times are on the tss and not the 3′ end now used as the anchor ) , while the two “missing” bands should reappear ( as polymerases have now reached the 3′ end ) ; they do . For example , comparison of Figure 4B and C shows that the first missing band/contact ( with d at 30 min in Figure 3B ) reappears in Figure 4C , as does the second ( with f at 85 min ) . Bands/contacts are also sensitive to DRB ( Figures 4B , C , grey boxes ) . This interpretation is reinforced by an analysis involving 5′ and 3′ anchors on another gene ( of similar length as SLC6A5 ) that lie on the same chromosome as SAMD4A . Thus , GCH1 is ∼0 . 8 Mbp away from SAMD4A and responds as rapidly to TNFα ( Figure S3F , G ) . When its 5′ and 3′ ends are used as anchors , a complex set of changing contacts ( and missing bands ) is again seen ( Figure S3A–C ) . We also confirmed that the tss of GCH1 lay next to the tss of TNFAIP2 at 10 min but not at 0 min ( Figure S9A ) . This is consistent with responding promoters coming together to the same factory when active . As all other contacts analyzed involve SAMD4A , these results also indicate that such reorganization is not peculiar to one long gene . If responding regions only lie together when transcribed , their nascent transcripts should also only be together at the appropriate times . To test this we used RNA FISH with pairs of probes each able to detect an intron within a single nascent transcript copied RNA transcript at its transcription site; colocalization of nascent transcripts copied from the two different genes then yields a yellow focus [9] , [19] . Yellow foci were given by the TNFAIP2 probe ( red ) and SAMD4A probes c , d , and e/f ( green ) at 10 , 30 , and 60 min post-induction ( Figure 5A–C ) . No such colocalization was seen at other times ( Figure S5 ) , when relevant regions were not being transcribed . As a control , we analyzed nascent transcripts copied from a non-responsive ( constitutively-active ) gene—RCOR1—that lies between SAMD4A and TNFAIP2 ( Figure S9A ) ; no yellow foci were detected ( Figure 5D ) . Just as 3C showed the templates lie together ( Figure 2 ) , RNA FISH confirms their transcripts also colocalize . We also investigated inter-chromosomal contacts 30 min post-induction , using probes targeting ( green ) SAMD4A region d and ( red ) SLC6A5 intron 1 ( close to the tss ) or intron 10 ( close to the 3′ end ) . When no 3C contacts between SAMD4A region d and the tss of SLC6A5 were seen ( Figure 4B ) , no yellow foci were detected ( Figure 5E; Figure S5C ) . But the “missing” 3C band was seen at 30 min using the 3′ terminus as anchor ( Figure 4C ) , and then yellow foci are seen ( Figure 5F ) . As a control , we analyzed nascent transcripts copied from another non-responsive ( constitutively-active ) gene—EDN1—that lies on a different chromosome; again , no yellow foci were seen ( Figure 5G ) . Electron microscopy reveals that nascent nucleoplasmic transcripts typically lie on the surface of ∼87 nm ( protein-rich ) factories [20] . To see if colocalizing transcripts encoded by the SAMD4A d:TNFAIP2 and SAMD4A d:SLC6A5 pairs lie this close together , we used a new approach that allows resolution beyond the diffraction limit of the light microscope [21]–[23] . We assume the red and green signals that yield a yellow focus ( e . g . , Figure S5B ) mark two sub-diffraction spots , fit Gaussian curves to their intensities , and measure the distance ( with 15 nm precision ) between peaks [23]; the distance between the two transcripts ranges from 7 to 102 nm , with a mean separation of 62 nm ( Figure 5H ) . This distribution is much like that seen when a pair of red and green points are repeatedly and randomly distributed in a 35 nm shell surrounding an 87 nm diameter sphere ( Figure 5H , orange line ) . [Subdiffraction-sized red/green fluorescent beads of 110 nm serve as a truly co-localizing control ( Figure S5B , left ) ; then , the distance between their red and green peaks is within the uncertainty of our measurements ( n = 8; not shown ) . ] These results are consistent with nascent transcripts copied from the two different genes lying on the surface of the same transcription factory .
We tested the two models illustrated in Figure 1 to address one fundamental assumption of modern molecular biology , namely that a transcribing polymerase tracks along its template as it makes its transcript . SAMD4A has a unique set of properties that make it particularly useful for this analysis; it can be switched on rapidly and synchronously by TNFα ( with approximately half the cells in the population responding ) , its length provides sufficient temporal and spatial resolution ( it takes ∼70 min to transcribe , and contains many restriction sites that facilitate the use of 3C to discriminate between contacts produced by different parts of the gene ) , and neither its sense or anti-sense strands encode other transcription units that might complicate analysis . 3C reveals that just the parts of SAMD4A being transcribed at a particular moment lie close to just the parts of three other genes being transcribed at that moment ( Figures 3 , 4 , S3 , and S8 ) . These inter-genic contacts occur infrequently , as expected [24]–[26] . RNA FISH confirmed that the relevant nascent RNAs lie together at the appropriate times ( Figures 5 and S5 ) , while “super-resolution” microscopy ( allowing measurements below the diffraction limit ) showed that the distance between the two transcripts is consistent with them lying within 35 nm of the surface of an 87 nm sphere ( Figure 5H ) . Such results are difficult—if not impossible—to explain if polymerases track . Rather , they are consistent with an alternative where two responding genes diffuse to an 87 nm factory to be transcribed by immobilized enzymes . Then , as the two genes are reeled in , only parts being transcribed at a given moment will lie transiently together [5] . These results beg many questions . For example , we were able to detect interacting sequences at a reasonable frequency simply by assuming the existence of factories dedicated to transcribing genes that respond rapidly to TNFα ( Figures S2 and S9 ) . If such specialized factories exist [27] , [28] , how many might there be in a nucleus , and how many are accessible to a gene like SAMD4A ? Fortunately , these questions will soon be answered , as techniques for analyzing all contacts made by any gene in a nucleus have been developed [29] . We also note that our results are consistent with others obtained from a recent genome-wide study; after stimulating human cells with estrogen and mapping contacts made by bound estrogen receptor-α ( using ChIP , 3C , and “deep” sequencing ) , contacting partners were often associated with bound RNA polymerase II [30] .
HUVECs from pooled donors ( Lonza ) were grown to 80%–90% confluency in Endothelial Basal Medium 2-MV with supplements ( EBM; Lonza ) , starved ( 18 h ) in EBM+0 . 5% FBS , and treated with TNFα ( 10 ng/ml; Peprotech ) for up to 85 min . In some cases , 50 µM 5 , 6-dichloro-1-β-D-ribofuranosylbenzimidazole ( DRB; Sigma-Aldrich ) was added 20 min before harvesting cells . 3C was performed as described [10] . In brief , 107 cells were fixed ( 10 min; room temperature ) in 1% paraformaldehyde ( Electron Microscopy Sciences ) , “Dounce”-homogenized , and membranes lyzed ( 30 min; 4°C ) using 0 . 2% Igepal ( Sigma-Aldrich ) . Nuclei were pelleted and resuspended in the appropriate restriction buffer , incubated ( 16 h; 37°C ) with SacI or HindIII ( 800 units/106 cells; New England Biolabs ) , diluted to 8 ml in ligation buffer , T4 DNA ligase added ( 4 , 000 units/106 cells; New England Biolabs ) , and incubated ( 48 h at 4°C , then 20 min at room temperature ) . After reversing cross-links ( 16 h; 65°C ) , DNA was purified by phenol extraction and ethanol precipitation , cut with BglII to reduce fragment length , and repurified . 71%–78% restriction sites in the template were cut by SacI or HindIII ( determined as in [31] ) . PCR conditions were adjusted so that reactions were within the linear range of amplification ( i . e . , ∼175 ng template/reaction; 1 . 75 mM MgCl2 , 1% dimethylsulphoxide , 10 pmoles of each primer , and GoTaq polymerase ( Promega ) ; 95°C for 2 min , then 34 cycles at 95°C for 55 s , 59°C for 45 s , and 72°C for 20 s , followed by one cycle at 72°C for 2 min ) ; amplimers were resolved on 2 . 5% agarose gels , stained with SYBR Green ( Invitrogen ) , and scanned using an FLA-5000 scanner ( Fuji ) . Identities of all 3C products were confirmed by DNA sequencing ( Geneservices , Oxford ) , except for those in Figure S8 ( where identities were confirmed by restriction digestion ) . Amplification efficiencies were examined using a control template generated by SacI or HindIII digestion of BAC clones covering GAPDH on HSA12 ( RP5-940J5; ImaGenes ) , SAMD4A , GCH1 ( RP11-170J16 , CTC-775N1 , CTD-2586I5 , CTD-2378G4; CHORI , Invitrogen ) , and TNFAIP2 ( CTD-2594N9; Invitrogen ) on HSA14 , SLC6A5 on HSA11 ( RP11-120F6; CHORI ) , and PTRF on HSA17 ( RP11-194N12; CHORI ) followed by ligation . This synthetic template was spiked ( to reach 175 ng/µl ) with HUVEC DNA cut with the relevant restriction enzyme and ligated . Other control templates included non-digested/ligated DNA and digested/non-ligated DNA ( both from 106 cells ) . Results shown were reproduced using at least two independently obtained templates .
|
We were all taught that an RNA polymerase becomes active by diffusing to a promoter , initiating transcription , and then tracking like a locomotive down the DNA template . We test this using tumour necrosis factor alpha ( TNFα ) to switch on transcription of two human genes which lie far apart on the genetic map and then measure how close the two are in 3D nuclear space . If what we were taught were true , there is no reason to expect the two genes to lie together . What we find—using two different techniques ( cutting/ligating nearby sequences , and super-resolution microscopy ) —is that the two genes are initially apart; then the parts of the genes being transcribed at a particular moment transiently come into close proximity . Our results are consistent with a model in which genes diffuse to a cluster of polymerases—a transcription factory—with transcripts being made as immobile polymerases reel in their respective templates . The DNA moves , not the polymerase .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"molecular",
"biology/transcription",
"initiation",
"and",
"activation",
"genetics",
"and",
"genomics/gene",
"expression",
"cell",
"biology/nuclear",
"structure",
"and",
"function",
"genetics",
"and",
"genomics/nuclear",
"structure",
"and",
"function",
"molecular",
"biology/transcription",
"elongation",
"biochemistry/transcription",
"and",
"translation",
"cell",
"biology/gene",
"expression"
] |
2010
|
Active RNA Polymerases: Mobile or Immobile Molecular Machines?
|
Information is encoded in neural circuits using both graded and action potentials , converting between them within single neurons and successive processing layers . This conversion is accompanied by information loss and a drop in energy efficiency . We investigate the biophysical causes of this loss of information and efficiency by comparing spiking neuron models , containing stochastic voltage-gated Na+ and K+ channels , with generator potential and graded potential models lacking voltage-gated Na+ channels . We identify three causes of information loss in the generator potential that are the by-product of action potential generation: ( 1 ) the voltage-gated Na+ channels necessary for action potential generation increase intrinsic noise and ( 2 ) introduce non-linearities , and ( 3 ) the finite duration of the action potential creates a ‘footprint’ in the generator potential that obscures incoming signals . These three processes reduce information rates by ∼50% in generator potentials , to ∼3 times that of spike trains . Both generator potentials and graded potentials consume almost an order of magnitude less energy per second than spike trains . Because of the lower information rates of generator potentials they are substantially less energy efficient than graded potentials . However , both are an order of magnitude more efficient than spike trains due to the higher energy costs and low information content of spikes , emphasizing that there is a two-fold cost of converting analogue to digital; information loss and cost inflation .
Information is encoded , processed and transmitted in neural circuits both as graded potentials ( continuous , analogue ) and action potentials ( pulsatile , digital ) . Although sensory and chemical synaptic inputs to neurons are graded [1] , in most neurons these are converted into a train of action potentials . This conversion overcomes the attenuation of graded signals that occurs as they are propagated over long distances within the nervous system [2] , and may prevent noise accumulation in neural networks because pulsatile signals are restored at each successive processing stage [3] , [4] . However , because spike trains use discrete pulses of finite precision they have a lower dimensionality than analogue voltage signals , reducing their signal entropy [4] . Consequently , spike trains can encode fewer states within a given time period than analogue voltage signals . This is borne out by experimental measurements that show the conversion of the graded generator potential into a spike train reduces the information rate [5]–[7] . Thus , non-spiking neurons that encode information as graded potentials typically have much higher information rates than spiking neurons [5] , [8] , [9] . A drop in the energy efficiency of information coding has also been suggested to accompany the conversion of graded to action potentials [3] , [10] . Neuronal energy consumption is dominated by the influx/efflux of ions , which must be pumped back across the cell membrane by the Na+/K+ ATPase consuming ATP [3] , [11] , [12] . These ion movements can incur substantial energy costs even in graded potential neurons [3] , [13] . However , the large Na+ influx during action potentials requires additional cellular energy to extrude , though the precise energy cost will vary among neuron types [11] , [14]–[16] . Our aim is to identify the causes of the loss of information and energy efficiency when graded potentials are converted to action potentials . Although some causes of information loss in spiking neurons have been studied previously , such as channel noise [17]–[19] or dimensionality reduction [6] , [20] , in most cases their effects on information rates have not been quantified . We quantified both information rates and energy efficiency using single compartment models . We compared the information rates , energy consumptions and energy efficiencies of spike trains with those of the generator potentials that triggered the spike trains , and of the graded response produced in the absence of voltage-gated Na+ channels . We find that three previously unreported effects reduce the information rate and efficiency of the generator potential by 50%; namely the finite durations of action potentials , and the noise and nonlinearity introduced by voltage-gated ion channels . The effect of channel noise on spike timing reduces the information rate and efficiency by <10% . We conclude that the conversion of graded signals to “digital” action potentials imposes two penalties; spikes increase energy costs and both spike coding mechanisms and the spike code reduce information rates . As a result energy efficiency falls by well over 90% .
To determine the effect of noise generated by the voltage-gated Na+ and K+ channels on the information rates of the spiking neuron model , we replaced either the stochastic Na+ or K+ channels with deterministic channels thereby eliminating this component of the channel noise . In comparison to the stochastic model , the deterministic Na+ channel model generated more reliable spike trains for a given stimulus ( Figure S1A , S2A ) . Similarly , replacing the stochastic K+ channels in the spiking neuron model with deterministic channels also generated more reliable spike trains for a given stimulus in comparison to the original spiking neuron model ( Figure S1A , S2B ) . We quantified differences in the reliability between the original stochastic spiking neuron model , the modified model with deterministic Na+/stochastic K+ channels , and the modified model with stochastic Na+/deterministic K+ channels . We compared the total entropy , noise entropy , information rate and information per spike for spike trains generated by low mean , high standard deviation stimuli or high mean , low standard deviation stimuli ( Figure 4 ) . All three models produced 50–57 spikes/s in response to the stimuli ( Figure 4A ) . In comparison to the original spiking neuron model with stochastic Na+ and K+ channels , the total entropy of the deterministic K+ channel model was lower by 1–7% , whereas the total entropy of the deterministic Na+ model was almost identical ( Figure 4B ) . The deterministic K+ channel model also had the lowest noise entropy , making the APs more reliable ( Figure 4C ) . Both models with deterministic ion channels had higher information rates than the original model because of their lower noise entropy , but the difference was just 7% , irrespective of the stimulus statistics ( Figure 4D ) . This suggests that channel noise has relatively little impact on the information rate of the 100 µm2 single compartments we modelled . Thus , in addition to channel noise and dimensionality reduction , there must be other sources of information loss . The information in the spike train comes from the generator potential ( Figure 5A ) . However , the generator potential is not equivalent to the voltage signals produced by the graded potential model , which lacks voltage-gated Na+ channels . We constructed an approximation of the generator potential , the pseudo-generator potential , by removing the action potentials from spike trains and replacing them with a 6 ms linear interpolation of the membrane potential , corresponding to the maximum action potential width ( Figure 5A ) . The pseudo-generator potential probability density function is distorted in comparison to the graded potential being narrower with a more pronounced peak because voltage excursions beyond threshold are truncated , the action potential being replaced with an interpolated response ( Figure 5A , B ) . For a particular stimulus the information rate of the pseudo-generator potential was intermediate between that of the spike trains and that of the graded potential model ( Figure 5C ) . The information rates of the pseudo-generator potential were highest ( 1094 bits/s ) with low mean , high standard deviation stimuli , 860 bits/s ( 366% ) higher than that of the corresponding spike trains but 1146 bits/s ( 51% ) lower than that of the corresponding graded potential ( Figure 5C ) . The information rates of the pseudo-generator potential were lowest ( 188 bits/s ) with low mean , low standard deviation stimuli . This lowest value was 158 bits/s ( 531% ) higher than that of the corresponding spike trains , but 352 bits/s ( 65% ) lower than that of the corresponding graded potential ( Figure 5C ) . What reduces the information rate of the pseudo-generator potential relative to the graded potential ? We identify three processes: the duration of the action potential and associated refractory period , and two effects caused by the presence of voltage-gated Na+ channels , noise and non-linearity . We will assess each of these processes , in turn . The action potential and accompanying refractory period creates a ‘footprint’ on the generator potential during which information is lost ( Figure 6A ) . To assess the impact of this ‘footprint’ on the information rate , we stimulated the graded model with a white noise stimulus ( Figure 1A , B ) to generate a set of graded responses from which we could estimate the signal , noise and information rate . These graded responses produced a high information rate ( 1427 bits/s ) . We then inserted 6 ms long sections of linear interpolation spaced at least 10 ms apart into the individual graded responses to mimic action potential footprints ( Figure 6B ) . We added between 10 and 80 linear interpolations per second into each response to represent the spike footprints at different firing rates and re-calculated the Shannon information rate ( Figure 6B ) [25] . Interpolations were added at exactly the same positions in all responses , termed deterministic interpolation ( Figure 6B ) , to represent the footprints of noise-free spikes and give an upper bound on signal entropy . The placement of the interpolations was then jittered by up to 4 ms ( Figure 6B ) , termed jittered interpolation , to represent reliable spike trains with low noise entropy . Finally , interpolations were placed randomly in each response ( Figure 6B ) , termed random interpolation , to resemble unreliable spike trains with high noise entropy . The Shannon information rate [25] was unaffected by the deterministic or jittered interpolation , irrespective of the number of interpolations inserted ( Figure 6C ) because it depends only upon the signal-to-noise ratio ( SNR ) and the response bandwidth [25] . Thus , inserting increasing numbers of interpolations , even when jittered , does not affect the Shannon information rate because these interpolations are inserted in identical ( deterministic ) or similar ( jittered ) positions , leaving the regions between the interpolations unaffected . Conversely , increasing the number of random interpolations reduced the Shannon information rate from 1427 to 485 bits/s ( Figure 6C ) because these interpolations add noise to the responses , thereby reducing the SNR . In addition to the Shannon information rate [25] , we calculated coherence-based information rates to determine the effect of the footprint on information loss from the stimulus ( see Methods ) . The coherence-based estimate of the information rate is a measure of linear dependence between the stimulus and the response , and describes different forms of signal corruption including non-linear distortion [29] . The coherence-based information rate decreased as the number of interpolations inserted increased for all three types of interpolation , deterministic , jittered and random ( Figure 6D ) . The coherence-based information rate dropped from 1148 bits/s with no interpolations to 346 bits/s with 80 interpolations . Although we inserted linear interpolations into the voltage responses , there is still a fluctuation at the corresponding position in the current stimulus . The mismatch between the interpolations and the stimulus may reduce the coherence-based information rate by inflating the non-linearity . To determine whether this is the case , we added linear interpolations at exactly the same positions to both the stimulus and the response , and recalculated the coherence-based information rate ( Figure 6D ) . This difference between the coherence-based information rates calculated with or without interpolations added to the stimulus as well as the response is the information lost due to the action potential footprint . For the same number of interpolations , all three types of interpolation , deterministic , jittered and random , had higher information rates ( between 177 and 516 bits/s ) with interpolations added to the stimulus than without ( Figure 6D ) . These coherence-based information rates were dependent upon the number of interpolations inserted . For example , inserting 10 interpolations reduced the information rate from 1148 bits/s to 1090 bits/s but inserting 80 interpolations reduced the information rate to 860 bits/s . Thus , the coherence-based method demonstrates that the action potential footprint blanks out information about the stimulus . This loss of information increases with spike rate from 5 . 3% at 10 Hz to 33 . 5% at 80 Hz . Channel noise affects sub-threshold potentials as well as spike timing and reliability [30] . We measured the standard deviation of the voltage noise at sub-threshold membrane potentials for the spiking neuron model , the deterministic Na+/stochastic K+ channel model , the stochastic Na+/deterministic K+ channel model and the graded neuron model ( Figure 7A ) . In the absence of an input stimulus , the voltage noise was generated entirely by the spontaneous opening and closing of the voltage-gated ion channels . The noise standard deviation of all the models was highest at the most depolarised potentials and dropped as the membrane potential was hyperpolarised towards the reversal potential of the K+ ions ( Figure 7A ) . Between −74 to −70 mV the voltage noise standard deviation was highest for the spiking neuron model and lowest for the stochastic Na+/deterministic K+ channel model . The voltage noise of the deterministic Na+/stochastic K+ was close to that of the spiking neuron model ( Figure 7A ) . However , near the K+ reversal potential of −77 mV the voltage noise of all three models containing stochastic K+ channels dropped as the driving force on K+ ions approached zero . The drop was less pronounced in the spiking neuron model because stochastic Na+ channels continued to produce noise . Below the K+ reversal potential , the voltage noise of all three models containing stochastic K+ channels increased ( Figure 7A ) , with the driving force on K+ ions . The voltage noise of the deterministic K+ channel model dropped as the membrane potential was hyperpolarised , even below K+ reversal potential , because the probability of spontaneous Na+ channel opening , the only source of channel noise , drops at hyperpolarised potentials . Indeed , the deterministic K+ channel model had the lowest voltage noise at holding potentials more depolarised than ∼−74 mV and more hyperpolarised than ∼−80 mV ( Figure 7A ) . Thus , although the noise generated by the spontaneous opening of both Na+ and K+ channels contributes to the voltage noise of the spiking neuron model , the K+ channel noise apparently makes the greater contribution at potentials between −74 to −70 mV . Note that the voltage noise standard deviation with both channel types together is less than the sum of the standard deviations of the individual channel types because their variances add . We assessed the impact of the sub-threshold voltage noise on the Shannon information rate by stimulating each model with a white noise current with a zero mean and low standard deviation ( μ = 0 , σ = 1 , τc = 3 . 3 ms ) . An additional tonic current was injected and adjusted to hold the mean membrane potential at either −77 or −70 mV . This tonic current prevented the models containing voltage-gated Na+ channels from reaching threshold , permitting a direct comparison of the effects of stochastic and deterministic channel combinations upon sub-threshold information coding . We calculated the Shannon information rate [25] of each model at the two mean potentials , −77 and −70 mV ( Figure 7B ) . The highest information rates of all the models occurred at the more hyperpolarised potential because the voltage noise was lower . Due to a distinct drop in voltage noise near the K+ reversal potential , the deterministic Na+/stochastic K+ channel model and the graded neuron model , attain the highest information rates of 3123 bits/s at −77 mV . These information rates were ∼30% greater than those of the sub-threshold spiking neuron model and the stochastic Na+/deterministic K+ channel model , which are lower because of voltage-gated Na+ channel noise . At −70 mV the increased voltage noise in all the models reduces their information rates ( Figure 7B ) . The information rate of the sub-threshold spiking neuron model dropped 86% to 321 bits/s . The sub-threshold information rates of both models with stochastic K+ channels dropped 63% to 1142–1168 bits/s , whilst the stochastic Na+/deterministic K+ channel model has the lowest voltage noise and , consequently , the highest sub-threshold information rate of 1288 bits/s . The drop in the information rates of all the models at the more depolarised holding potential shows the substantial effect of channel noise upon the sub-threshold and graded potentials . The combination of both stochastic Na+ and stochastic K+ ion channels in the spiking neuron model reduce the information content of the sub-threshold potential relative to the graded neuron model by 24% at −77 mV to 73% at −70 mV . Voltage-gated ion channels introduce non-linearities [31] , [32] that could reduce the information content of the generator potential by distorting the voltage signal . We assessed the sub-threshold effect of non-linearity on each of the models , at −77 mV and −70 mV , using the coherence-based information rates we previously calculated to assess the impact of the action potential footprint ( Figure 6D ) . Higher coherence-based information rates indicate better reconstruction of the original stimulus , based solely on linear decoding principles [29] . In the spiking neuron model the coherence-based information rates dropped by more than 63% as the holding potential becomes more depolarised i . e . , from 1027 bits/s at −77 mV to 382 bits/s at −70 mV ( Figure 7C ) . This fall indicates a decline in the quality of linear reconstruction . By comparison , the stochastic Na+/deterministic K+ model was the least affected by depolarisation , the coherence-based information rates dropping by just 1 . 5% . For the model with deterministic Na+/stochastic K+ and the model with only stochastic K+ channels , the coherence-based information rates drop ∼4 . 2–4 . 8% at the more depolarised potential ( Figure 7C ) . Increasing the holding potential to −68 mV causes all three models containing voltage-gated Na+ channels to produce spikes , making them increasingly non-linear ( data not shown ) . In addition to coherence-based information rates , we used the normalised root mean squared error ( nRMSE ) between the original stimulus and the reconstructed stimulus to assess the effect of non-linearity . An nRMSE value that tends towards zero represents perfect reconstruction [29] . The nRMSE increased as the membrane potential increased indicating a drop in the quality of reconstruction ( Figure 7D ) ; the increase in nRMSE was largest for the sub-threshold spiking model ( 67% ) but the nRMSE of the three other models also increased by 8–13% . This decline in reconstruction quality is due to an increase in the open channel probability with depolarization . For the models containing voltage-gated Na+ channels , the voltage threshold for eliciting an action potential is close to −68 mV . At −70 mV the increase in the numbers of open voltage-gated Na+ channels increases positive-feedback and , consequently , the magnitude of the non-linearity . A fluctuating input stimulus superimposed upon the holding current also reduces the distance from the voltage threshold , though the effect of this on reconstruction will depend on the magnitude and polarity of the fluctuations . Using linear systems analysis ( see Methods ) , we assessed how much of the input ( current ) can be predicted from the response ( voltage ) by reconstructing the input stimulus current . We find that when the graded voltage response was used for the reconstruction based on linear decoding the predicted input stimuli were most coherent , with the lowest nRMSE ( Figure S3C , D ) and the highest coherence-based information rates ( Figure S3C , E ) . The reconstruction accuracy ( nRMSE and coherence based information ) of the pseudo-generator potentials was lower than that of the graded potentials ( Figure S3B , D , E ) . The highest nRMSE and , consequently , the lowest coherence-based information rate was obtained from reconstructions based on action potentials ( Figure S3A , D , E ) , although these were only marginally worse than reconstructions based on pseudo-generator potentials ( Figure S3 ) . Thus , voltage-gated Na+ channels distort both the subthreshold ( pseudo-generator ) and suprathreshold responses so that the incoming stimulus current cannot be accurately reconstructed using just a linear decoder . Neuronal information rates are constrained by extrinsic noise in the input stimuli , as well as by intrinsic noise generated by ion channels [33] , [34] . To investigate this constraint , we added broadband Gaussian noise to the white noise input stimulus . This enabled us to quantify and compare the effect of extrinsic noise upon the information rates of the spiking model , the pseudo-generator potentials from the spiking model and the graded model . In our simulations , although the presence of the extrinsic noise source facilitates a marginal increase in precision of the APs for inputs with low standard deviations , it does not alter the variability of the APs , consequently noise-aided enhancement of mutual information is absent ( cf . McDonnell et al . [35] ) . The amount of extrinsic noise was altered to produce an input stimulus with either a low or a high SNR input stimulus ( Equations 2 and 3; SNR = 2 or 20 ) . The SNR is defined as the ratio of the signal power to the noise power . In our simulations , we decreased the SNR by increasing the noise power ( see Methods; Equation 3 ) . For the spiking model , increasing the input noise produces a relatively small increase in total entropy ( ∼5% , SNR = 2; ∼2% , SNR = 20 ) ( Figure S4A ) but a relatively large increase in noise entropy ( ∼180% , SNR = 2; ∼50% , SNR = 20 ) ( Figure S4B ) , and this produces a significant drop in the mutual information ( ∼40% , SNR = 2; ∼10% , SNR = 20 ) ( Figure S4C , S5A ) . The information rates of the pseudo-generator potentials also decrease with increased extrinsic noise ( Figure S5B ) . The loss in relation to the noise-less stimulus is greater in the pseudo-generator potentials ( ∼69% , SNR = 2; ∼29% , SNR = 20 ) with higher standard deviation input signals . A 10-fold increase in the input SNR caused a 133% increase in information rate , from 335 bits/sec ( SNR = 2 ) to 780 bits/sec ( SNR = 20 ) , compared to 1094 bit/sec in the absence of extrinsic noise . Likewise , the information rates of the graded model were reduced by up to 73% for low SNR input signals ( SNR = 2 ) and by up to 36% for high SNR input signals ( SNR = 20 ) ( Figure S4C ) , the higher quality input signal ( SNR = 20 ) causing the information rate to increase from 595 to 1422 bits/sec . Thus , the information rates of the spiking model were the least affected by the extrinsic noise whilst those of the graded model were the most affected ( Figure S5A–C ) . The energy consumption of each model was determined from the K+ ion fluxes across the membrane needed to generate the voltage signals , as the number of ATP molecules hydrolyzed by the Na+/K+ pump [12] . This pump maintains the ionic concentration gradients that generate electrical responses and operates stoichiometrically , pumping back 2 K+ ions for every ATP molecule that it consumes [36] . The energy consumption of the spiking neuron model is strongly correlated with its firing rate ( Figure 8A ) because the energy consumption of an action potential is high compared to the consumption between action potentials . Higher standard deviation stimuli evoke larger membrane potential fluctuations , eliciting more action potentials and , therefore , consuming more energy . Consequently , the high mean , high standard deviation stimuli that evoked the highest firing rates also incurred the highest energy consumption , 3 . 9*108 ATP molecules/s ( Figure 8A ) . Low mean stimuli with high standard deviations consume 3 . 1 times more energy than stimuli with low standard deviations but for high mean stimuli it is just 1 . 4 times more ( Figure 8A ) . This is because the standard deviation of signal fluctuations has less of an effect upon the average firing rate with high mean input stimuli . Pseudo-generator membrane potentials consume less energy than the spiking neuron model . Indeed the maximum energy consumption of the pseudo-generator potentials is 6 . 4*107 ATP molecules/s , almost an order of magnitude less than the spiking neuron model ( Figure 8B ) . Like the spiking model , when the pseudo-generator potential model is driven with a high mean stimulus , increasing the stimulus standard deviation increases energy consumption . But , unlike the spiking model , when the stimulus mean is low , increasing its standard deviation reduces energy consumption . Low mean , high standard deviation stimuli consume less energy because they hyperpolarise the membrane potential by 10 mV or more below the resting potential , and this reduces the number of open K+ channels ( Figure S6A , B ) . Conversely , with high mean stimuli the maximum peak-to-peak voltage of the compartment is approximately the same , irrespective of the standard deviation ( Figure S6A , B ) . The greater energy consumption of the high standard deviation is due to the 1 . 6-fold greater numbers of open K+ channels , which cause a doubling of the mean K+ current at equivalent membrane potentials , thereby inflating the energy consumption . The energy consumption of the graded model showed the same trends as the pseudo-generator potentials ( Figure 8C ) . Again , less energy is consumed in response to low mean high standard deviation stimuli than to low standard deviation stimuli , due to an 85% decrease in the number of open K+ channels ( Figure S7A , B ) . In contrast , at high means , high standard deviation stimuli consumed 64% more energy than low standard deviation stimuli ( Figure 8C ) because high input standard deviations open greater numbers of K+ channels ( Figure S7A , B ) . We calculated the energy efficiency of information coding by dividing the information rates of the spiking neuron model , the pseudo-generator potentials and the graded neuron model by their corresponding energy consumptions . The energy efficiency of the spiking neuron model was highest ( 8 . 4*10−7 bits/ATP molecule ) for low mean , high standard deviation stimuli and lowest ( 3 . 8*10−8 bits/ATP molecule ) for high mean , low standard deviation stimuli ( Figure 9A ) . This 22-fold difference in energy efficiency was accompanied by a 23-fold difference in information rate . Thus the coding of low mean , high standard deviation stimuli was most efficient because these stimuli generated the highest information rates with firing rates , and therefore energy costs , similar to high mean , low standard deviation stimuli ( Figure 9A ) . In other words , energy efficiency rises with information per spike . Indeed , in all models , spiking , pseudo-generator potential , and graded , increasing input stimulus mean reduced energy efficiency because it increased the mean level of response without introducing more information ( Figure 9A , B ) . As expected , the energy efficiency of all three models improved when the information rate increased in response to an increase in stimulus standard deviation at a given stimulus mean ( Figure 9A , B ) . For example at low means , the spiking model's information rate increased by 689% with a concomitant increase in efficiency of 151% . For the pseudo-generator potentials information increased by 482% , and efficiency increased by 889% and in the graded neuron efficiency increased by 363% and information increased by 315% . Both pseudo-generator ( 8 . 0*10−5 bits/ATP ) and graded potential ( 1 . 3*10−4 bits/ATP ) models were 95–156 times as energy efficient as the spiking model ( 8 . 4*10−7 bits/ATP ) , when all models were compared with low mean , high standard deviation inputs . At higher information rates the energy efficiency of both the pseudo-generator and graded potentials improved substantially ( Figure 9B ) . However , the graded potentials achieved higher information rates than the pseudo-generator potentials and in this regime they were as much as 1 . 6 times more energy efficient at 1 . 3*10−4 bits/ATP molecule . The addition of extrinsic noise did not affect this general pattern of relationships between input stimuli , information rate and energy efficiency in the three models . However , by reducing the information rates of all three models the extrinsic noise reduced the energy efficiency for any given input stimulus ( Figure 9A , B ) . For example , adding noise to the inputs reduced the efficiencies of the pseudo-generator potentials by 71% for a low quality input ( SNR = 2 ) and by 26% for a high quality input ( SNR = 20 ) . Similarly , the efficiency of the graded potential model dropped by 74% at low SNR and by 36% for high input SNR . Given that extrinsic noise only marginally altered the energy consumption , it decreases efficiency by decreasing the amount of information that can be coded .
Analogue voltage signals in non-spiking neurons and generator potentials in spiking neurons typically have higher information rates than spike trains [5]–[9] . This information loss is a consequence of a change in coding strategy; non-spiking neurons and generator potentials encode information as a continuous analogue voltage signal whereas spiking neurons use discrete pulses of finite precision and width , limiting the number of states that can be coded within a given time period . However , there are also biophysical causes of this information loss , and these were the focus of our study . Spiking neurons can be lossless encoders of band-limited inputs if their spike rates exceed the Nyquist limit [37] , both at the level of a single neuron or across a population of neurons [38] , [39] . But below this limit information loss occurs and is affected by the factors we have examined . Our simulations show that voltage-gated Na+ channels , which are necessary for action potential generation , are the primary biophysical cause of information loss in sub-threshold potentials because they increase intrinsic noise and introduce non-linearities . Indeed , this information loss in sub-threshold potentials is greater than the information loss in spike generation attributable to voltage-gated Na+ channels . Further information loss in the sub-threshold potential occurs because each action potential obscures the generator potential , reducing its information content . This suggests that the biophysical factors we identify have their major impact upon sub-threshold information processing . Comparing the energy efficiencies of our models , spike trains consume an order of magnitude more energy than graded or pseudo-generator potentials for a given stimulus . This result emphasizes the two-fold penalty of action potentials on coding efficiency; lower information rates and higher energy costs . Graded and generator potentials consume similar amounts of energy , the primary determinant of which is the input mean , but due to their lower information rates generator potentials are less energy efficient than graded potentials . Our models contained voltage-gated ion channels with the same biophysical properties as those found in the squid giant axon because well-established kinetic models exist for them [40] , [41] . Different channel kinetics will alter channel noise [42] , affect the shape of the action potential [11] and alter the information rates of a spiking model [43] . However the main effects of channel noise in our models are on the graded and generator potentials . Previous modeling studies have used squid voltage-gated Na+ channels to show that they increase sub-threshold noise [30] , but did not quantify their effect on sub-threshold information rates . We find that the noise from voltage-gated Na+ channels and voltage-gated K+ channel noise substantially decreases the information rate of the generator potential . This finding suggests that the high densities of voltage-gated Na+ channels at the spike initiation zone [44] , as well as voltage-gated Na+ channels and Ca2+ channels in dendrites and dendritic spines [45] , [46] could also reduce the information rate of sub-threshold signals , and this could have a deleterious effect on information processing . Our models suggest that action potential duration ( including the absolute refractory period ) is an important source of information loss , imposing a lower limit on the interspike interval and preventing the spike initiation zone from integrating new information for a brief period . In vivo many neurons have considerably higher spike rates than our models , which had moderate spike rates below approximately 90 Hz . At these high spike rates , substantial portions of the information would be lost from the generator potential , promoting narrower action potentials and sparse codes that require relatively few action potentials [47] . However , many neurons use signals that are considerably longer than typical action potentials such as bursts and plateau potentials [48] that obscure far more of the generator potential and incur a greater information loss . This emphasizes the importance of these long-duration signals as indicators of high salience signals . The non-linearity of all the models incorporating voltage-gated Na+ channels increases with the sub-threshold depolarization because the positive feedback generated by the Na+ channels increases as the threshold approaches . Thus , at sub-threshold levels the Na+ channels distort the voltage waveform . This distortion could reduce the information content of the sub-threshold potentials , though this depends upon whether the transformation of any synaptic metric ( current , conductance , etc . ) into the voltage waveform is linear in the sub-threshold regime . Linear as well as non-linear mapping may occur between the synaptic input and the resultant voltage waveform [20] , [49] . Voltage-gated Na+ channels may constitute one such non-linearity , distorting the synaptic input [50] . In such cases , although a linear decoder cannot fully represent and recover the input information , a decoder relying on higher order features of the membrane potential may prevent any information loss ( also see [39] ) . Our use of current rather than conductance as the input stimulus ignores the energy cost associated with conductance inputs , which will reduce the energy efficiency of information coding of all the models . Conductance inputs close to the spike initiation zone will also alter the membrane time constant and affect action potential initiation [51] , [52] . Consequently , conductance inputs will affect the bandwidth and temporal precision of all the models and the maximum spike rate of the spiking neuron model [53] . The synaptic channels needed to implement the conductance changes will also contribute noise to the models [54] , reducing their information rates . By incorporating extrinsic noise , however , we have shown that the relationships we have found will remain qualitatively similar . The squid giant axon action potentials that we modeled consume substantially more energy than other vertebrate and invertebrate action potentials [11] , [14]–[16] , inflating the energy consumption of the spiking neuron model and reducing its efficiency . Nevertheless , the efficiency drop that occurs when generator potentials are converted to action potentials is substantial and will remain , albeit with a smaller difference . The topological class of model ( e . g . Type I or Type II ) may also influence energy consumption through the dynamics and time course of the ionic and synaptic currents determining the threshold manifold [55] . Indeed , minimizing metabolic consumption in single compartment models [55] leads to the leak and the inward currents competing with each other even before reaching the spiking threshold , via a Hopf bifurcation ( Type II ) . This causes an increase in energy consumption forcing the optimal action potentials to steer away from such bifurcations; gradient descent on metabolic consumption leads to saddle-node bifurcations as in Type I cortical neurons ( unpublished observation – BS , personal communication – Martin Stemmler , [55] ) . The energy consumption of the graded potential neurons will also be affected by changes in the biophysical properties of voltage-gated ion channels , though this is unlikely to substantially affect the relationship between the input stimuli and the energy consumption . Our models systematically explored combinations of the mean and standard deviation of a Gaussian input . Those spike trains with the lowest information rates and bits per spike were evoked by low standard deviation stimuli , whereas high standard deviation stimuli evoked consistently higher information rates for a given mean stimulus . Consequently , across all our models there was no systematic relationship between the mean spike rate and the information rate , total entropy , noise entropy or coding efficiency ( bits per spike ) . Indeed , the highest and lowest information rates and coding efficiencies were found at similar spike rates . However , these findings are specific to the type of stimuli we used , a randomly varying input signal superimposed on an offset . It is more usual to find that the information rate increases with spike rate whilst the coding efficiency declines because the entropy per spike falls [56] , [57] . Non-Gaussian naturalistic stimuli vary more widely than do Gaussians . These larger excursions make the voltage response more nonlinear and engage adaptation mechanisms that , if they affect the signal and noise differently , can change the information rates of both graded potentials and the spike trains they generate [58] . Although there are methods that could allow us to compare the coding and metabolic efficiency of analogue and spiking responses to natural stimuli [6] , each modality has its own statistics . Even within a modality different classes of neuron have distinctive firing patterns because they select different components of the input ( e . g . retinal ganglion cells [57] ) . Faced with many particular cases , we chose to start with a general stimulus that identifies factors , such as input signal to noise , that are widely applicable . As a case in point , in many neurons the mean and standard deviation of the input stimuli and the extrinsic noise are often correlated [59]–[61] . For example , extrinsic noise in synaptic inputs is often correlated with their number and strength , and hence signal amplitude [62]–[64] . Thus , the stimulus space investigated with our models exposes relationships between energy efficiency and information rate that are broadly applicable to a number of different types of neuron . In particular , our models demonstrate that the energy efficiency of spiking neurons can be improved by reducing the mean input and increasing the standard deviation of the signal . Graded neurons achieve this by using predictive coding to eliminate the mean and amplify the remaining signal to fill their output range [65] , [66] and these procedures increase both their coding efficiency and their energy efficiency [3] . Our findings demonstrate that spiking neurons can do likewise . Taken together , our analyses show that the biophysical mechanisms involved in action potential generation contribute significantly to the information loss that accompanies the conversion of a graded input to a spike train . Although we cannot directly relate the proportions of information loss to specific mechanisms , it seems likely that the action potential ‘footprint’ and sub-threshold voltage-gated Na+ channel noise are the major sources of information loss . Viewed as a cost-benefit trade-off , action potentials incur penalties ( information loss and energy cost ) that are , presumably , balanced against being able to transmit information over considerable distances and preventing noise accumulation during successive processing stages . Reducing the distances over which information is transmitted in the nervous system may favor less conversion of graded signals into spike trains [67] . However , problems associated with accumulating noise during successive processing stages [4] may remain severe . Thus , even in some highly miniaturized nervous systems , neurons with action potentials are likely to be necessary [67] . In conclusion , our modeling of single compartment neurons confirms that a critical step in neural coding , the conversion of an analogue sub-threshold signal to a series of discrete “digital” pulses , is accompanied by substantial information loss . We show that voltage-gated Na+ channels , critical components for the conversion of analogue to digital , reduce the information in sub-threshold analogue signals substantially , and that this loss is compounded by interference from action potentials . Thus , the first step in a hybrid processing strategy to increase efficiency [4] , [68] , the analogue processing of inputs , is compromised by mechanism used for the second step , the conversion of analogue to digital , and this calls for strategic placement of the spike initiation zone [69] . Some neurons appear to mitigate a small fraction of the loss of information that accompanies the conversion of analogue to digital by transmitting both analogue and digital [70]–[72] . Information may be encoded in the height and width of action potentials [71]–[74] suggesting that spiking neurons may transmit more information than is calculated by treating them as digital pulses . Even in these cases , however , the ‘footprint’ of the action potentials and sub-threshold voltage-gated Na+ channel noise are still likely to cause substantial information loss .
We used a single compartment stochastic Hodgkin-Huxley model of the squid giant axon for our simulations [40] , [41] . The model supporting spiking contained two voltage-gated ion channels , transient Na+ and a delayed rectifier K+ along with the leak conductance , while the model producing purely graded signals contained delayed rectifier K+ and leak conductances . The dynamics of the membrane potential is governed by a set of activation and inactivation variables m , h and n with the current balance equation , ( 1 ) Cm is the membrane capacitance , are the conductance of the Na+ , K+ and leak channels respectively , are the respective reversal potentials , is a time dependent current stimulus and is the input ( extrinsic ) stimulus noise current . is zero for no input noise simulations . The variables m , h and n follow first order kinetics of the form , where is the steady-state ( in ) activation function and is the voltage-dependent time constant . The model was driven using a time dependent current – , a 300 Hz Gaussian white noise , filtered using a 40th order Butterworth filter . The voltage resonant frequency of the squid axon model can vary between 100 Hz at 10°C to 250 Hz at 20°C [75] . Therefore , we selected the input cut-off frequency at 300 Hz that is slightly more than the output 3 dB cut-off frequency encompassing the frequency response expected out of an underdamped second-order response ( see Figure 3 in Guttman et al . [75] ) . The mean and the standard deviations of the stimulus were varied in the range 1–10 µA/cm2 , enabling comparison to earlier work studying channel noise and its effects on information rates [76] . The stimulus was presented for 1 second and each set of simulations consisted of 60 such trials . is an unfiltered broad-band Gaussian white noise with , ( 2 ) where noise variance is computed using ( 3 ) Ω denotes the signal-to-noise ratio ( SNR ) . All Gaussian random numbers were generated using the Marsaglia's Ziggurat algorithm [77]; uniform random numbers were generated using Mersenne Twister algorithm [78] . Deterministic equations were integrated using the Euler-algorithm while stochastic differential equations were integrated using the Euler-Maruyama method , both with a step size of 10 µs . Parameter values are given in Table S1 . Our simulations incorporate Na+ and the K+ voltage-gated ion channels without cooperativity ( Figure S8 ) so that the state transition matrix evolves according to a Markov process [79] , [80] . We track the numbers of channels that were either closed or open [79] using the Gillespie algorithm [81] . The Na+ and the K+ channels had 13 states with 28 possible transitions among these states −20 transitions for the Na+ channels and 8 for the K+ channels . As an example , in time interval δt , the probability that the K+ channel remains in state k is , where γk depicts the sum of all transition rates from state k to any possible successive state . During the interval δt no other ion channel changes its state such that the probability of the ion channels remaining in the same state in the time interval δt is , ( 4 ) is the number of Na+ voltage-gated ion channels in state [i , j] , [nk] is the number of K+ voltage-gated ion channels in state [k] , γij is the total transition rate from state and γk is the total transition rate from state . The transition rate for a particular ion channel state is chosen by drawing a pseudo-random number r1 from a uniform distribution [0 , 1] and defining ttrans as . The Gillespie algorithm then selects which of the 28 possible transitions occur in the time interval ttrans [79] , [81] . The conditional probability of a particular transition j that occurs in the time interval δt is given by , ( 5 ) Here , aj is the product of transition rate associated with transition j and the number of channels in the original state of that transition . The denominator in Eqn . ( 5 ) is equal to λ . The particular transition rate is selected by drawing a random number r2 from the uniform distribution [0 , 1] and fixing ψ as , ( 6 ) The number of voltage-gated ion channels in each state was updated and the membrane potential calculated . An identical algorithm was used for the channel noise in the compartment containing only K+ voltage-gated channels . Both information-theoretic and linear system analysis are a common place in neuroscience [82] , but before providing a detailed exposition for each of these methods , we justify our use of them . The channel capacity for a Gaussian channel [25] , [56] allows us to place an upper bound on the Shannon information encoded in the generator potentials under the assumption of an additive Gaussian noise . On the other hand , the “direct method” [26] is a minimal assumption method to derive an estimate of the reduction in entropy per unit time per spike . Although these two calculations enable us to quantify the information loss separately within each domain ( graded and spiking ) , a more appropriate comparison would employ the same metric permitting direct comparison between domains . The Wiener filter [9] , [56] permits such a comparison , allowing us to test the fidelity of both the analog and the pulsatile signals using identical linear optimal filtering , giving a lower bound on the information present in the response ( e . g . to linearly decode the input stimulus ) . Thus , if inputs were linearly mapped onto outputs then the information rates from “direct method” and the “Wiener filter” analysis would be identical [82] . The lower our reconstruction error the better our generative model of the output is . There are several methods that have been used to quantify information rates in spiking neurons . These include histogram based “direct method” [26] , context-tree Markov Chain Monte Carlo ( MCMC ) [83] , metric space method [84] , binless method [85] , compression entropy [6] , among others . We have used the widely employed “direct method” to measure the entropy of the responses , primarily due to its simplicity and the separation of mutual information into separate terms capturing variability ( spike train entropy ) and reproducibility ( noise entropy ) [26] . The spike train entropy quantifies the variability of the spike train across time . The noise entropy on the other hand , measured the reproducibility of the spike train across trials . These quantities were dependent upon the temporal resolution with which the spikes were sampled , Δt and the size of time window , T . We present a different stimulus current in each subsequent trial ( unfrozen noise ) to calculate the spike train entropy , while using presentations of the same stimulus current in each subsequent trial ( frozen noise ) to calculate the noise correlation . We divided the spike train to form K-letter words ( K = 2 , 4 , 6 , 8 , 12 , 16 , 24 , 32 , 48 or 64 ) , where K = T/Δt . We used the responses from the unfrozen noise session , to estimate the probability of occurrence of particular word , P ( W ) . We estimated the total entropy as , ( 7 ) We estimated the probability distribution of each word at specified time durations , t so as to obtain P ( W|t ) . Entropy estimates were then calculated from these distributions and the average of the distributions at all times were computed to yield the noise entropy as , ( 8 ) 〈〉 indicates average over time . The information was then computed as , ( 9 ) The spike train entropy and the conditional noise entropy diverge in the limit of Δτ→0 , their difference converges to the true finite information rate in this limit [26] . Therefore , we used bias correction methods such that the estimation of entropy was less prone to sampling errors [86] . Using Δt = 1 ms , we varied the spike trains to form words of different lengths . Using these entropy estimates , we extrapolated to infinite word length from four most linear values of the curve of entropy against the inverse of word length . We used an upper-bound method to calculate the maximum information transferable by the nonspiking responses [25] , [56] . This method assumes that the neuronal response and the neuronal noise had independent Gaussian probability distributions in the frequency domain and the noise was additive in nature . In the presence of additive non-Gaussian noise such a method provides us with an upper bound on the channel capacity that is dependent on the entropy power of the non-Gaussian noise distribution [25] , [87] , [88] . We defined the stimulus S as the mean neuronal response obtained from a frozen noise experiment . The noise in each trial was calculated by removing the average response from the individual responses Ri . Owing to Gaussian assumptions , it required enough data to estimate the mean and variance of the Gaussian probabilities . The actual information might be lower than this bound because a Gaussian distribution has the highest entropy for a given variance . In our simulations , both the response and the noise had an approximately Gaussian distribution . We obtained the mean response power spectrum and the noise power spectrum using the multi-taper spectral estimator and computed their ratio to be the signal-to-noise ratio ( SNR ) [29] . This is then used to compute the information for the Gaussian channel as , ( 10 ) For our simulations , the limits of the integral were taken from k1 = 0 Hz to k2 = 300 Hz . The integral was evaluated using trapezoidal rule . We performed stimulus reconstruction to test how noise affects the coherence of a linear system [9] , [56] . The method involved finding a linear temporal filter to minimize the difference between the real and the reconstructed stimulus . We followed Haag and Borst [89] in the derivation of this filter using Gaussian unfrozen noise as the stimulus set . We used 60 trials that consisted of 1 second period of unfrozen white noise si ( t ) to obtain the spike trains ri ( t ) in the form of 1's and 0's with 10 µs resolution . These time domain signals were Fourier-transformed to obtain complex functions Si ( f ) and Ri ( f ) . Two filters were obtained , either by normalizing the cross-power spectral densities ( CPSD ) of the stimulus and the spike response by the stimulus power spectral density ( PSD ) ( forward filter ) or the spike power spectral density ( reverse filter ) as demonstrated below , with angle brackets ( < > ) indicating averages over trials , ( 11 ) ( 12 ) Using the reverse filter , we estimated the stimulus , as the product between Ri ( f ) and Greverse ( f ) , ( 13 ) The quality of the estimate was evaluated by computing a filter between the original stimulus and the reconstructed stimulus; this is simply the coherence function ( γ2 ( f ) ) as shown below , ( 14 ) ( 15 ) The coherence results have been cross-validated using a 65–35 split between the training set and the test set i . e . , we used the first 65% of the trials to calculate the reverse filter and then checked its validity on the next 35% of the trials by computing the final filter ( Gfnal ( f ) ) or the actual coherence ( γ2 ( f ) ) . Reconstruction quality was measured using two metrics . First , normalized root mean squared error ( nRMSE ) between the original stimulus and the reconstructed stimulus was calculated as , ( 16 ) A nRMSE value that tends towards zero represents perfect reconstruction . Second , we calculated a coherence based information rate where a higher value indicates better reconstruction , ( 17 ) Energy consumption in our model is defined as the amount of ATP expended during the encoding of the band-limited stimulus current . The Na+–K+ pump hydrolyses one ATP molecule for three Na+ ions extruded out and two K+ ions imported into the cell [11] . We determined the total K+ current by separating the leak current into a K+ permeable leak current and adding it to the delayed rectifier K+ current . We computed the number of K+ ions by integrating the area under the total K+ current curve for the duration of stimulus presentation . In order to derive the energy consumption we calculated the number of ATP molecules used by multiplying the total K+ charge by NA/ ( 2F ) , where NA is the Avogadro's constant and F is the Faraday's constant .
|
As in electronics , many of the brain's neural circuits convert continuous time signals into a discrete-time binary code . Although some neurons use only graded voltage signals , most convert these signals into discrete-time action potentials . Yet the costs and benefits associated with such a switch in signalling mechanism are largely unexplored . We investigate why the conversion of graded potentials to action potentials is accompanied by substantial information loss and how this changes energy efficiency . Action potentials are generated by a large cohort of noisy Na+ channels . We show that this channel noise and the added non-linearity of Na+ channels destroy input information provided by graded generator potentials . Furthermore , action potentials themselves cause information loss due to their finite widths because the neuron is oblivious to the input that is arriving during an action potential . Consequently , neurons with high firing rates lose a large amount of the information in their inputs . The additional cost incurred by voltage-gated Na+ channels also means that action potentials can encode less information per unit energy , proving metabolically inefficient , and suggesting penalisation of high firing rates in the nervous system .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"neuroscience",
"biology",
"computational",
"neuroscience"
] |
2014
|
Consequences of Converting Graded to Action Potentials upon Neural Information Coding and Energy Efficiency
|
Cognition can reveal itself in the pupil , as latent cognitive processes map onto specific pupil responses . For instance , the pupil dilates when we make decisions and these pupil size fluctuations reflect decision-making computations during and after a choice . Surprisingly little is known , however , about how pupil responses relate to decisions driven by the learned value of stimuli . This understanding is important , as most real-life decisions are guided by the outcomes of earlier choices . The goal of this study was to investigate which cognitive processes the pupil reflects during value-based decision-making . We used a reinforcement learning task to study pupil responses during value-based decisions and subsequent decision evaluations , employing computational modeling to quantitatively describe the underlying cognitive processes . We found that the pupil closely tracks reinforcement learning processes independently across participants and across trials . Prior to choice , the pupil dilated as a function of trial-by-trial fluctuations in value beliefs about the to-be chosen option and predicted an individual’s tendency to exploit high value options . After feedback a biphasic pupil response was observed , the amplitude of which correlated with participants’ learning rates . Furthermore , across trials , early feedback-related dilation scaled with value uncertainty , whereas later constriction scaled with signed reward prediction errors . These findings show that pupil size fluctuations can provide detailed information about the computations underlying value-based decisions and the subsequent updating of value beliefs . As these processes are affected in a host of psychiatric disorders , our results indicate that pupillometry can be used as an accessible tool to non-invasively study the processes underlying ongoing reinforcement learning in the clinic .
There is fast-growing interest to understand how the pupil , as a non-invasive proxy of neuromodulation [1] , relates to cognition . Already since the 1960s , pupil dilation has been associated with the expenditure of cognitive effort [2 , 3] . In more recent years , its relation to decision-making has been investigated extensively . These studies show that the pupil dilates during periods of uncertainty about incoming , task-relevant information [4–7] and after the occurrence of unexpected events that generate surprise [8–11] . Also in a gambling task , where unexpected outcomes were tied to reward , was pupil dilation associated with surprise [12] . However , in that particular study choices did not influence the outcomes of the gambles; making it impossible to learn from the outcomes of choices . In real world encounters , people can learn from the outcomes of their choices and use this information to optimize behaviors or maximize reward . Several studies have shown that pupil dilations track such reward learning processes . During classical or Pavlovian learning [13] , the pupil dilates in response to cues that predict reward [14–16] and tracks changes in reward expectations [17] . In situations that require decisions to obtain reward , baseline pupil diameter prior to a choice [18] , as well as task-evoked dilations [19] predicted whether a choice would be either exploratory or exploitatory , hence , predicting the sensitivity to choose the option with the highest expected reward . While these findings suggest that the pupil provides a promising marker of several processes and states associated with learning from reward , it remains unclear how the pupil relates to the underlying process of reinforcement learning ( RL ) . Understanding of this relationship is important , as it could open up the possibility to continuously monitor the underlying cognitive processes that shape learning and decision-making based on reward . This knowledge would greatly increase the clinical impact of pupil size recordings , which already has shown some promising results in studies involving Parkinson’s patients [20 , 21] . However , it is unclear how these processes interact with other cognitive processes such as attention , cognitive effort , and uncertainty . Here , we investigated pupil size fluctuations during value-based learning and decision-making , using a computational RL model to identify the specific influences of value-related computations on pupil size . We measured pupil size while thirty-four participants performed a probabilistic RL task , consisting of a separate learning and transfer phase . ( Fig 1A and 1B and Methods ) [22] . In the learning phase , the reliability of choice outcomes varied across three learning pairs with different reward probabilities ( AB , 80:20; CD , 70:30; EF , 60:40 ) . As participants gradually learned to choose the best option in each pair , these different reward probabilities created varying degrees of choice difficulty , uncertainty and value expectations across choices . In a subsequent transfer phase , participants were then presented with novel stimulus pair combinations ( i . e . , AB , CD , EF , AC , AD , AE , AF , BC , BD , BE , BF , CE , CF , DE , and DF ) and asked to select the most rewarding option based on previous learning . As no choice feedback was provided in the transfer phase , it allowed us to measure how choices were guided by previously acquired reinforcement values , and how this information generalized to entirely new choice situations . We fitted a hierarchical Bayesian version of the Q-learning RL algorithm [23] to participants’ choices in the learning phase to describe value-based choices and outcome evaluations ( Fig 1C and Methods ) [24–27] . Bayesian hierarchical parameter estimation results in more precise and stable parameter inference compared to procedures using individual-level maximum likelihood [26 , 28–30] , and is therefore a preferred modelling approach . The Q-learning algorithm describes value-based decision-making using two functions: a choice function and an outcome function . The choice function calculates the probability of choosing one option ( Q-chosen ) over the other ( Q-unchosen ) , based on one’s sensitivity to value differences , or explore-exploit tendency ( β; Fig 1D , left panel ) . The outcome function then computes the magnitude by which the reward prediction error ( RPE ) changes value beliefs about the chosen option , scaled by the learning rate ( α; Fig 1D , right panel ) [31] . As value beliefs are differently updated after positive and negative outcomes [32–34] via different striatal learning mechanisms [35–37] , we defined separate learning rate parameters for positive ( αGain ) and negative ( αLoss ) choice outcomes [24 , 32 , 33 , 38] . Our computational approach allowed us to investigate the potential utility of the pupil as a proxy for value-based decision-making and value belief updating , across two levels . First , we describe participants’ choice behavior using parameters that embody core computational RL principles . These parameters provide a strong handle to investigate how inter-individual differences in value-based learning and decision-making relate to pupil responses . Second , by simulating the learning process we could investigate how pupil size depended on trial-to-trial fluctuations in underlying computational variables such as value beliefs , uncertainty and reward prediction errors . That is , our experimental paradigm allowed us to map pupil responses onto separable computational components both across participants and trials , using linear systems analysis techniques [39 , 40] .
Participants learned the stimulus-reward contingencies well , as they correctly learned to select the higher reward probability option in all three pairs ( P ( correct ) above chance , all Ps < . 001; Fig 2A ) . Performance was best in the most reliable choice pair ( AB ) and decreased progressively as the feedback reliability of choice pairs decreased from CD to EF: smaller differences in the reward probability ratios increased the number of errors ( F ( 2 , 66 ) = 14 . 45 , P< . 001 , η p 2 = . 19 ) and response times ( F ( 2 , 66 ) = 5 . 5 , P = . 006 , η p 2 = . 04 ) . In the transfer phase , choices were guided by the previously learned reward probabilities . Here , participants made more errors ( F ( 2 , 66 ) = 49 . 3 , P< . 001 , η p 2 = . 53 ) and were slower ( F ( 2 , 66 ) = 34 . 6 , P< . 001 , η p 2 = . 12 ) when confronted with option pairs with small value differences ( Fig 2B ) , consistent with earlier studies [32 , 41 , 42] . The Q-learning model simulated participants’ choice behavior well ( Fig 2C ) when using the fitted learning rates ( αGain , αLoss ) and explore-exploit ( β ) parameter ( Fig 2D ) . In accordance with behavior , the estimated value beliefs were highest for A and lowest for B ( Fig 2E ) with differences in value beliefs being largest for AB , followed by the CD and EF pair ( F ( 2 , 66 ) = 20 . 63 , P< . 001 , η p 2 = . 39 ) . We next investigated whether the pupil was sensitive to the cognitive processes supporting value-based decisions . To do so , we first characterized the average pupil response pattern across subjects epoched around two separate moments in the trial: leading up to , and immediately after the moment of choice and around the moment of feedback . Around the moment of a choice , a biphasic pupil response was observed that was characterized by dilation starting ≈1s . prior to the moment of the behavioral report ( Fig 3A ) . This upwards response reflected the unfolding decision process [5 , 43] and was followed by late pupil constriction ( ≈1s . post-choice ) . After receiving choice feedback , again a biphasic pupil response was observed that was characterized by early dilation ( ≈1s . post-event ) and late constriction ( ≈2s . post-event; Fig 3B ) . Across individuals , the observed choice- and feedback-evoked pupil responses corresponded differentially to the underlying processes driving value-based decision-making . As shown in Fig 3C , left panel , pupil dilation at the moment of a choice was uniquely predicted by an individual’s sensitivity to value differences , or explore-exploit tendency ( β; permutation test , P = . 006; S1A Fig ) , indicating that a greater tendency to exploit high value options ( high β ) related to a stronger dilatory response ( Fig 3C , right panel ) . Feedback-related dilation and constriction correlated inversely with an individual’s positive , but not negative , learning rate ( S1B Fig ) , suggesting that this parameter selectively scaled the amplitude of the feedback-evoked pupil response . Indeed , as shown in Fig 3D , left panel , the feedback-evoked response amplitude was uniquely predicted by an individual’s positive learning rate ( αGain; permutation test , P = . 017 ) , indicating that slower updating of value beliefs after positive feedback predicted a stronger feedback-evoked response ( low αGain; Fig 3D , right panel and S1D Fig ) . In sum , pupil responses evoked by choice and feedback differentially predicted the underlying processes supporting value-based decisions in the learning phase . The tendency to exploit high value options ( β ) predicted stronger pupil dilation leading up to a value-driven choice , whereas less updating of value beliefs after positive feedback ( αGain ) predicted an amplified feedback-related response . These relations are consistent with the tenets of the Q-learning model , in which the explore-exploit parameter determines the outcome of a value-driven choice and learning rates affect how much value beliefs are updated after receiving choice feedback . We observed that across-subject variability in pupil responses was explained by model parameters that describe the underlying processes driving value-based decision-making . But do pupil responses also reflect the ongoing reinforcement learning process during value learning ? In a next step , we investigated the extent to which trial-to-trial fluctuations in variables describing ongoing value-based decision-making were reflected in pupil responses . In the learning phase , prior to reaching a value-driven choice , pupil dilation correlated positively with the value difference between options ( cluster P< . 001 , 2 . 0s . pre-event until -0 . 07s . pre-event , Fig 4A , upper panel ) , indicating that larger value differences elicited larger pupil dilation before the choice . Specifically , the pupil dilated as a function of trial-by-trial value beliefs of the chosen , but not the unchosen option ( paired t-test , t ( 33 ) = 6 . 98 , P< . 001; Fig 4B , upper panel ) , revealing that pupil dilation uniquely reflected the value belief determining the upcoming choice . To rule out the possibility that condition differences ( i . e . AB , CD , EF ) instead of trial-by-trial fluctuations in chosen value beliefs explained pupil dilation prior to a choice , we estimated their independent effects on pupil size in a single regression analysis . We observed no differences between conditions in average pupil dilation prior to a choice ( Fig 4A , lower panel ) . This also excluded the hypothesis that pre-choice pupil dilation was driven by uncertainty [5 , 44] or cognitive effort [2 , 3] , as we did not observe significantly more dilation in the most uncertain , hence , most effortful EF pair . In all pairs , pre-choice pupil size correlated positively with chosen value ( F ( 2 , 66 ) = 19 . 76 , P< . 001 , η p 2 = . 15; Fig 4B , lower panel ) irrespective of condition type ( F ( 2 , 66 ) = 1 . 8 , P = . 17 ) . Thus , prior to reaching a value-driven choice , the pupil tracked subtle differences in value beliefs about the upcoming choice , while dilation did not reflect uncertainty or cognitive effort driven by condition differences . Next , we asked whether value beliefs also modulated pre-choice pupil dilation in the subsequent transfer phase , where choices were based on previously acquired reinforcement values . In contrast to the learning phase , pupil dilation prior to a value-driven choice was not predicted by previously learned value differences between options ( Fig 4C , upper panel ) . Indeed , a repeated measures ANOVA with the factors phase ( learning , transfer ) and value ( chosen , unchosen ) indicated that only during learning , but not during transfer , pre-choice pupil dilation was modulated by value beliefs about the upcoming choice ( F ( 1 , 33 ) = 6 . 9 , P = . 013 , η p 2 = . 06 ) . This interaction effect was not explained by differences in tonic pupil size fluctuations between the experimental phases ( S2 Fig ) , which can impact the magnitude of phasic pupil responses [1 , 40] . Additionally , we investigated the observed interaction using a Bayesian repeated measured ANOVA . Compared to the null model , the model that incorporated both main factors and their interaction received most support from the data as indicated by BF = 55 . 4 . This is regarded very strong evidence in favour of the alternative model [45] . This model also convincingly received most support from the data compared to all candidate models , as indicated by BF = 16 . 9 . Lastly , post-hoc test that directly compared chosen and unchosen value beliefs showed that they did not differently modulate pre-choice pupil size in the transfer phase ( paired t-test , <1 , Fig 4B , upper panel ) . Nor did either variable’s mean correlation with pre-choice pupil dilation differ from zero ( t<1 for chosen and unchosen value ) . Together , these findings provide compelling evidence that chosen and unchosen value beliefs differently modulate pupil dilation during decision formation in the learning compared to the transfer phase . However , immediately after a value-based choice , learned value beliefs negatively predicted pupil dilation both in the learning ( cluster P = . 007 , 0 . 68s . pre-event until 1 . 5s . post-event; Fig 4A , upper panel ) and transfer phase ( cluster P = . 003 , -0 . 02s . until 1 . 48s . post-event; Fig 4C , upper panel ) . Now smaller , instead of larger , value differences elicited larger post-choice pupil dilation , suggesting that the difficulty of a recent choice , or the choice conflict it generated , drove pupil size upward . Indeed , we observed a similar post-choice pupil response pattern when regressing choice conflict on the basis of the experimental reward probabilities on pupil size ( Fig 4C , lower panel ) , indicating that post-choice pupil dilation was modulated by choice conflict , consistent with an earlier report [42] . These model-based trial-to-trial analyses show that when engaged in active reinforcement learning , pupil dilations differentially reflect value beliefs and choice conflict at different points in time . Prior to value-based choices , pupil size uniquely reflected value beliefs about the upcoming choice , where stronger dilations predicted higher value beliefs . This pattern of pre-choice value dilations was absent in the subsequent transfer phase where rewards could not be obtained , indicating that apparently similar pupil dilations prior to value-based choices can index different cognitive processes during and after reinforcement learning . Only during active reinforcement learning , we observed that choice-related pupil dilation reflected value beliefs about the upcoming choice . If the pupil reliably tracked the ongoing reinforcement learning process , it should also provide information about the evaluation of a recent choice outcome . In the last step , we therefore investigated how feedback-related pupil responses covaried with the degree to which outcomes violated value beliefs about a recent choice . We observed larger feedback-related pupil dilation ( Fig 5A ) after choices between options with small value differences . Specifically , early post-feedback dilation correlated negatively with differences in value beliefs of recently presented options ( cluster P< . 001 , -1 . 5s . pre-event until 1 . 78s . post-event; Fig 5B ) . We furthermore verified that these feedback-related dilations were not driven by feedback valence ( S3A and S3B Fig ) . In contrast to dilation in the choice interval , dilation in the feedback interval was explained by fluctuations in trial-by-trial value beliefs of both the chosen and the unchosen options , in opposite directions ( Fig 5C ) . Thus , lower beliefs about the chosen and higher beliefs about the alternative option both increased dilation , indicating that uncertainty about the value of a recent choice modulated feedback-related dilation . In support of this , trial-by-trial chosen and unchosen value beliefs explained feedback-related dilation already prior to receiving feedback , which suggested that uncertainty about the outcome of a value-based decision drove pupil size . Lastly , outcomes that violated value beliefs did not elicit larger feedback-related dilations ( S3 Fig ) , excluding the hypothesis that these modulations of the feedback response reflected surprise . Importantly , whereas value beliefs about a recent choice affected early dilation , the degree to which outcomes violated those beliefs modulated late feedback-related pupil constriction . As shown in Fig 5B , signed RPEs correlated positively with late feedback-related pupil constriction ≈2s . after receiving feedback ( cluster P< . 001 , 1 . 8s . until 3 . 0s . post-event ) . This correlation indicated that worse-than-expected outcomes ( -RPEs ) elicited stronger pupil constriction compared to better-than-expected outcomes ( +RPEs ) . To summarize , we observed a biphasic feedback-related pupil response that tracked the evaluation of a recent value-based choice . Early pupil dilation was modulated by uncertainty about the value of options , as choices between similarly valued options increased dilation the most . Late pupil constriction was modulated by the violation of current value beliefs , as worse-than-expected outcomes elicited stronger pupil constriction compared to better-than-expected ones .
The present results provide the novel insight that the pupil reliably tracks the underlying cognitive processes of learning and decision-making based on reward . When engaged in active reinforcement learning , but not when choice value was already internalized , the pupil showed two distinct response patterns . Prior to reaching a value-driven choice , pupil dilations scaled with trial-by-trial value beliefs about the upcoming choice and were diagnostic for an individual’s sensitivity to choose the option with the highest expected outcome . Feedback about the choice subsequently evoked a biphasic evaluation response . Early pupil dilation scaled with uncertainty about the value of recent choice options , whereas subsequent pupil constriction scaled with the violation of current choice value beliefs , or signed reward prediction errors . Earlier studies have shown that pupil dilations can reflect variables or states related to reward learning [14–16 , 20 , 21] . Our results extend these findings by showing that pupil responses reliably track choice value computations during both decision formation and decision evaluation . The specificity of our findings outlines how the pupil can be used to index the ongoing reinforcement learning process . These results could greatly increase the clinical impact of pupil size recordings , as our findings suggest that pupil responses can be used to monitor the ( affected ) ongoing reinforcement learning process . Specifically , single-trial fluctuations in pupil dilation prior to choice signaled the value of the to-be-chosen option , but not the alternative one . This indicates that during decision formation , pupil dilation specifically reflected the value that was driving the choice . Could these value-driven dilations reflect the effects of cognitive effort [2 , 3] , or uncertainty [4 , 5 , 8] that are known to affect pupil size ? While effects of cognitive effort are typically studied with tasks in the domain of cognitive control ( e . g . [46–49] ) , it is generally found that pupil dilation increases with increasing task demands . Presumably , dilations reflect the effort exerted in reaction to difficult or demanding situations [50] . In our study , this would translate to the following hypothesis: the most difficult choice condition ( i . e . the most unreliably reinforced stimulus pair EF ) should elicit greater pupil dilation . Our findings do not agree with this hypothesis , as choices in the difficult EF pair were not preceded by stronger pupil dilation . Neither can our findings be explained by effects of uncertainty about a value-based choice , as higher value beliefs , indicating more certainty about choice value , elicited greater dilation . What our findings indicate is that higher value beliefs about the upcoming choice led to stronger reward expectations [20 , 21] , or lower risk assessment [12] that increased pupil dilation prior to a value-based choice . Furthermore , an individual’s sensitivity to value differences between presented options , as quantified by the β parameter of our model , predicted the amount of pupil dilation exactly at the moment of a value-based choice . Individuals that were sensitive to small value differences showed stronger choice-locked pupil dilation and made more exploitatory choices , which also led to better task performance [25] . Optimal task performance [51 , 52] as well as the tendency to exploit in dynamic environments [19 , 53] have previously been associated with increased choice-related pupil dilation . The observed relationship could therefore reflect either one of these processes , as choosing a high value option can result from accurate option value representations , or from the general tendency to favor exploitation over exploration [54] . Future studies that measure pupil size during value-based decision-making in a reversal learning paradigm may be able to disentangle these two alternative explanations , as optimal task performance would then depend on changing decision strategies over time . After receiving feedback , a biphasic feedback-related pupil response tracked two different evaluation processes associated with the outcome of a choice . First , early dilation was scaled by the uncertainty associated with the outcome of a value-based choice . This was further evidenced by the observation that choices between closely valued options triggered uncertainty-related pupil dilation already prior to the moment of feedback . This suggests pupil dilation reflected value uncertainty when participants anticipated the outcome of a choice . While our study is the first to relate feedback-related pupil dilation directly to uncertainty about internal choice value beliefs , these findings are consistent with studies that relate pupil dilation to perceptual decision uncertainty , driven by observer’s internal noise [6 , 7] . Second , late pupil constriction was explained by signed reward prediction errors , reflecting how much an outcome violated current value beliefs about the chosen option . Lower-than-expected choice outcomes resulted in stronger pupil constriction compared to higher-than-expected ones , a response pattern consistent with value coding [55] . As the reward prediction error term multiplied with the learning rate updates the value of the chosen stimulus , reward prediction error responses in the pupil after feedback are consistent with our finding that chosen value uniquely modulated pupil dilation prior to a choice . We can only speculate about the similarity to the reward prediction error firing pattern of phasic dopamine neurons [56–59] that briefly activate after higher-than-expected outcomes and deactivate after lower-than-expected outcomes . Alternatively , the observed correlation between late feedback-related pupil constriction and signed reward prediction errors could be driven by differences in saliency between unexpected positive and negative outcomes . It has been shown that contrast-based stimulus saliency modulates the magnitude of transient pupil responses , with more salient stimuli evoking a larger peak-to-peak ( i . e . larger dilation and constriction ) pupil response [60 , 61] . While we controlled for the physical stimulus saliency in our experiment , some experimental events could have been more salient than others . Due to reinforcement learning , unexpected negative outcomes occurred less often than unexpected positive outcomes , which might have rendered them subjectively more salient events . Thus , increased subjective saliency could explain the stronger feedback-related pupil constriction observed after unexpected negative events . We found that pupil responses systematically tracked key components of reinforcement learning , however , important differences were observed with a later recall phase . Only during learning , but not during transfer , was pupil dilation prior to choice modulated by choice value; a difference that may indicate different underlying cognitive processes that drive value-based choices [62 , 63] . Why was this the case , in a task where participants had to make value-based decisions in both experimental phases ? One important difference between the learning and transfer phase was the presentation of choice feedback , thus , the ability to learn from choice outcomes . In the learning phase , only three stimulus pairs were presented and each choice was followed by feedback to allow learning . In the transfer phase , our objective was to assess already internalized choice value by confronting participants with novel stimulus pair combinations of previously acquired reinforcement values . Hence , in this phase , choices were not followed by feedback , and choice value representations could not be changed . Dopamine , particularly in the striatum [64] , plays an important role during reinforcement learning [65] . Striatal dopamine strengthens actions that lead to rewarding outcomes and weakens those that lead to aversive ones [22 , 32 , 36] . It thereby flexibly adapts behavior to maximize future reward . In the transfer phase , value beliefs are consolidated and dopamine no longer plays an important role in learning or modulating choice behavior . Information used to make a value-based choice can be retrieved from memory , guided by structures that encode learned value representations , such as the ventromedial prefrontal cortex [63 , 66] . Our finding that pupil dilation signaled upcoming choice value only during learning , could mean that the pupil is particularly sensitive to contingency learning [67] , thus , to learning the mapping between actions and outcomes . Whether , and in what way , dopamine modulates the pupil response has yet to be determined , but several lines of research show promising results [7 , 20 , 21 , 68 , 69] . It is likely that such modulations occur via interactions with the noradrenergic locus coeruleus ( LC ) , a brainstem nucleus that is often linked to pupil dilation in micro-stimulation and decision-making studies [68 , 70–73] . As these studies also found that activity in other brain areas correlated with pupil responses [60 , 61 , 70 , 71] , including the dopaminergic ventral tegmental area [68] , this further suggests multiple ( interacting ) sources driving pupil size . In support of interactions between the noradrenergic and dopaminergic system in modulating pupil size , the LC and dopaminergic midbrain structures have dense reciprocal connections and receive a common top-down projection from the prefrontal cortex [74] . Moreover , both systems play important , yet different , roles in reward-learning and motivational behavior [73 , 75] , which suggest they both might play important roles in modulating the pupil during reinforcement learning . While we observed the pupil to systematically track key components of reinforcement learning , could these results alternatively be explained by effects of attention ? The pupil has long been used as an index of attention , as the amount of attention paid to a stimulus determines the amount of transient pupil dilation [76–79] . In our study , when choices where made , attention was most likely shifted toward the chosen option , which could have driven phasic pupil dilation during choice . While this explanation fits our finding in the learning phase , where pre-choice pupil dilation scaled with the value of the to-be chosen option , it does not fit with the absence of this relation in the transfer phase , even though participants paid attention to the chosen options , as they performed the task well . Neither does attention -and its relation to phasic pupil dilation- explain why feedback-related pupil constriction in the learning phase scaled with how much an outcome violated choice value beliefs . Thus , while attention likely plays an important role in value-based decision-making [42 , 80–82] , the observed patterns of results here cannot be solely explained by effects of attention . To conclude , our study provides evidence that the pupil is a reliable indicator of value-based decision-making during active reinforcement learning . Pupil responses signaled the processing of value up to a choice and the subsequent evaluation of choice outcomes in terms of uncertainty and violations of value beliefs . There were several aspects to our approach that enabled us to establish these specific relations and to move beyond previous work linking the pupil to reward . First , our computational approach enabled us to characterize the full temporal profile of value-based decisions in the pupil , thereby relating different decision and evaluation processes to different components of the pupil response . These relationships could only be obtained using a ridge regression deconvolution approach that enabled us to disentangle the multiple underlying cognitive processes that impacted pupil size within a single regression analysis . Second , these specific relations could only be established with the use of classical learning theories that provided us access to participants’ developing value beliefs , and the underlying choice considerations thought to support value-driven decisions . This also highlights our use of internal , or subjective , value estimates to relate to the pupil . This contrasts previous studies that used externally derived value estimates to investigate reward-related effects on pupil size [8 , 12 , 20 , 21 , 69] . Lastly , our study describes the temporal evolution of reinforcement learning in the pupil , thereby providing evidence that the pupil can be used to non-invasively track the reinforcement learning process as it takes place . Future studies that combine functional brain imaging and pupillometry will have to further specify the brain areas that contribute to the value-based pupil response .
The Ethics Committee of the Vrije Universiteit Amsterdam approved this study . Forty-two healthy participants with normal to corrected to normal vision completed the experiment ( 10 males; mean age = 24 . 9; age range = 18-34 years ) . They were paid 16€ for 2 hours of participation and earned an additional performance bonus ( mean = 10 . 2€ , SD = 1 . 8 ) . The ethical committee of the Vrije Universiteit Amsterdam approved the study and written informed consent was obtained from all participants . Eight participants were excluded from analyses due to the following reasons: inadequate fixation to the center of the screen ( N = 4 ) , reporting more than three unique stimulus pairs in the learning phase ( N = 1 ) and ( almost ) perfect choice accuracy in the learning phase , which complicated behavioral model fitting ( N = 3 ) , resulting in a total of 34 participants for the analyses . Participants were seated in a dimly lit , silent room with their head positioned on a chin rest , 60 centimeters away from the computer screen . They received written information about the general purpose of the experiment , after which they completed a 30-trial practice session of the learning phase . Subsequently , participants completed for the learning phase 6 runs of 60 trials each ( 360 trials in total , 120 presentations of each stimulus pair ) , with small breaks in-between runs . After each run , the earned number of points was displayed . At the end of the learning phase , the total number of earned points was converted into a monetary bonus . Directly after the learning phase , participants entered the transfer phase . They completed 5 runs of 60 trials each ( 300 trials in total , 20 presentations per stimulus pair ) , with small breaks in-between runs . Overall choice accuracy was displayed at the end of the transfer phase . Stimuli were presented on a 21-inch Iiyama Vision Master 505 MS103DT with a spatial resolution of 1024 x 768 pixels , at a refresh rate of 120Hz , with mean luminance 60 cd/m2 . Experiments were programmed in OpenSesame and data analysis were performed using custom software written in Python , using Numpy ( v1 . 11 . 2 ) , Scipy ( v0 . 18 . 1 ) , FIRDeconvolution ( v0 . 1 . dev1 ) , Hedfpy ( v0 . 0 . dev1 ) , MNE ( v0 . 14 ) and PyStan ( v2 . 14 ) packages . Luminance effects on pupil size were minimized by keeping the background luminance of the display constant . Color stimuli were near-isoluminant to each other and the background ( set via a flicker-fusion color calibration test carried out once at the start of the experiment ) . To account for further luminance bias effects , each participant had a unique color pair ( red-blue; yellow-dark blue; green-magenta ) to reward probability mapping ( AB , CD , EF ) that was counterbalanced in order ( e . g . red-blue or blue-red for AB ) . In each learning phase trial , participants continuously fixated on a central white fixation dot . After 500ms ( SD = 200ms ) , two colored stimuli ( 1 . 26°x1 . 26° visual angle ) appeared at the horizontal meridian left and right from the central fixation dot at a distance of 5 . 04° visual angle . Participants made a choice for one of the options using the ‘K’ ( left choice ) and ‘L’ ( right choice ) keys . A choice was highlighted by a small dark gray arrow ( 150ms ) pointing in the direction of the chosen option . After a random interval drawn from a Gaussian distribution with a mean of 1500ms ( SD = 300ms ) , the choice was followed by auditory feedback , indicating reward ( +0 . 1 points; 500ms “correct” sound ) or no reward ( 500ms; pure sine tone at 300Hz ) . Omissions or response times ( RTs ) longer than 3500ms were followed by a neutral tone ( 500ms; pure sine tone at 660Hz ) . Inter-trial intervals were drawn from a Gaussian distribution with a mean of 3000ms ( SD = 300ms ) Trials of the transfer phase followed the same trial structure as trials in the learning phase , but had a shorter duration as choices were not followed by feedback . Choices and RTs were recorded for all trials in the learning and transfer phase . RT on every trial was computed as the time from onset of the stimulus pair until the choice ( key press ) . Trials with RTs below 150ms or above the RT deadline of 3500ms were removed from all analyses . As a choice between two options in the learning phase was never necessarily “correct” , we defined the selection of the optimal option ( more reinforcing option of the presented pair ) as a correct choice . For the transfer phase , value conflict on a particular trial was defined on the basis of the experimental reinforcement value difference between the presented stimuli , where smaller value differences were associated with higher conflict . Choices during the learning phase were fit with a reinforcement learning ( “Q-learning” ) model [23 , 83] . The Q-learning model has an extensive theoretical background in which decision-making is explicitly evaluated [84 , 85] and has been successfully applied in a range of domains . Examples of this are genetics [32 , 86] , clinical settings [54 , 63 , 87–89] , attention [24] , decision bias [34] and risk [38] . For each option , the model estimates its expected value , or “Q-value” , on the basis of individual sequences of choices and outcomes . All Q-values were set to 0 . 5 before learning . After each choice , the chosen option’s Q-value is updated by learning from feedback that resulted in an unexpected outcome , which is captured by the RPE , ri ( t ) − Qi ( t ) . Thus , the Q-value for option i on the next trial t is updated depending on the outcome , r , using the following formula: Q i ( t + 1 ) = Q i ( t ) + { α G a i n [ r i ( t ) − Q i ( t ) ] if r =1 α L o s s [ r i ( t ) − Q i ( t ) ] if r =0 ( 1 ) where parameters 0 ≤ αGain , αLoss ≤ 1 represent positive and negative learning rates , respectively , that determine the magnitude by which value beliefs are updated depending on the RPE . We modeled separate learning rates , as different striatal subpopulations are involved in positive and negative feedback learning [35–37 , 90] and individuals tend to learn more from positive feedback [24 , 25 , 34] . Modeling two learning rates was further validated by comparing this hierarchical Q-learning model to a hierarchical Q-learning model with only one learning rate . Model selection was based on individual AIC and BIC values and supported the use of two learning rates , as indicated by lower AIC and BIC values ( mean AIC 1α = 234 , mean AIC 2α’s = 218; mean BIC 1α = 242 , mean BIC 2α’s = 230 ) . Given the Q-values , the probability of selecting one option over the other ( e . g . selecting option A over B ) was described by a softmax choice rule: P A ( t ) = e x p ( β · Q A ( t ) ) e x p ( β · Q B ( t ) ) + e x p ( β · Q A ( t ) ) ( 2 ) Here , 0 ≤ β ≤ 100 , or the explore-exploit parameter , described the sensitivity to option value differences , where larger β values indicates greater sensitivity , and more exploitative choices , for options with relative higher reward values . The Q-learning model was fit using a Bayesian hierarchical fitting procedure , where individual parameter estimates were drawn from group-level parameter distributions that constrained the range of possible individual parameter estimates . This procedure allowed for the simultaneous estimation of group-level and individual-level parameters [26 , 91] , thereby capitalizing on the statistical strength offered by the degree to which participants are similar with respect to the model parameters as well as taking into account individual differences [92] . As shown in Fig 1C , our model was implemented following [24 , 25] . Variables ri ( t-1 ) ( outcome for participant i on trial t-1 ) and chi ( t ) ( choice of participant i on trial t ) were obtained from the behavioural data . Per-participant parameter estimates αGi ( αGain participant i ) , αLi ( αLoss participant i ) and βi ( β participant i ) were modeled using a probit transformation z’i ( α G i ′ , α L i ′ , β i ′ ) . The probit transformation is the inverse cumulative distribution function of the normal distribution that can be used to specify a binary response model . z’i were drawn from group-level normal distributions with mean μz′ and standard deviation δz′ . A normal prior was assigned to group-level means μz’∼N ( 1 , 0 ) and a uniform prior to the group-level standard deviations δz’∼U ( 1 , 1 . 5 ) [26] . The Bayesian hierarchical model was implemented in STAN [93] and fit to all trials of the learning phase that fell within the correct response time window 150ms ≤ RT ≤ 3500ms , ( mean = 99 . 5% of trials , SD = 0 . 8% ) . Multiple chains were generated to ensure convergence , which was evaluated by the Rhat statistic [94] . The Rhat statistic confirmed convergence of the fitting procedure ( i . e . , all Rhats were equal to 1 . 0 ) . We also tested whether the derived per-participant parameters could simulate choices that were qualitatively similar to the observed choices originally used for fitting . Here , choices were simulated 100 times for each participant using the mode of the derived per-participant parameter distribution . Simulated choice accuracy was averaged across simulations and evaluated against the observed choice data ( Fig 2C ) . The modes of the per-participant posterior parameter distributions were selected to describe individual positive and negative learning rates ( αGain , αLoss ) and relative reward sensitivity ( β ) . In the learning phase , these per-participant parameter estimates were used to quantify Q-values and RPEs on each trial . Specifically , we quantified for each trial the value of the option that was chosen and the alternative unchosen option . In the transfer phase , when participants did not receive feedback about their choices , we investigated how previously learned value related to pupil responses during value-based decisions . To do so , we selected the final Q-value estimates for each option ( i . e . at the end of the learning phase ) and used these values to quantify for each trial the value of the chosen and unchosen stimulus , given the individual sequences of choices . All obtained single-trial variables were used as covariate regressors in a deconvolution analysis ( described below ) , to investigate how they dynamically varied with trial-by-trial fluctuations in transient pupil responses in the learning and transfer phase . The diameter of the pupil was recorded at a 1000Hz using an EyeLink 1000 Tower Mount ( SR Research ) . The eye-tracker was calibrated prior to each run . Blinks and saccades were detected using standard EyeLink software with default settings and Hedfpy , a Python package for preprocessing eye-tracking data . Periods of data loss during blinks were removed by linear interpolation , using an interpolation time window of 200ms before until 200ms after a blink . Blinks not identified by the manufacturer’s software were removed by linear interpolation around peaks in the rate of change of pupil size , using the same interpolation time window . The interpolated pupil signal was band-pass filtered between 0 . 05Hz and 4Hz , using third-order Butterworth filters , z-scored per run , and resampled to 20Hz . As blinks and saccades have strong and relatively long-lasting effects on transient pupil size [40 , 95] , these influences were removed from the data , as follows . Blink and saccade regressors were created by convolving all blink and saccade events with their standard Impulse Response Function ( IRF ) [40 , 96 , 97] . These convolved regressors were used to estimate their responses in a General Linear Model ( GLM ) , after which we used the residuals of this GLM for further analysis . For the subsequent deconvolution analysis , trials were removed in which participants made a saccade towards either of the two presented colored stimuli ( i . e . saccades exceeding 3 . 3° visual angle away from fixation ) to ensure that pupil responses were not affected by eye movements ( percentage removed trials , mean = 4 . 8%; SD = 4 . 5%; range = 0 . 0%-16 . 3% ) . We implemented the deconvolution analysis using cross-validated ridge regression , which allows one to find the general solution to a least-squares problem that would be unstable due to multicollinearity of regressors [98] . Ridge regression penalizes , or shrinks , regression coefficient weights towards zero to reduce the estimation variance on the coefficients: β ^ r i d g e = ( X T X + λ I ) − 1 X T y ( 3 ) Here , y is the pupil time series signal and X is the design matrix consisting of a set of vectors that contain ones at all sample times relative to the event timings of which we estimated the pupil response , and zeros elsewhere . The identity matrix , I , is multiplied by λ ≥ 0 , a tuning parameter that controls the strength of the penalty term . If λ = 0 , the linear regression solution is obtained , λ = ∞ , β ^ r i d g e = 0 . To obtain for each participant the optimal λ value , we applied cross validation on the pupil time series data . Here , the pupil data was divided into a training and test set . A weight matrix was obtained for each λ value ( range = 0 ≤ λ ≤ 1 ) , using the training set , and was used to predict the test set . This process was repeated for 20 different selections of training and test sets , and the best λ value was selected based on its prediction accuracy . The resulting regression , β ^ r i d g e , contained the deconvolved pupil responses of all separate event types . Nonparametric cluster-based permutation t-tests [99–101] were used to test for significant regression coefficients and to correct for multiple comparisons over time . Briefly , for each time point of a time series signal , t-tests were performed on each set of across-subject coefficient values . The cluster size was determined by the number of contiguous timepoints for which the t-test resulted in P< . 05 . The observed cluster size was then compared to a random permutation distribution of maximal cluster sizes: the proportion of random clusters resulting in a larger size than the observed one determined the P-value , corrected for multiple comparisons . To assess the effects of chosen and unchosen value covariates on pupil size across the decision interval , we summed each regressor’s coefficient values locked to the start ( option onset ) and locked to the end of the decision interval ( the moment of choice ) , while discarding their post-choice effects . We normalized the summed regressor coefficient values by the number of samples they explained of the pupil time series signal . The resulting averaged , normalized regressor coefficient values were used in a repeated measures ANOVA to test for main and interaction effects on pupil size , both for the learning and transfer phase . Across-subject analyses of the relation between pupil responses and computational model parameters were calculated using bootstraps [102] . We randomly drew with replacement 10 , 000 new pupil size—model parameter estimate pairs which were used in the across-subject GLM . From the resulting bootstrapped regression coefficients , 68% confidence intervals were calculated using a percentile approach . P-values calculations were based on a two-sided hypothesis test , with the P-value being the fraction of the bootstrap distribution that fell below ( or above ) 0 .
|
It has long been known that the pupil dilates when we decide . These pupil dilations have predominantly been linked to arousal . However , reward-related processes may trigger pupil dilations as well , as dilations have been linked to activity in the dopaminergic midbrain , a region important for reward processing and reinforcement learning . Using a learning task and a computational model to quantitatively describe the cognitive processes that drive reinforcement learning behavior , we show that the pupil closely tracks different aspects of the reinforcement learning process . Prior to making a value-based choice , pupil dilation reflected the value of the soon-to-be-chosen option . After receiving choice feedback , early dilation reflected uncertainty about the value of recent choice options , while late constriction reflected how strongly an outcome violated current value beliefs . These findings provide the novel insight that the pupil can be used to track value-based decision-making , opening up a new method for online tracking of reinforcement learning processes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"learning",
"medicine",
"and",
"health",
"sciences",
"neurochemistry",
"chemical",
"compounds",
"decision",
"making",
"social",
"sciences",
"neuroscience",
"learning",
"and",
"memory",
"organic",
"compounds",
"hormones",
"cognitive",
"psychology",
"mathematics",
"cognition",
"amines",
"neurotransmitters",
"catecholamines",
"discrete",
"mathematics",
"combinatorics",
"dopamine",
"sensory",
"physiology",
"human",
"learning",
"chemistry",
"visual",
"system",
"biochemistry",
"psychology",
"eye",
"movements",
"permutation",
"organic",
"chemistry",
"physiology",
"biogenic",
"amines",
"biology",
"and",
"life",
"sciences",
"sensory",
"systems",
"physical",
"sciences",
"cognitive",
"science",
"attention"
] |
2018
|
How pupil responses track value-based decision-making during and after reinforcement learning
|
This study was undertaken in five onchocerciasis/lymphatic filariasis ( LF ) co-endemic local government areas ( LGAs ) in Plateau and Nasarawa , Nigeria . Annual MDA with ivermectin had been given for 17 years , 8 of which were in combination with albendazole . In 2008 , assessments indicated that LF transmission was interrupted , but that the MDA had to continue due to the uncertain status of onchocerciasis transmission . Accordingly , assessments to determine if ivermectin MDA for onchocerciasis could be stopped were conducted in 2009 . We evaluated nodule , microfilarial ( mf ) skin snip , and antibody ( IgG4 response to OV16 ) prevalence in adults and children in six sentinel sites where baseline data from the 1990s were available . We applied the 2001 WHO criteria for elimination of onchocerciasis that defined transmission interruption as an infection rate of <0 . 1% in children ( using both skin snip and OV16 antibody ) and a rate of infective ( L3 ) blackflies of <0 . 05% . Among adult residents in sentinel sites , mean mf prevalence decreased by 99 . 37% from the 1991–1993 baseline of 42 . 95% ( 64/149 ) to 0 . 27% ( 2/739 ) in 2009 ( p<0 . 001 ) . The OV16 seropositivity of 3 . 52% ( 26/739 ) among this same group was over ten times the mf rate . No mf or nodules were detected in 4 , 451 children in sentinel sites and ‘spot check’ villages , allowing the exclusion of 0 . 1% infection rate with 95% confidence . Seven OV16 seropositives were detected , yielding a seroprevalence of 0 . 16% ( 0 . 32% upper 95%CI ) . No infections were detected in PCR testing of 1 , 568 Simulium damnosum s . l . flies obtained from capture sites around the six sentinel sites . Interruption of transmission of onchocerciasis in these five LGAs is highly likely , although the number of flies caught was insufficient to exclude 0 . 05% with 95% confidence ( upper CI 0 . 23% ) . We suggest that ivermectin MDA could be stopped in these LGAs if similar results are seen in neighboring districts .
Onchocerciasis is a leading cause of visual impairment and blindness in many developing countries . The main complications of this infection are severe eye disease that leads to blindness and skin disease characterized by papular or hypopigmented lesions and intense itching . The disease is caused by the filarial nematode Onchocerca volvulus that is transmitted by Simulium species black flies , the most common vector in sub-Saharan Africa being S . damnosum [1] . Fertilized female worms found in subcutaneous fibrous nodules , known as onchocercomas , release microfilariae ( mf ) , the pre larval from of the parasite . These are picked up by the vector when blood feeding and eventually develop into the third stage larvae ( L3 ) that are infectious to humans . The mf are also the cause of the major associated eye and skin morbidity of onchocerciasis . There are no epidemiologically important animal reservoirs [2] . Mass drug administration ( MDA ) with ivermectin is the WHO recommended strategy for the control of onchocerciasis [3] . Ivermectin is a potent microfilaricide that also has a limited effect on the viability and reproductive capabilities of adult onchocercal worms; females are able to regain their ability to produce microfilariae three to six months after ivermectin treatment [4] . This has meant that repeated treatment is needed in order to suppress the manifestations of the infection over time . Computer simulations have suggested that repeated ivermectin treatment could suppress transmission to a point at which the adult parasite population is no longer able to maintain itself , thereby interrupting transmission . While early models estimated this time at about 25 years [5] , more recent evidence has suggested that 5 to 15 years of mass treatment with ivermectin could interrupt transmission depending on the treatment strategy and the initial force of infection [6] , [7] . In Plateau and Nasarawa states in north-central Nigeria , onchocerciasis is endemic in 12 of the 30 districts ( called in Nigeria Local Government Areas - LGAs ) in the two states . Mass drug administration with ivermectin ( MectizanR , donated by Merck & Co . , Inc . , NJ , USA ) in these 12 LGAs began in 1992 and has maintained >80% reported coverage of the treatment eligible population since 1995 [8] , [9] . Beginning in 2000–2001 , ivermectin was combined with albendazole in all 12 onchocerciasis endemic LGAs to treat co-endemic lymphatic filariasis ( LF ) [10] . In 2008 , after 7–8 years of treatment , King et al conducted a survey to determine if LF transmission had been interrupted [11] . The King survey concluded that five of the original 12 co-endemic LF/oncho LGAs had interrupted LF transmission . MDA , however , was not approved by the Federal Ministry of Health to be stopped due to co-endemic onchocerciasis; accordingly the required post treatment surveillance for LF could not be launched . In an effort to ascertain whether onchocerciasis transmission had been interrupted after 17 years of MDA in these five LGAs , and to determine if all ivermectin-based MDA could be halted , the onchocerciasis study reported here was conducted in 2009 . MDA has continued in these five LGAs through 2013 , with ivermectin and albendazole ( 2010–2012 ) and then with ivermectin alone ( 2013 ) . The survey was structured to ascertain changes in microfilaria and nodule rates in villages for which baseline data were available , dating from the early 1990s , before ivermectin MDA was launched . In addition , we focused on two key criteria for halting MDA as outlined in the 2001 WHO document , “Certification of Elimination of Human Onchocerciasis: Criteria and Procedures , ” [12]; 1 ) onchocerciasis infection rates in children of <0 . 1%; and 2 ) entomological infective ( L3 ) larvae prevalence in the S . damnosum black fly vector <0 . 05% . These criteria were operationally adjusted to accommodate population challenges in assessing children by Lindblade in the Americas [13] , [14] , Higazi in Sudan [15] and Katabarwa , et al in Uganda [16]:
The State Ministries of Health for both Plateau and Nasarawa approved the surveys . The protocol was reviewed by the Emory Institutional Review Board ( IRB ) and considered as standard monitoring and evaluation , ( e . g . , deemed to be non-research under Federal Regulations 45 CFR Section 46 . 102 ( d ) ) . Participants provided oral consent for their examinations , and parents who brought their young children for examination provided oral assent . Acceptance or denial of consent/assent was documented on survey forms . Persons >90 cm height and women who were not pregnant or lactating , were treated with ivermectin during MDA according to national program guidelines . Persons acting as human attractants for blackfly catches were told about the personal risks and community benefits of participation and given the option to opt-out of participation at any time without repercussions . Catchers were not compensated by the program . Plateau and Nasarawa states are located in central Nigeria and have an estimated 4 million residents , most of whom live in agricultural villages . Plateau state was split into two ( the eastern half forming Nasarawa state ) in 1997 . Administratively , the two states are divided into 30 LGAs: 17 in Plateau and 13 in Nasarawa . All 30 LGAs are LF endemic , and 12 of these also have meso-hyperendemic onchocerciasis ( estimated nodule rates >20% ) . The five LF/onchocerciasis co-endemic LGAs where King et al determined that LF transmission had been eliminated are Karu , Kokona , Bokkos , Bassa and Jos East . These LGAs are located along the northwestern “oncho belt” of Plateau and Nasarawa states ( Figure 1 ) . Baseline data on nodule and/or mf prevalence were available for fourteen villages . Six of these were considered sentinel sites ( sentinel villages ) originally designated to be followed serially to measure the impact of the program . The other eight villages had baseline data for occasional ‘spot checks . ’ The six sentinel villages consisted of two in Bassa and one in each of the other four LGAs ( Table 1 ) . The spot check villages consisted of three in Karu , two in Kokona , and one in each of the other three LGAs . A list of these villages is shown in Table 1 , together with baseline data collected prior to launching ivermectin MDA , between 1991 and 1993 . Baseline assessments were conducted on an average of 30 male adults , ≥20 years of age , and resident for at least five years , who were selected at random from among volunteers for evaluation . Sentinel sites were more likely ( 83% ) to have baseline data that included mf prevalence than were spot check villages ( 25% ) . Nodule assessments were conducted ( by palpation ) to find subdermal masses that met the characteristic clinical appearance of onchocercomas; rates were calculated as the total number of persons having any nodules divided by the total number examined . Skin snips were obtained from the left and right iliac crest using a field sterilizable punch biopsy instruments ( corneoscleral punches ) . The skin snips ( about 1–3 mg in weight ) were incubated in normal saline and examined after 24 hours by microscopy ( 40× ) for mf . The Plateau State Ministry of Health launched mass drug administration ( MDA ) with ivermectin in 1992 , with the assistance of the River Blindness Foundation ( RBF ) . The program treatment coverage goal was to reach at least 80% of the eligible population using community based distributors ( CBDs ) selected by the individual communities and trained by the RBF and Ministry staff . Ivermectin was provided to the CBD in the presence of the local chief at a time selected by the community leaders [8] . CBDs were then given 2–4 weeks to complete drug distribution and to report back to the Ministry of Health . Treatments were then verified by Ministry of Health staff . Communities with the highest endemicity were targeted first: During the first year of treatment , 217 , 289 persons were treated in 636 communities , 72% of the treatment goal of 300 , 000 . By 1994 , treatments had increased to 448 , 521 in 939 communities , 75% of the treatment goal of 600 , 000 . In 1995 , the program achieved 87% of its treatment goal of 620 , 000 and has maintained >80% reported coverage ever since [9] , [10] . In 1996 , the strategy of MDA changed to the ‘Community Directed Treatment with Ivermectin’ ( CDTI ) strategy of the African Program for Onchocerciasis Control ( APOC ) . In 2000 , albendazole was added to the ivermectin MDA to treat lymphatic filariasis which was co-endemic with onchocerciasis . Treatment records do not allow disaggregating figures for the five LGAs in question , but overall reported coverage in the two state area remained >80% of the treatment eligible population for all years . In 2003 a cluster coverage survey based on questionnaires provided independent treatment figures from treatment eligible persons drawn from a sample frame that included all 30 LGAs . The details of that survey were reported by Richards ( 2011 ) [10] . Questionnaire results showed an ivermectin/albendazole treatment coverage of 72 . 20% ( 95% CI 65 . 5–79 . 0 ) of the treatment eligible population in all 30 LGAs . Sentinel Village Surveys: All 6 villages were surveyed at 12–14 months after the last round of MDA for prevalence of nodules , mf and onchocerciasis antibody ( OV16 ) . All eligible residents of the six sentinel villages ( total population 13 , 512 , range 692 to 3 , 840 ) were invited to participate and all those who arrived on the appointed day were enrolled . Name , age , years of residency , and last year of ivermectin treatment were all recorded . Nodule assessments were conducted by palpation in a manner similar to that used in the 1990s baseline surveys . If nodules were found , their locations were noted on a chart and the number recorded . Nodule rates were again calculated as the total number of persons having any nodules divided by the total number examined . The mf sampling methodology differed from the baseline assessments , because of concern about HIV and other viral infection risks when using field sterilization of corneoscleral punches . Therefore , in the 2009 surveys , disposable sterile needles and blades were used . After cleaning the skin with alcohol , the tip of a sterile needle was used to elevate 3–4 mm of skin over the right and left iliac crest . A sterile blade was then used to remove the skin at its base . The skin snip was then transferred to sterile saline solution in a 96 well plate and a second sample was taken from the left side using the same procedure . The blade and needle were then discarded . Skin snips were fixed in formalin after 24 hours incubation at room temperature and subsequently examined under a microscope ( 40× ) for O . volvulus mf . Skin snips were obtained from the left and right iliac crest of adults ( ≥20 years ) and children ( 3–12 years ) . Blood was collected by finger pricking . Blood samples were placed on Whatman No . 2 filter paper ( Sigma-Aldrich , St . Louis , MO , USA ) , air dried , separated by sheets of paper , and stored in plastic bags in a cooler until they were returned to the laboratory in Jos and stored at 4°C . Blood samples were processed for IgG4 antibody against the OV16 recombinant antigen using a standard ELISA as described [13] , [14] . ELISA testing was performed at The Carter Center laboratory in Jos , the capital of Plateau State . The 2001 WHO criteria for demonstration of interruption of transmission require an infection prevalence of <0 . 1% in children , with 95% confidence . This requires a sample size of at least 3 , 000 children . To reach this number , expanded surveys in children were conducted in the ‘spot-check’ villages and using the same methodology as above , children were examined for nodules and tested for mf ( skin snip ) and OV-16 ( blood spot ) . Spot check villages were selected based on their having had high baseline ( nodule ) prevalence in the 1990s ( see Results section ) . The results were combined with those from children in the sentinel villages . Capture sites were established at Simulium vector breeding points near the 6 sentinel villages . Villagers were asked to identify three to four persons that would serve as attractants on catch teams and two capture sites per village were established . Fly-catchers were from the sentinel villages and were supervised by the LGA ministry of health personnel . Fly catches took place during the peak black fly breeding season , from mid-June through August , 2009 . Each site had four capture days per month , each consisting of hourly collections ( 50 minutes of catches and 10 minutes of rest ) between 7:00am and 5:50pm . Flies were captured in tubes before they could bite , labeled , and preserved with 95% ethyl alcohol . Flies were analyzed at The Carter Center lab in Jos . Analysis was done using O-150 polymerase chain reaction ( PCR ) and PCR products detected by ELISA to determine the presence of O . volvulus DNA [13] , [14] , [15] . Flies were grouped according to village into pools of 100 using positive controls . Fly heads and bodies were analyzed separately to determine if there was any infected flies ( L1 or L2 in the bodies ) or infective flies ( L3 in the heads ) . If positives were found , the pool was run again for confirmation . L3 results were used to determine the seasonal transmission potential ( STP ) , which is the theoretical number of L3s a person receives as the result of infective vector bites during the high transmission season . Persons resident for <5 years , or children <5 years who had not resided all their lives in the community , were excluded in the final analysis . Mean comparison , chi square , and 95% confidence intervals for mf , nodules , and OV-16 were determined using STATA 11 ( SataCorp LP , College Station , TX ) . A P value of <0 . 05 was considered statistically significant . Confidence intervals for blackfly analysis were calculated in PoolScreen 3 . 0 ( University of Alabama , Birmingham , AL ) .
Baseline results from the 1990s are shown in Table 1 . Of the six sentinel villages surveyed , 3 had nodule prevalence data and 5 had mf data . Of the eight spot check villages , all 8 had nodule results and 2 had mf results . From 11 of the 14 villages , 328 persons were assessed for nodules and from 7 of the 14 villages , 210 persons were assessed for mf . The total nodule prevalence in the three sentinel villages was 47 . 10% ( range 7 . 40–83 . 30% ) and the total mf prevalence was 42 . 90% ( range 33 . 30–45 . 60% ) in the five sentinel villages with results . The total nodule prevalence in the eight spot check villages was 39 . 80% ( range 16 . 10–46 . 60% ) and the total mf prevalence in the two spot check villages was 24 . 60% ( range 9 . 70–40 . 00% ) . Overall , mean nodule prevalence was 41 . 80% ( range 7 . 40–83 . 30% ) and the mean mf prevalence was 37 . 62% ( range 9 . 70%–55 . 70% ) . Thirteen ( 93% ) of the 14 villages were classified as being meso-hyperendemic ( mf or nodule rates being >20% ) , the only exception being Gada ( in Jos East ) which was hypoendemic ( nodule prevalence of 16% ) . In the six sentinel villages , the total sample was 2 , 197 ( 739 adults and 1 , 458 children ) . Among the adults ( aged ≥20 ) , the mean age was 43 years and 43 . 12% were male ( n = 329 ) . The mean number of years resident in the village was 37 years , and 92 . 18% reported having taken ivermectin in the past . The majority of adults reported their occupation as “farmer” ( 68 . 62% ) , followed by “housewife” ( 19 . 09% ) , student ( 6 . 01% ) , “business” ( 2 . 94% ) , “other” ( 1 . 87% ) , civil servant ( 0 . 67% ) , and herdsman ( 0 . 40% ) . Only 8 adults ( 4 male , 4 female ) in the six sentinel villages had nodules , resulting in a prevalence of 1 . 08% ( village range 0–4 . 00% ) ; overall , this was a 97 . 60% reduction ( p<0 . 001 ) from the 1991 baseline ( 97 . 41% amongst males alone as measured at baseline ) . Only two adults had positive skin snips , resulting in an mf prevalence of 0 . 27% ( village range 0–1 . 82% ) , a 99 . 30% reduction ( p<0 . 001 ) over baseline ( Table 2 ) . Twenty six adults had OV16 IgG4 antibody ( 3 . 52% , with village range 0–10 . 90% ) , which was 13 times the mf rate . Both individuals who were positive by skin snip were OV16 positive . Of the 8 adults with nodules , zero had mf in skin and 2 ( 25% ) were OV16 positive . Among children in sentinel villages , mean mf and OV16 rates were 0% and 0 . 41% respectively ( 6 OV16 positives , village range 0%–0 . 79% ) ( Table 3 ) . The combined results for children and adults in the sentinel villages are shown on the lower panel of Table 2 . The mean nodule prevalence was 0 . 36% ( village range 0%–1 . 2% ) , the mean mf prevalence was 0 . 09% ( range 0%–0 . 57% ) , and the mean OV16 prevalence was 1 . 46% ( range 0%–3 . 70% ) . A total of 4 , 451 children ages 3 to 12 were examined in sentinel ( 1 , 458 children ) and spot check ( 2 , 993 children ) villages: 47 . 60% were female , the mean age was 8 years , and the mean number of years reported as resident was 8 years with 98 . 20% having been resident their entire lives . Among children who would have been eligible for ivermectin treatment during the previous year ( i . e . ≥6 years ) , 90 . 20% reported having taken the drug . No positive skin snips or nodules were found ( Table 3 ) . In contrast , seven children were OV16 positive , resulting in a prevalence of 0 . 16% ( village range 0–0 . 79% ) . OV16 rates were significantly higher in the sentinel villages ( 0 . 42% , range 0–0 . 79% ) compared to the spot check villages ( 0 . 03% , range 0–0 . 49% ) ( p<0 . 002 ) . Adults in sentinel villages ( Table 2 ) were more likely to be OV16 positive ( p<0 . 001 ) and mf positive ( p<0 . 01 ) than children ( Table 1 ) . There was no statistical difference in the prevalence of OV16 positive children by age group ( p = 0 . 342 ) , but older children , ages 10 to 12 , had a seroprevalence that exceeded the 0 . 1% threshold ( 0 . 29% ) while children under 10 were right on the 0 . 1% threshold ( Table 4 ) . Of the seven OV16 positive children , only one , aged 10 ( resident of Kamwai village , Bokkos LGA ) , had not been resident their entire life and had moved there at the age of 4 . All but two of the OV16 positive children ( ages 4 and 6 ) had received at least one round of ivermectin . The total number of black flies captured in the six sampled village was only 1 , 568 . The highest number of flies caught in a single village was 568 in Kamwai and the lowest was 3 in Anacha . PCR runs on the head and body pools were negative for O . volvulus DNA . Since there were no positive heads , there were no infective larva and therefore the STP was 0 .
Onchocerciasis and lymphatic filariasis ( LF ) have overlapping endemicity in many parts of Africa , and the ‘stop MDA’ decision for one is influenced by the transmission status for the other . This challenge was described by Katabarwa for the Wadalai focus of onchocerciasis in Nebbi District , Uganda [16] . In that case , ivermectin MDA for onchocerciasis could not be halted since ongoing transmission of LF required that MDA with ivermectin and albendazole continue . Katabarwa noted , “As elimination of onchocerciasis becomes more of a prospect in Africa , coordination of onchocerciasis and LF elimination efforts is essential in foci such as Wadelai where co-endemicity exists so that elimination of both diseases can be achieved in an integrated fashion , allowing similar interventions to be halted at the same time . ” ( Katabarwa et al ( 2012 ) , page 6 ) The situation in Plateau and Nasarawa States in Nigeria is the reverse side of the ‘oncho/LF stop MDA’ challenge . In 2009 King et al . showed that LF transmission had been interrupted in ten LGAs in the two state area , but the Federal Ministry of Health gave permission for stopping MDA only in five ‘LF only’ LGAs [11] . Permission was not granted to stop MDA in the five co-endemic oncho/LF LGAs until such time as onchocerciasis ‘stop MDA’ surveys could be completed effectively postponing post treatment surveillance for LF indefinitely . There are two sets of epidemiological criteria for stopping onchocerciasis MDA currently in use: 1 ) The 2001 WHO Geneva criteria ( used by the Onchocerciasis Elimination Program for the Americas [13] , Sudan [15] , and Uganda [16] , which focus on demonstrating very low infection rates in young children as the primary support for claiming absence of recent transmission - very similar to the LF TAS survey approach [17] ) ; and 2 ) the African Program for Onchocerciasis Control ( APOC ) criteria [18] based on work done in studies in Senegal and Mali [19] and in Kaduna , Nigeria [20] . In Senegal and Mali , Diawara , et al . used their observations to validate the ONCHOSIM model predicting transmission interruption would occur when overall village mf prevalence in a transmission zone fell below <5 . 0% in 100% of villages and <1 . 0% in 90% of villages [19] . Tekle et al ( 2012 ) reported that the annual ivermectin distribution program had achieved these mf thresholds in Kaduna state in Nigeria , a state that borders both Plateau and Nasarawa [20] . In our study we found epidemiological evidence suggesting that onchocerciasis transmission in the five oncho/LF LGAs of concern had been broken considering all the above criteria . Our surveys were designed to be most robust in demonstrating the 2001 WHO Geneva guideline requirement of very low infection rates in children . Skin snip prevalence was 0 in over 4000 children aged 3–12 years examined ( upper 95% CI<0 . 01% ) in our study . Community wide skin snip prevalence met APOC stop MDA requirements in all six sentinel villages , with total community ( e . g . , combined adults and children skin snip results ) mf prevalences each being <1% , and no individual community being over 5% . Of note , APOC has subsequently conducted an independent evaluation of onchocerciasis throughout the two state endemic area and reported 100% negative skin snip findings from a more robust community sample [21] Both WHO and APOC entomologic criteria for the onchocerciasis stop MDA decision seek infective ( L3 ) rates in vectors <0 . 05% ( with 95% confidence ) , requiring a sample size of at least 6000 flies . Unfortunately , our teams had difficulty capturing the required 6000 flies within the time frame and budgetary allowances of the study . We established 10 capture sites in four of the five LGAs of interest ( Kokona , Bassa , Bokkos , and Jos East ) with four capture days ( 7:00am and 5:50pm per day ) per month during the black fly breeding season ( mid-June through August ) ( e . g . , 10 capture days per site ) . In other words , 100 person/days ( and over 1000 person/hours ) of human landing capture time was only sufficient to collect 1 , 568 Simulium damnosum s . l . flies , just 2 . 5% of the needed sample . To reach the WHO/APOC requirement , we would have had to increase our field activities to 400 person days ( sixteen capture days per month ) , well beyond our logistical capacity . One recommendation from our study is that new approaches are needed for capturing Simulium vectors of onchocerciasis if these current entomological guidelines are to be met by the many programs seeking to stop MDA for onchocerciasis in this part of Nigeria . Operations research is needed to determine how to improve capture numbers without increasing the time required of human attractants . New black fly traps that are being developed may be an answer for doing this [22] . We found all 1 , 568 S . damnosum s . l . bodies and heads tested by PCR to be negative for O . volvulus DNA; as noted , this result is not sufficient to statistically exclude the WHO/APOC 0 . 05% infectivity threshold with 95% confidence . Despite this finding we are recommending that these results be used in support of a ‘stop MDA’ decision for the five LGAs . Important to note is that there was no evidence of any infection in the vector bodies , where earlier stage larvae ( L1 , L2 ) are found . Earlier ( L1 and L2 ) stages in the flies are likely to be 3–5 times more frequent than L3s , and absence of any infection is as powerful as a ‘xenodiagnostic’ demonstrating the absence of mf in humans in these areas , and highly consistent with our skin snip results . The evidence provided here ( albeit entomologically incomplete ) supports a conclusion that transmission of onchocerciasis had likely been broken in 2009 . Given the fact that four more years of MDA have been provided since this study , and another MDA is planned for 2014 , we think that interruption has been achieved now and that MDA could be stopped in 2015 . Our results , however , apply only to the five oncho/LF co-endemic LGAs discussed . Another onchocerciasis survey is likely needed to make a decision on the remaining oncho/LF LGAs . Such a survey is would need to pay particular attention to collecting a sufficient number of black flies . This paper and its companion by King et al . [11] provide an example of the complexities that encompass the stop MDA decisions in oncho/LF co-endemic areas . We favor the establishment of a Nigeria expert advisory committee to the FMOH , similar to the one the Ugandan Ministry of Health has established ( the Ugandan Onchocerciasis Elimination Expert Advisory Committee ) [16] . Such a committee would be very helpful in reviewing in detail the often incomplete , imperfect , or highly nuanced data sets , and so give best advice to the FMOH on where and when to halt interventions . As we have seen in this example , the onchocerciasis and LF programs should work closely together , perhaps through a single independent advisory review committee having combined oncho/LF responsibility to give a more rational approach to expediting the stop MDA process . Our study gave us the opportunity to compare skin snip infection rates with the OV16 IgG4 ELISA . Among adults in the sentinel villages , OV16 antibody prevalence was 13 times greater than skin snip rates and 3 . 2 times greater than the nodule rates . All mf positive adults were OV16 positive , but the correlation with nodules ( which were clinically onchocercomas but not parasitologically confirmed ) was not as good . This is likely due to the well-recognized fact that as prevalence of onchocerciasis decreases , the specificity of clinical onchocercomas drops [23] . The fact that OV16 rates were low , 3 . 5% among long term resident adults in villages documented as being meso-hyperendemic in the 1990s , is evidence that the OV16 IgG4 response wanes over the years as transmission is broken and the adult O . volvulus parasites die and are not replenished . The OV16 prevalence in children was higher than the skin snip prevalence , with no OV16 positive child having a positive snip . This could be due to the fact that their infections were below the sensitivity of a skin snip , or that their antibody response represented a recent exposure or a single sex infection , rather than a patent one . In the future , the use of PCR in skin snips from children who are OV16 positive might be considered as a better confirmatory test [24] . We also documented the classical increase of antibody positivity with increasing age . Given this , the general approach in the future should be to sample children aged under ten years for OV16 assessments [24] , [16] . If we focus on children aged under ten , the OV16 antibody rate positivity rate was 0 . 10% ( n = 3 , 050 , 0 . 29% upper 95%CI ) . This raises another important issue , which is that the 2001 WHO guidelines do not distinguish ‘infection rates’ determined by nodule prevalence , mf prevalence , serological prevalence , or DEC patch test . Clearly more clarification is needed for the 1/1 , 000 infection benchmark . In this study we chose mf prevalence since this was the clearest representation of active infection of the three indices used ( mf , nodule and OV16 antibody ) . The current data leave us asking ourselves ‘Is the OV16 antibody prevalence threshold ( when used alone as an indicator of infection rates ) of <0 . 1% too high of a standard for transmission interruption ? ’ Aside from the aforementioned weakness of the study in having insufficient numbers of black fly vectors to exclude the critical 0 . 05% infectivity threshold , there were at least three other weaknesses in this study . First , we compared the 1990s baseline nodule and mf rates determined in small samples of adult men with a larger follow up adult study that included adult women ( 57% of the sample ) . We also calculated 2009 community mf prevalence based on a sample that excluded the age groups 13–19 years . We do not believe either of these shortcomings dramatically altered the results or our conclusions , and we note that prevalence did not differ significantly between men and women in the 2009 sample . Second , the study was conducted in five LGAs selected based on the results of an earlier LF assessment by King . These LGAs may or may not be representative of the overall onchocerciasis transmission zone; it is uncertain because we do not know the subspecies of S . damnosum and its flight range . The worst case scenario would be a large transmission zone ( e . g . , far reaching flight range ) that includes all the endemic LGAs in Plateau and Nasarawa as well as LGAs in neighboring Kaduna state . However , sentinel data ( not presented here ) from other onchocerciasis endemic LGAs in Plateau and Nasarawa show similar negative results ( particularly in children ) , and published evidence from Kaduna suggests that transmission interruption is likely there as well [20] . Thus there is reason to suspect that transmission of onchocerciasis has been broadly interrupted by the MDA programs , even in the worst case scenario . This study indicates the need , in onchocerciasis/LF co-endemic areas , to coordinate field assessments for the stop MDA assessments . If onchocerciasis assessments can be done together with LF transmission assessments surveys ( TAS ) [17] time and money can be saved in making a required joint programmatic decision to stop ivermectin based MDA . We urge that new tools and associated operations research be undertaken promptly to allow data that can be obtained concurrently and ideally from the same samples and age groups , to allow the stop MDA decision by technical committees with sufficient expertise in the epidemiological dynamics and accepted WHO guidelines for each disease .
|
Both lymphatic filariasis and onchocerciasis are treated with ivermectin-based mass drug administration ( MDA ) regimens in Africa . Where the infections are co-endemic , ivermectin treatments cannot be stopped until both infection transmission cycles are broken . This report follows a previous determination that the LF transmission cycle had been interrupted in five districts ( LGAs in Nigeria ) but evidence was needed on the status of the onchocerciasis transmission cycle prior to halting MDA . In this report we determined ( based on WHO guidelines ) that most likely the transmission of onchocerciasis has been interrupted in Plateau and Nasarawa States and we conclude that ivermectin MDA could be stopped .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"public",
"and",
"occupational",
"health",
"helminth",
"infections",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"plant",
"science",
"filariasis",
"infectious",
"disease",
"epidemiology",
"epidemiology",
"global",
"health",
"plant",
"pathology",
"biology",
"and",
"life",
"sciences",
"lymphatic",
"filariasis",
"onchocerciasis",
"tropical",
"diseases",
"neglected",
"tropical",
"diseases",
"parasitic",
"diseases"
] |
2014
|
Status of Onchocerciasis Transmission after More Than a Decade of Mass Drug Administration for Onchocerciasis and Lymphatic Filariasis Elimination in Central Nigeria: Challenges in Coordinating the Stop MDA Decision
|
Nucleotide Excision Repair ( NER ) , which removes a variety of helix-distorting lesions from DNA , is initiated by two distinct DNA damage-sensing mechanisms . Transcription Coupled Repair ( TCR ) removes damage from the active strand of transcribed genes and depends on the SWI/SNF family protein CSB . Global Genome Repair ( GGR ) removes damage present elsewhere in the genome and depends on damage recognition by the XPC/RAD23/Centrin2 complex . Currently , it is not well understood to what extent both pathways contribute to genome maintenance and cell survival in a developing organism exposed to UV light . Here , we show that eukaryotic NER , initiated by two distinct subpathways , is well conserved in the nematode Caenorhabditis elegans . In C . elegans , involvement of TCR and GGR in the UV-induced DNA damage response changes during development . In germ cells and early embryos , we find that GGR is the major pathway contributing to normal development and survival after UV irradiation , whereas in later developmental stages TCR is predominantly engaged . Furthermore , we identify four ISWI/Cohesin and four SWI/SNF family chromatin remodeling factors that are implicated in the UV damage response in a developmental stage dependent manner . These in vivo studies strongly suggest that involvement of different repair pathways and chromatin remodeling proteins in UV-induced DNA repair depends on developmental stage of cells .
A network of DNA damage response ( DDR ) mechanisms protects organisms against the continuous genotoxic stress induced by reactive metabolites and other genotoxic agents , such as environmental contaminants and ultraviolet ( UV ) radiation from the sun [1] . The DDR network consists of several DNA repair mechanisms , cell cycle checkpoints and cellular senescence and apoptotic signaling cascades . Nucleotide Excision Repair ( NER ) is a DNA repair mechanism that is able to remove a wide variety of helix-destabilizing DNA lesions including those induced by UV light . Eukaryotic NER is a highly conserved multi-step process , involving more than 25 proteins , of which the principal molecular mechanism has been dissected in detail [1] , [2] . NER is initiated by two distinct DNA damage recognition mechanisms which use the same machinery to repair the damage . Damage in the transcribed strand of active genes is repaired by Transcription Coupled Repair ( TCR ) , which depends on recruitment of the ATP-dependent chromatin remodeling protein Cockayne Syndrome protein B ( CSB ) and the WD40 domain containing protein Cockayne Syndrome protein A ( CSA ) to the site of damage [3]–[5] . TCR is thought to be activated by stalling of elongating RNA polymerase II during transcription [3] , [6] . Damage in other , non-transcribed sequences of the genome is repaired by Global Genome Repair ( GGR ) , which requires detection of the lesions by the UV-damaged DNA-binding protein ( UV-DDB ) complex and a complex containing Xeroderma Pigmentosum group C protein ( XPC ) , human homolog of RAD23 ( hHR23 ) and Centrin-2 [7]–[9] . The XPC protein is essential for the initiation of GGR and subsequent recruitment of other NER factors [10] , [11] . The majority of XPC is found in complex with the hHR23B protein , while only a fraction copurifies with the redundant hHR23A protein . Both hHR23 proteins are thought to stabilize XPC and stimulate its function [12]–[14] . Although HR23B is not essential for in vitro NER , in vivo damage is poorly repaired in cells lacking hHR23B [12] , indicating that hHR23B is essential for proper NER function . Following detection of a lesion , either via GGR or TCR , the transcription factor IIH ( TFIIH ) is recruited to open the DNA helix around the damage in an ATP-dependent manner using its Xeroderma Pigmentosum group B and D ( XPB and XPD ) helicase subunits [1] , [2] . Next , Xeroderma Pigmentosum group A ( XPA ) and Replication Protein A ( RPA ) are recruited to stabilize the repair complex and properly orient the structure-specific endonucleases Xeroderma Pigmentosum group F ( XPF ) /Excision Repair Cross-Complementing protein 1 ( ERCC1 ) and Xeroderma Pigmentosum group G ( XPG ) to excise the damaged strand . The resulting ∼30 nt single strand DNA gap is filled by DNA synthesis and ligation . In mammals , congenital defects in GGR and TCR lead to an increased sensitivity towards DNA damaging agents such as UV irradiation . Inherited mutations in GGR genes cause Xeroderma Pigmentosum , which is characterized by extreme UV-sensitivity and skin cancer predisposition [15] . Hereditary TCR deficiency causes Cockayne syndrome , which leads to entirely different features such as severe but variable neurodevelopmental symptoms and premature aging . In contrast to mammals , specific TCR defects in yeast have only a marginal effect on DNA damage resistance , despite a relatively larger proportion of the genome that is transcriptionally active [16] . Current knowledge of NER does not provide an explanation for the pleiotropic phenotypic expression of NER-deficiencies . Despite detailed insight in the molecular mechanism of NER , many aspects of the in vivo UV-induced DNA damage response ( UV-DDR ) are still unclear . It is for instance not well understood how NER functions in nucleosomal DNA and in different tissues of developing organisms . Therefore , a full understanding of the complete UV-DDR and its interplay with NER in living organisms is imperative . The nematode C . elegans seems well suited to analyze the complete UV-DDR in vivo in more detail , because of its short lifetime , well-characterized biology and its amenable use to identify new genes involved in the UV-DDR . Several studies have specifically addressed the role of NER proteins in the UV-DDR in C . elegans . Knockdown of the C . elegans orthologs of mammalian CSB , XPA and XPF increases sensitivity to UV irradiation [17]–[21] . Furthermore , it was shown that the XPA and XPC orthologs function in the C . elegans germ line to induce cell cycle arrest and apoptosis in response to UV irradiation [22] . Together , these studies suggest that NER function is highly conserved in C . elegans . However , a thorough analysis of the function of NER and , more specifically , the role of the GGR and TCR subpathways in response to UV irradiation in different tissues during development has not been performed . In this study , we make use of mutations in the C . elegans RAD23 , XPC and CSB orthologs to show that during early development , in germ cells and embryos , GGR is the major pathway involved in the response to UV irradiation . Defective GGR leads to inefficient lesion removal in germ cells , specific defects in germ cell development and embryonic death after UV irradiation . Intriguingly , in juvenile and adult animals TCR is the major NER pathway involved in the UV response . Analysis of the UV response of embryos shows that , during development , TCR gradually becomes more important than GGR . Finally , we exploit C . elegans to identify novel genes involved in the UV-DDR , specifically in the TCR-related UV response . Our results reveal four genes implicated in SWI/SNF and four genes implicated in ISWI ATP-dependent chromatin remodeling whose involvement in the UV-DDR changes during development .
To study the UV-DDR in the context of a whole organism , we tested UV-B sensitivity of mutant C . elegans at different developmental stages . Our initial experiments showed that UV-B irradiation produced better reproducible phenotypes than UV-C irradiation ( data not shown ) , most likely due to the fact that UV-B penetrates deeper through the multiple cell layers of C . elegans . First , we tested UV sensitivity of animals carrying mutations in the general NER genes xpa-1 , xpg-1 , xpf-1 and ercc-1 . Alleles of xpa-1 and xpf-1 , but not xpg-1 and ercc-1 , have been previously described . xpa-1 ( ok698 ) encodes a putative null allele of the C . elegans ortholog of mammalian XPA and was shown to cause severe sensitivity to UV irradiation [20] , [22] . him-9 ( e1487 ) is an allele of xpf-1 , encoding the C . elegans ortholog of mammalian endonuclease XPF [23] . tm1682 and tm1670 are two alleles of xpg-1 , the ortholog of the mammalian endonuclease XPG and have not been described before . tm1682 represents a deletion of the first two exons of xpg-1 , probably creating a knock-out allele , but also of part of the last exon of the adjacent glycosyl hydrolase gene tre-1 ( Figure 1A ) . Thus , to rule out an effect of tre-1 , in our analysis we also included tm1670 , which represents a deletion that is predicted to remove exon 2 and a large part of exon 3 , encoding for a truncated 679 amino acids in stead of 829 amino acids protein ( Figure 1A ) . Since most of the N-terminal nuclease domain is deleted , the resulting protein is expected to be non-functional . tm2073 represents a deletion in the conserved Rad10 domain of ercc-1 , the C . elegans ortholog of mammalian ERCC1 which is in complex with XPF , and is predicted to encode a loss-of-function allele ( Figure 1B ) . To address the specific contribution of the TCR and GGR pathways in the UV-induced DDR in vivo , we also analyzed C . elegans strains carrying mutations expected to affect either pathway specifically . The genome of C . elegans encodes an ortholog of the GGR-specific mammalian HR23A and HR23B genes , called rad-23 . This gene is predicted to encode for a 372 amino acids protein having similar domain organization as mammalian HR23A and HR23B proteins ( Figure 1C; [24] ) . The rad-23 ( tm2595 ) allele represents a deletion of the major parts of exon 2 and exon 3 and an insertion of 28 basepairs . Since tm2595 deletes both UBA domains and the XPC binding domain and is predicted to encode a truncated protein of 96 aa , this allele is likely a functional null allele . The TCR-specific mammalian CSB gene is represented in C . elegans by csb-1 [17] . This gene encodes a 957 amino acids protein containing a SNF2-like ATPase domain , similar to human CSB ( Figure 1D ) . The csb-1 ( ok2335 ) allele consists of a 1620 bp deletion which removes exon 5 and 6 and the largest parts of exon 4 and 7 . This allele is predicted to encode a truncated protein of 513 amino acids , which is likely a functional null since most of the SNF2 domain is deleted . To test UV sensitivity of germ cells , adult animals were irradiated and allowed to recover for 24 hours , after which they were put on fresh plates to lay eggs for 3–4 hours ( Figure S1A ) . ‘Germ cell and embryo survival’ was measured by determining the percentage of eggs that hatched over the total amount of eggs laid . As expected , we found that the core NER factors xpa-1 , xpg-1 , xpf-1 and ercc-1 were necessary for germ cells and embryos to survive even relatively low doses of UV irradiation ( Figure 2A and 2B ) . Next , we tested UV sensitivity of rad-23 ( tm2595 ) and csb-1 ( ok2335 ) mutants . Functional rad-23 appeared to contribute only partially to UV resistance ( compared to xpa-1 ) , whereas , surprisingly , csb-1 did not seem to contribute at all ( Figure 2C ) . Similar results were obtained using eggs laid immediately after irradiation or after different recovery periods up to 51 hrs after irradiation ( data not shown ) . This suggests that a similar UV response , involving general NER factors and rad-23 , but not csb-1 , acts in all developing germ cells , oocytes and early embryos . The specific contribution of rad-23 but not csb-1 suggests that germ and early embryonic cells depend mainly on the GGR pathway of NER to overcome the effects of UV irradiation . Alternatively , it could be that csb-1 is not involved in TCR in C . elegans or that TCR defects are not associated with UV sensitivity . To test whether GGR and TCR act redundantly in the germ line , or whether csb-1 is not involved in UV-damage repair or survival , we generated animals carrying mutations in both rad-23 and csb-1 . Irradiation of rad-23; csb-1 double mutants showed that these animals are more UV sensitive than rad-23 single mutants and as sensitive as animals carrying mutations in general NER genes ( Figure 2C ) . This suggests that both TCR and GGR are active in germ cells . In mammals , RAD23 functions in GGR as part of a heterotrimeric complex containing also Centrin-2 [7] , [25] and XPC [8] , [26] . The genome of C . elegans contains an ortholog of XPC , xpc-1 , for which only recently , during the course of our experiments , a good loss-of-function allele became available . This allele , tm3886 , represents a 24 bp insertion and 474 bp deletion in exon 2 , probably causing a truncated protein ( Figure 1E ) . To confirm that the specific UV sensitivity of rad-23 is caused by a defect in GGR , we tested the phenotype of the novel xpc-1 mutation . xpc-1 single mutants showed a similar UV sensitivity in the germ line as rad-23 single and rad-23; xpc-1 double mutants , whereas xpc-1; csb-1 double mutants were as UV sensitive as rad-23; csb-1 double mutants ( Figure 2D ) . These results are in line with our previous findings and with the idea that in C . elegans , similar as in mammals , RAD-23 and XPC-1 function in complex during GGR . Based on our results ( summarized in Table 1 ) , we hypothesize that in the germ line GGR plays an essential role in UV survival , whereas TCR only has a secondary , partially redundant function to GGR ( Figure 2E ) . Furthermore , our results are in agreement with the idea that , similar as in mammals , rad-23/xpc-1 and csb-1 act in parallel pathways , GGR and TCR , that converge on a common pathway to repair DNA damage . Previously , it was found that ionizing and UV irradiation both induce apoptosis of a fraction of the pachytene germ cells of C . elegans [22] , [27] , which are located near the gonad tube bend ( Figure 3B , first image ) . Functional xpa-1 was shown to be required for induction of apoptosis [22] , [27] , suggesting that the NER process itself is necessary to activate the apoptotic machinery . To test whether induction of germ cell apoptosis requires functional GGR or TCR , we measured induction of apoptosis in the pachytene germ cells of wild type , xpa-1 , rad-23 , csb-1 and rad-23;csb-1 mutants in response to UVB irradiation . In contrast to wild type animals , xpa-1 mutants exhibited severely reduced apoptosis induction after UVB , as observed previously ( Figure 3A; [22] ) . Furthermore , we found that in csb-1 mutants apoptosis was induced at wild type levels , whereas in rad-23 mutants apoptosis induction was mildly reduced . No apoptosis induction after UV irradiation was observed in rad-23; csb-1 double mutants , similar as in xpa-1 mutants . These results indicate that both the GGR and the TCR pathway are required to induce germ cell apoptosis in response to UV . Together with the mild decrease in apoptosis induction in rad-23 mutants , this is in line with our previous results showing that GGR , acting partially redundant with TCR , is the main NER pathway in the germ line of C . elegans . Surprisingly , UV irradiation does not induce , but even seems to inhibit apoptosis in xpa-1 and rad-23; csb-1 mutants , and less efficiently in rad-23 mutants . In unirradiated animals , germ cell apoptosis is thought to be a developmental mechanism to maintain germ line homeostasis [28] . Following UV irradiation , NER-dependent apoptosis of pachytene stage germ cells may serve to eliminate damaged cells . After exiting pachytene stage , undamaged germ line nuclei progress to complete meiosis and are fertilized as oocytes in the proximal part of the gonad ( reviewed in [29]; Figure 3B , first image ) . Next , fertilized oocytes initiate embryogenesis . Thus , it was interesting to follow the fate of UV-damaged pachytene germ cells in NER proficient and deficient animals . In wild type and csb-1 animals , oocytes in the proximal part of the gonad appeared morphologically normal after UV irradiation . In contrast , in xpa-1 , rad-23 and rad-23; csb-1 mutants , the morphology of oocytes was drastically altered over time after UV irradiation ( Figure 3B , arrowheads ) . Further analysis using DAPI staining to visualize chromatin condensation associated with specific meiotic developmental stages revealed that in xpa-1 , rad-23 and rad-23; csb-1 mutant germ cells failed to progress to the oocyte stage for at least up to 30 hrs after irradiation ( Figure 3C , arrowheads , and Figure 3D; data not shown ) . In contrast , in wild type animals and csb-1 mutants morphologically normal diakinesis stage oocytes were readily recognizable at all time points after UV irradiation ( Figure 3C and 3D and data not shown ) . These results suggest that in UV irradiated animals lacking functional XPA-1 or RAD-23 maturation of meiotic germ nuclei is impaired . Further down the gonad tube , the general morphology of embryos in utero was also severely compromised in xpa-1 , rad-23 and rad-23; csb-1 mutants ( data not shown ) , suggesting extensive embryonic cell death . This latter finding is in agreement with the fact that fewer eggs are laid with increasing dosages of UV irradiation ( data not shown; [30] ) and that eggs which are laid show increased mortality rates . Possibly , the lack of UV-induced apoptosis in these mutants leads to a reduced clearance of UV-damaged cells which results in defects in meiotic maturation , morphological changes and ultimately cell death . Together , these results confirm that in germ cells GGR , but not TCR , is the dominant NER pathway necessary to overcome the genotoxic effects of UV irradiation . To investigate whether the UV hypersensitivity of germ cells of xpa-1 and rad-23 mutants is accompanied or caused by reduced DNA repair , we measured UV damage removal . To this end we applied immunofluorescence to visualize Cyclobutane Pyrimidine Dimers ( CPDs ) , the most abundant UVB-induced DNA lesions [31] . As shown in Figure 4 , 18 hours after irradiation a virtual complete removal of CPDs from gonad nuclei was observed in wild type and csb-1 animals , but not in xpa-1 , rad-23 and rad-23; csb-1 mutants . These results further corroborate the notion that GGR is the major NER pathway in germ cells of C . elegans . To investigate whether the observed GGR dependence of the UV response is restricted to germ cells or whether it is a common feature of C . elegans cells , we determined UV sensitivity of later developmental stages . We found that early developmental stages of C . elegans are more sensitive to UV irradiation than later stages , in line with what was previously described ( data not shown; [21] , [30] ) . To score UV sensitivity of L1 larvae , we developed an assay in which survival of UVB-irradiated L1 larvae was measured ( see materials and methods and Figure S1B ) . Survival was scored by determining the percentage of animals capable of growing to adulthood over the total amount of animals in response to UV irradiation . We found that xpa-1 , xpg-1 , xpf-1 and ercc-1 L1 larvae were extremely sensitive to UV and arrested development completely in response to relatively low UV doses ( Figure 5A and 5B ) . This developmental arrest is possibly caused by a damage-induced block in transcription , causing breakdown of RNA polymerase II , as was shown following UVC irradiation of xpa-1 mutants [20] . However , at the UVB doses we used ( up to 160 J/m2 ) we were unable to confirm breakdown of RNA polymerase II ( data not shown ) . To our surprise , we found that csb-1 L1 larvae , but not rad-23 L1 larvae , were more sensitive to UV than wild type animals ( Figure 5C ) , opposite to what was observed in germ cells . Similar to rad-23 mutant germ cells , csb-1 L1 larvae showed an intermediate UV sensitivity in between wild type animals and general NER mutants . Again we found that rad-23; csb-1 double mutants were more sensitive than either rad-23 or csb-1 single mutant alone and were comparable to general NER mutants ( Figure 5C ) . Although rad-23 mutant L1 larvae did not show increased lethality , they did appear to develop slightly slower in response to UV irradiation ( data not shown ) . Next , we also tested the recently available xpc-1 mutant . xpc-1 single and rad-23; xpc-1 double mutants did not show enhanced UV sensitivity compared to wild type animals ( Figure 5D ) . rad-23; xpc-1 double mutants even showed a mild but reproducible increase in UV survival . Other functions of rad-23 , besides NER [32] , [33] , might account for this observation , although at the moment we do not understand how these might stimulate UV survival . Importantly , xpc-1; csb-1 double mutants showed extreme UV sensitivity comparable to that of general NER mutants and the rad-23; csb-1 double mutant ( Figure 5D ) . Similar results were obtained using older larval stages and young adults instead of L1 larvae ( data not shown ) . Together , these results ( summarized in Table 1 ) suggest that in contrast to germ cells , TCR is the major NER pathway acting in juvenile and adult C . elegans tissues to counteract the effects of UV irradiation ( Table 1 , Figure 5E ) . The GGR pathway seems to act partially redundantly to the TCR pathway . The observed difference in UV survival of rad-23/xpc-1 and csb-1 during development suggests that as a germ cell grows to become an L1 larva , a switch occurs that favors the dependence on one pathway over the other . To test at which developmental stage csb-1/TCR becomes the primary UV survival pathway instead of rad-23/GGR , we collected eggs from adult animals by hypochlorite treatment and irradiated these at different time points after collection ( Figure S1C ) . We found that in early eggs , rad-23 function is still essential for optimal UV survival ( Figure 5F; 1 hr , 20 J/m2 ) , similar to germ cells and embryos . However , in time rad-23 function became gradually dispensable while csb-1 function was more and more essential for optimal UV survival ( Figure 5G; 4 and 8 hr , 40 J/m2 ) . Note that during later time points slightly higher UV doses had to be used due to the fact that early embryos are more UV sensitive than later stage embryos . This phenomenon might be due to growth causing less UV penetrance or higher tolerance of transcription and replication blocking lesions [30] . Irradiation of eggs collected by egg laying gave similar results as eggs collected by hypochlorite treatment ( data not shown; [30] ) . In summary , these results suggest that during embryogenesis , before hatching , GGR gradually becomes less and TCR becomes more important for C . elegans to cope with the toxic effects of UV exposure . The developmental difference between TCR- and GGR-dependent UV-sensitivity of C . elegans suggests that developmental-stage dependent regulatory genes specifically involved in either pathway could be identified using C . elegans . Recently , we have successfully used C . elegans to show that Heterochromatin Protein 1 ( HP1 ) , represented by hpl-1 and hpl-2 in C . elegans , is involved in the UV-DDR [34] , suggesting a role for chromatin condensation status in UV survival . This implies that proteins involved in chromatin dynamics , e . g . chromatin remodeling and epigenetics , may be implicated in the UV-DDR . These proteins are expected to play important roles in controlling the efficiency of DNA repair , by regulating the access to DNA as well as checkpoint signaling associated with DNA repair [35] . CSB itself is an ATP-dependent chromatin remodeling factor , which is thought to alter nucleosome structure to enable repair [36] , [37] . In yeast , fruit flies and mammals , several different ATP-dependent chromatin remodeling complexes , e . g . the SWI/SNF , the ISWI , the NuRD , the CHD and the INO80 families , have been identified , some of which have been implicated in the DDR [35] , [36] . To test whether these remodeling complexes are involved in the developmental stage-dependent UV-DDR in C . elegans , we set up a screen in which we systematically tested L1 larvae UV sensitivity of animals in which subunits of these major remodeling complexes or genes carrying motifs predicted to be involved in ATP-dependent chromatin remodeling were knocked down either by mutation or RNAi ( Table S1 ) . UV survival of L1 larvae in which proteins of the NuRD , the CHD and the INO80 chromatin remodeling family were knocked down closely mimicked that of wild type larvae ( data not shown ) , suggesting no involvement in the UV-DDR . In contrast , knockdown of four proteins of the ISWI family and four proteins of the SWI/SNF family resulted in increased UV-sensitivity ( Table 1 , Figure 6 ) . We tested two partial loss-of-function alleles of the ISWI/SMARCA5 chromatin remodeling ATPase subunit isw-1 [38] . isw-1 ( n3297 ) animals showed reproducible sensitivity to UV irradiation ( Figure 6A ) , but isw-1 ( n3294 ) animals did not ( Figure S2A ) . Surprisingly , isw-1 ( n3297 ) carries a missense mutation within a non-conserved region of the gene while isw-1 ( n3294 ) encodes a missense mutation in a conserved DEXD/H box helicase domain required for chromatin remodeling activity [38] . Since isw-1 null mutants are not viable , we additionally knocked down isw-1 using RNAi and confirmed that isw-1 functions in the UV-DDR ( Figure S2B ) . Furthermore , deletion alleles of hda-2 ( ok1479 ) and hda-3 ( ok1991 ) , which represent orthologs of human class I histone deacetylase [39] , and mutation of him-1 , the C . elegans ortholog of human cohesin protein SMC1 , which are all found in complex with human ISWI/SMARCA5 [40] , also increased UV-sensitivity ( Figure 6A ) . To confirm the significance of these findings , we reproduced the observed UV sensitivities in multiple independent experiments . Knockdown of hda-2 and hda-3 by RNAi was also attempted , but was found to produce variable results ( data not shown ) , possibly because efficacy of the RNAi was not always optimal . As the e879 allele used for him-1 was described to be temperature-sensitive [41] , we additionally tested him-1 mutants at 25°C and found them to be more UV sensitive than at 20°C ( Figure S3A ) . This increased UV sensitivity at the restricted temperature further confirmed that this gene is indeed implicated in the UV-DDR . Next , we tested animals carrying a temperature sensitive missense mutation ( os13 ) in the SWI2/SNF2 chromatin remodeling ATPase subunit psa-4 [42] , a putative ortholog of human BRM/SMARCA2 . Animals tested at a permissive temperature ( 20°C ) were found to be mildly sensitive to UV ( Figure 6B ) , whereas animals tested at a nonpermissive temperature ( 25°C ) showed a strongly enhanced UV sensitivity ( Figure S3B ) . Additionally , mutations in other subunits of SWI/SNF remodeling complexes , e . g . the SMARCC1 ortholog psa-1 ( os22 and ku355; [42] , [43] ) , the PBRM1 ortholog pbrm-1 ( tm415 ) and the SMARCB1 ortholog snfc-5 ( ok622 ) also increased UV-sensitivity ( Figure 6B , Figure S2C ) . As both psa-1 alleles were described to be temperature-sensitive , we tested both alleles at 25°C and found them to confer even stronger UV-hypersensitivity than at 20°C ( Figure S3B ) . The UV hypersensitivities of all SWI/SNF mutants were reproduced in multiple , independent experiments , corroborating their significance . Furthermore , knockdown of pbrm-1 and snfc-5 using RNAi also mildly increased UV sensitivity ( data not shown ) . In summary , these results implicate the ISWI and SWI/SNF chromatin remodeling complexes in the UV-DDR of C . elegans . Mutation or RNAi-mediated knockdown of other members of both ATP-dependent chromatin remodeling complexes ( Table S1 ) had no effect , possibly because RNAi was not efficient or because these factors do not play a role in the UV-DDR . Involvement of some factors could not be tested due to lethality . In addition to the L1 larvae survival experiment , we tested whether the eight identified genes are also involved in the UV-DDR of germ cells and embryos . Since both isw-1 mutants did not lay sufficient eggs on a regular basis , we tested isw-1 involvement using RNAi . Knockdown of the isw-1 and psa-4 ATPase subunits of ISWI and SWI/SNF chromatin remodeling complexes , and of the cohesin member him-1 , rendered germ cells sensitive to UV ( Table 1 , Figure 7 ) . However , mutation of the other ISWI and SWI/SNF subunits had no significant effect on UV survival . These results suggest that ISWI and SWI/SNF chromatin remodeling activity is involved in UV survival of germ cells and embryos , but the response in germ cells seems to involve other subunits than the response in L1 larvae . The specific UV sensitivity of L1 larvae but not germ cells caused by knockdown of certain chromatin remodeling genes suggests these genes might be involved in TCR but not GGR . If this is the case , knockdown of these genes in a GGR-deficient background could lead to an even more pronounced UV sensitivity , similar as observed for the rad-23; csb-1 double mutants . Likewise , genes that affect UV sensitivity in L1 larvae as well as germ cells might be generally involved in NER , in both TCR and GGR . Inactivation of these genes in a GGR- or TCR-deficient background should not lead to increased UV sensitivity . To test this , we inactivated isw-1 , which affects sensitivity in germ cells and L1 larvae , and pbrm-1 , which only affects L1 larvae sensitivity , in rad-23 and csb-1 mutants . RNAi-mediated knockdown of isw-1 in rad-23 and csb-1 animals did not lead to significantly enhanced UV sensitivity compared to the respective controls , in both the L1 as well as the germ cell and embryo survival assay ( Figure 8A ) . Only a mild , but reproducible increase in UV sensitivity was observed in the germ cell and embryo sensitivity of rad-23 mutants in which isw-1 was knocked down . Most of these results , however , are in line with the idea that isw-1 has a general regulatory role in the UV-DDR but not specifically in either TCR or GGR . Next , we created double mutants for pbrm-1 and rad-23 or csb-1 and compared their UV sensitivity to respective controls ( Figure 8B ) . This showed L1 larvae UV sensitivity of pbrm-1; rad-23 double mutants was comparable to rad-23 single mutants and less severe to that of pbrm-1 single mutants . Unexpectedly , L1 larvae UV sensitivity of pbrm-1; csb-1 double mutants was enhanced compared to csb-1 and pbrm-1 single mutants . These results , which were reproduced in independent experiments , suggest in L1 larvae rad-23 is epistatic to pbrm-1 , while pbrm-1 and csb-1 act synergistically to protect against UV exposure . In germ cells and embryos no difference in UV sensitivity between pbrm-1; rad-23 and pbrm-1; csb-1 double mutants and their respective controls was observed . In conclusion , although our results clearly indicate a function for pbrm-1 , isw-1 and the other chromatin remodeling genes in the UV-DDR , their precise mode of action is still ambiguous and might not be simply confined to either TCR or GGR .
The genetic analysis presented in this paper strongly suggests that NER functions mechanistically similarly in the nematode C . elegans as it does in mammals . We and others [17]–[20] , [22] find that functional loss of core NER factors renders animals hypersensitive to UV light . Similar as in mammals , NER can be initiated by two distinct pathways , GGR and TCR , which depend on rad-23/xpc-1 and csb-1 , respectively . The clear difference between rad-23/xpc-1 and csb-1 UV sensitivities during development and the enhanced UV sensitivity in rad-23/xpc-1; csb-1 double mutants makes it unlikely that the RAD-23 and XPC-1 proteins are involved in both TCR and GGR . Therefore , C . elegans NER seems distinct from NER in budding yeast , where RAD23 and RAD4 ( yeast orthologs of hHR23 and XPC , respectively ) play a role in TCR as well [44] , [45] . Importantly , we observe that the involvement of GGR and TCR in C . elegans is developmentally regulated and differs between germ and somatic cells ( Table 1; Figure 9 ) . This developmental regulation was not noticed before in eukaryotes , but might be important for understanding the etiology of different mammalian syndromes associated with NER deficiencies . Following our analysis of the UV-DDR in C . elegans , we identify eight genes involved in ATP-dependent chromatin remodeling that function in the UV-DDR , depending on the developmental stage . Together , our data suggests C . elegans is a powerful model organism to study UV-induced DNA repair and to identify novel genes involved in this process . We provide evidence that in germ cells , oocytes and early embryo's GGR is the main DNA repair pathway conveying UV resistance . Our analyses of UV-survival , CPD repair , pachytene cell apoptosis and pachytene stage exit all indicate that rad-23 and xpc-1 are necessary and sufficient for germ cells to overcome the effects of UV irradiation . However , it is not exactly clear how UV irradiation of germ cells leads to the embryonic death that is measured in the germ cell and embryo survival assay ( Figure 2 and S1A ) . It is tempting to speculate that the lack of UV-induced apoptosis and defective pachytene stage exit leads to embryonic death . However , animals lacking the C . elegans p53 ortholog also show no UV-induced apoptosis , but have wild type levels of embryonic UV survival [22] . Furthermore , animals carrying a gain-of-function mutation ( n1950 ) in the core cell death pathway gene ced-9 , also do not show radiation-induced apoptosis [27] and do not show enhanced UV-induced embryonic lethality ( unpublished results ) . These observations indicate that lack of apoptosis and embryonic death are not necessarily linked . Our results confirm previous observations that in pachytene cells lacking functional XPA-1 apoptosis is not induced after irradiation [22] . This might imply that active NER is necessary to signal the presence of DNA damage to the apoptotic machinery , via the generation of NER-intermediates such as single stranded DNA [46] . Analysis of the rad-23 and csb-1 single and double mutants suggests that GGR or TCR alone is sufficient to induce apoptosis , although via TCR , e . g . in the rad-23 mutant , it seems to be slightly less efficient . Lack of functional GGR and TCR together inhibits induction of apoptosis . These results contrast the apoptotic response observed in cultured mammalian cells , which undergo increased apoptosis after irradiation when NER , and specifically TCR in differentiated cells , is impaired [47]–[49] . In these cells it is believed that persistence of damage in the transcribed strand of active genes triggers apoptosis . In undifferentiated mouse embryonic stem cells UV-irradiation induces apoptosis in NER-deficient XP-A cells but not in GGR-deficient XP-C cells [47] . Thus , it might be that in undifferentiated cells , similar to C . elegans germ cells , a trigger derived from GG-NER or a repair intermediate is necessary to set off an apoptotic response , contrary to the mainly TCR-driven apoptotic response of differentiated cells . An alternative explanation for the lack of apoptosis in NER deficient C . elegans germ cells might be that UV causes defects in cell cycle progression . Because of this , cells might not reach the late pachytene stage in which they can become apoptotic . Our results indicate that in C . elegans the involvement of GGR and TCR in survival of UV-induced DNA damage changes during development ( Figure 9 ) . A similar developmental change was described for the homologous recombination ( HR ) and non-homologous end-joining ( NHEJ ) repair pathways in C . elegans [50] . The error-free HR pathway is mainly active in germ cells and dividing somatic cells , while the error-prone NHEJ pathway becomes predominantly active in non-dividing somatic cells . This difference is probably to ensure that the genome integrity of germ cells and dividing cells is maintained , while genomic damage in non-dividing cells can be tolerated . Similarly , GGR may act in germ cells to ensure that the entire genome is free of lesions . TCR is only necessary to maintain active genes in non-dividing somatic cells . These findings exemplify the advantage of using C . elegans as in vivo tool to study the DNA repair response and are in line with similar observations in mammalian cells . Terminally differentiated human neurons appear to lose the ability to repair DNA lesions throughout the genome whereas they retain the ability to repair active genes [51] . Furthermore , in undifferentiated mouse embryonic stem cells the contribution of GGR to UV survival is larger than that of TCR , whereas in partially differentiated mouse embryonic fibroblasts the contribution of TCR is larger than that of GGR [47] . Although GGR is the major pathway contributing to survival in germ cells , we observed that TCR is also active but not essential for survival in these cells . Vice versa , in later developmental stages TCR is essential for survival , while GGR is also active but not essential for survival . The differences in activity of both repair pathways in later stages correlates to previous observations showing that in adult C . elegans highly transcribed and poorly transcribed genes are both repaired , although highly transcribed genes more efficient [52] . It is still unclear what causes the developmental switch from GGR to TCR . A possible mechanism might be that the switch occurs simultaneously with the onset of transcription in embryos , since TCR depends on transcription . However , transcription takes place in pachytene germ cells as well [29] and is initiated in the embryo already several hours before the csb-1 UV sensitivity becomes visible . A second mechanism might be that the switch is linked to proliferation , as the csb-1 UV sensitivity becomes visible when most cell divisions in the embryo have been completed [53] . However , oocytes , which depend on rad-23 , do not divide , while L1 larvae , in which cell division resumes , depend on csb-1 . A third mechanism might be the availability of RAD-23 and CSB-1 at the site of damage . Although both rad-23 and csb-1 appear to be expressed in all cells throughout development ( data not shown; [17] , [52] ) , there might be a delicate balance between RAD-23 and CSB-1 availability at the site of damage which is for instance influenced by chromatin-dependent accessibility of DNA . This hypothesis , however , does not correlate with the fact that the UV-DDR depends on rad-23 in all different cells of the germ line , while these cells differ significantly with regard to chromatin compaction . Finally , it might simply be that different processes are involved in survival and cell death when comparing germ cells to later stage somatic cells . Part of the UV sensitivity may result from direct interference of photolesions with vital processes such as transcription and replication . However , UV sensitivity may also be partially caused by extensive chromosomal aberrations which are caused by UV irradiation in C . elegans [54] . Germ cells might die from UV irradiation because global genome DNA damage , which is not repaired in a rad-23 genetic background , interferes with meiotic progression and early cell divisions . Later stage somatic cell types probably arrest due to block of transcription , which is persistent in a csb-1 genetic background [20] . Recent studies have highlighted the role of ( ATP-dependent ) chromatin remodeling in DNA repair , mainly focusing on the double-strand break response [35] , [36] . Using a dedicated genetic screen we identified eight genes implicated in chromatin remodeling whose involvement in the UV-DDR was unknown or at least ambiguous . Several lines of evidence suggest these genes genuinely function in the UV-DDR , instead of indirectly influencing UV survival because of their involvement in other processes such as transcription . First , inactivation of five genes caused UV hypersensitivity specifically in L1 larvae while inactivation of three other genes also caused germ cell hypersensitivity ( Table 1 ) . This specific difference between L1 and germ cell UV response would not be expected if UV hypersensitivity resulted indirectly from the impairment of other processes . Second , many other genes whose knockdown probably causes pleiotropic phenotypes ( see Table S1 ) were not found to be involved in the UV response . This also argues for a specific role of the eight identified chromatin remodeling genes in the UV-DDR . Finally , comparisons to literature and other DNA repair mechanisms suggest these genes might facilitate access of proteins to DNA or be involved in DNA damage signaling ( see discussion below ) . The mild UV hypersensitivity of the chromatin remodeling mutants , which contrasts the severe hypersensitivity of NER mutants , is in line with such a regulatory role . We identified four genes implicated in ISWI-dependent chromatin remodeling , isw-1 , hda-2 , hda-3 and him-1 . Mutation of him-1 was shown before to cause UV sensitivity [21] , while isw-1 , hda-2 and hda-3 were also identified in previous damage response screens [55] , [56] . The human isw-1 ortholog SMARCA5 is part of a chromatin remodeling complex that includes the hda-2/-3 ortholog HDAC1 and the cohesin subunit him-1 ortholog SMC1 [40] . Therefore , our findings suggest that an ISWI/cohesin complex involving these proteins is involved in the UV-DDR . However , since these proteins participate in several different other protein complexes , they might regulate the UV response independently of each other . This is suggested by the fact that isw-1 and him-1 loss-of-function causes sensitivity in germ cells , embryo's and L1 larvae , whereas hda-2 and hda-3 loss-of-function only affects L1 larvae . Alternatively , it could be that different ISWI/Cohesin complexes regulate different aspects of the UV-damage response that differ between germ cells and somatic cells and only involve hda-2/hda-3 in somatic cells ( Figure 9 ) . Several previous observations support a role for ISWI/Cohesin in the UV-DDR . For instance , the Drosophila ACF complex , containing the isw-1 ortholog ISWI , was found to facilitate NER in dinucleosomal DNA in vitro [57] . Furthermore , SMC is known to be phosphorylated following ionizing or UV irradiation and is thought to play a role in the S-phase checkpoint response in mammalian cells [58] , [59] . The evolutionary conserved function of ISWI/Cohesin activity in different DNA damage responses in different species suggests it is involved in one or more steps which are common among DNA damage pathways and possibly involve slightly different complexes: ( i ) ISWI and/or cohesin may function to mediate a DNA damage induced checkpoint response and ( ii ) ISWI and/or cohesin may function to facilitate efficient repair by altering chromatin structure . Follow-up functional studies will be required to explore the exact molecular role of ISWI/cohesin in the UV-DDR . Our analysis further implicated four genes involved in SWI/SNF mediated chromatin remodeling in the UV-DDR . pbrm-1 , psa-1 and snfc-5 , orthologs of human PBRM1 , SMARCC1 and SMARCB1 , respectively , only showed UV sensitivity when irradiated as L1 larvae , similar to hda-2 and hda-3 . Since the L1 larvae survival assay seems specific for TCR , this would suggest that these genes are specifically involved in TCR or a TCR-associated process ( Figure 9 ) . However , our genetic analysis of pbrm-1; rad-23 and pbrm-1; csb-1 double mutants suggests that pbrm-1 acts in parallel to csb-1 but not rad-23 in L1 larvae . To clarify these seemingly contradicting results , more detailed follow-up experiments to determine the precise function of pbrm-1 are necessary . psa-4 , ortholog of human BRM/SMARCA2 , showed also UV sensitivity in the germ line , indicating that it might have a more general role in the UV-DDR . Possibly , different ATP-dependent chromatin remodeling complexes play a role during TCR compared to GGR , or throughout development , while they may share some of the same subunits . In mammals , several different SWI/SNF-like complexes have been identified containing either BRM/SMARCA2 , the ortholog of psa-4 [42] , or BRG1 as ATPase subunit . Furthermore , involvement of other subunits such as SMARCC1 ( psa-1 ) , PBRM1 ( pbrm-1 ) and SMARCB1 ( snfc-5 ) also differs between different SWI/SNF complexes . SWI/SNF chromatin remodeling has been implicated in the UV-DDR before , but the exact mechanism by which it functions remain unknown . Mammalian cells lacking SMARCB1 or the BRM-paralog BRG1 are hypersensitive to UV irradiation , possibly because SWI/SNF functions in the checkpoint response [60] , [61] . Yeast SWI/SNF chromatin remodeling , on the other hand , was shown to stimulate excision repair in vitro and in cells , possibly because of rearrangement of chromatin at a damaged site to allow repair [62] , [63] . Therefore , it remains unclear whether SWI/SNF chromatin remodeling directly participates in the repair of a lesion or whether it modulates the checkpoint response , or whether it functions in both processes but involves complexes of different composition . We expect that the identification of specific SWI/SNF genes involved in the UV-DDR will lead to a better understanding of the role of SWI/SNF in the DNA repair mechanism . In summary , our analysis showed that C . elegans is especially well suited to genetically dissect genes and pathways involved in the UV-DDR at different stages of development . Based on the observed evolutionary conserved role of UV-DDR in C . elegans , it is expected that further analysis using the nematode will increase our understanding of how this response is organized in living organisms .
All strains were cultured according to standard methods [64] . Alleles used were csb-1 ( ok2335 ) , ercc-1 ( tm2073 ) , hda-2 ( ok1479 ) , hda-3 ( ok1991 ) , him-1 ( e879 ) , him-9 ( e1487 ) , isw-1 ( n3294 ) , isw-1 ( n3297 ) , pbrm-1 ( tm415 ) , psa-1 ( ku355 ) , psa-1 ( os22 ) , psa-4 ( os13 ) , rad-23 ( tm2595 ) , snfc-5 ( ok622 ) , xpa-1 ( ok698 ) , xpc-1 ( tm3886 ) and xpg-1 ( tm1670 ) . snfc-5 , xpa-1 , xpc-1 , ercc-1 , rad-23 and csb-1 mutants were backcrossed four times , pbrm-1 was backcrossed three times . Double mutants were genotyped using PCR ( primer sequences available upon request ) . RNAi bacteria were obtained from the Caenorhabditis elegans RNAi feeding library ( Geneservice ) . Control RNAi was vector pPD129 . 36 ( a gift from A . Fire ) . To measure UV sensitivity of germ cells and early embryos , staged young adults were washed and transferred to empty agar plates ( Figure S1A ) . Next , animals were irradiated at the indicated dose using two Philips TL-12 ( 40W ) tubes emitting UVB light , after which they were transferred to plates plated with OP50 bacteria . Following a 24 hr recovery period , animals were allowed to lay eggs for 2–3 hrs on fresh 6 cm plates containing food . In each experiment , for each dose 6 plates containing 3–5 adults per plate were used . The number of eggs laid was determined and 24 hours later the number of unhatched eggs , to calculate the survival percentage . To measure UV sensitivity of eggs or L1 larvae , eggs were collected from gravid adult animals by hypochlorite treatment and transferred to fresh plates seeded with HT115 ( DE3 ) bacteria ( Figure S1B and S1C ) . HT115 ( DE3 ) bacteria were specifically used because these bacteria form a uniform thin lawn on NGM plates , which increases reproducibility of the survival assay , as the thicker lawn formed by OP50 bacteria was found to partially shield C . elegans from UV irradiation . We did not observe any typical effects using HT115 ( DE3 ) bacteria related to the UV sensitivity of animals . To measure egg survival ( Figure 5D ) , animals were irradiated at indicated time points following hypochlorite treatment . The number of unhatched and hatched eggs was determined 24 hours later to calculate the survival percentage . To measure L1 larvae survival , animals were irradiated 16 hrs ( at 20°C ) after hypochlorite treatment . Animals that developed beyond the L2 stage ( survivors ) and animals that arrested development at the L1/L2 stage were counted to determine survival percentage . For experiments performed at 25°C ( Figure S3 ) , animals were cultured at 20°C and transferred to 25°C 45 minutes before irradiation . Hypochlorite treatment had no effect on survival rates ( data not shown ) and similar results were obtained by regular egg laying . Statistical analysis was performed using a one-way ANOVA test . To visualize CPD DNA damage , gonads were extruded by cutting the heads and tails of young adult animals using a fine gauge needle . Gonads were fixed in 3% paraformaldehyde , 0 . 1% Triton X-100 for 15 minutes , washed and permeabilized 2 times 10 minutes in PBS , 0 . 1% Triton X-100 . Next , gonads were incubated for 5 minutes in PBS , 0 . 07 M NaOH , to denature DNA . Gonads were then washed in PBS , 0 . 5% BSA , 0 . 15% glycin and incubated >2 hrs with CPD antibody ( Cosmo Bio Co . ) in PBS , 0 . 5% BSA , 0 . 15% glycin . Subsequently , animals were washed 2 times 10 minutes in PBS , 0 . 1% Triton X-100 and incubated >2 hrs with Alexa488 fluorescent secondary antibody ( Molecular Probes ) . Finally , animals were mounted on a glass slide using Vectashield with DAPI ( Vector laboratories ) . For DAPI staining , animals were fixed , permeabilized and mounted on a slide using Vectashield with DAPI . Images in Figure 3C and Figure 4 were acquired using a Zeiss LSM 510 META confocal microscope . Images in Figure 3B 1 were acquired using a Zeiss Axio Imager . Z1 and Nomarski optics . To determine germ line apoptosis , staged young adult animals were irradiated using 160 J/m2 UVB . Six hours later germ cell apoptosis was scored using Nomarski optics .
|
Nucleotide Excision Repair ( NER ) removes many forms of helix-distorting DNA damage which interfere with transcription and replication , including those induced by UV irradiation . NER is initiated when damage is sensed during transcription , i . e . Transcription-Coupled Repair ( TCR ) , or when damage is sensed in non-transcribed genomic sequences , i . e . Global Genome Repair ( GGR ) . Although the molecular mechanism of the core NER is known , it is not well understood how the UV response functions in living organisms and which additional mechanisms are involved to regulate its efficiency . Therefore , we exploited the small soil nematode C . elegans to study the UV response in a living organism . Using different NER–deficient animals , we found that in early development mainly GGR , but in later development mainly TCR is active in the UV response . Furthermore , we identified several new chromatin remodeling factors , whose involvement in the UV response also differs during development and which are thought to regulate efficiency of the UV response by altering chromatin structure . Our studies show that C . elegans is very well suited to genetically analyze the UV response during different developmental stages and in different tissues in a living animal .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"and",
"genomics/gene",
"discovery",
"developmental",
"biology",
"molecular",
"biology/dna",
"repair",
"cell",
"biology/cellular",
"death",
"and",
"stress",
"responses"
] |
2010
|
Involvement of Global Genome Repair, Transcription Coupled Repair, and Chromatin Remodeling in UV DNA Damage Response Changes during Development
|
Here we describe a chemical biology strategy performed in Staphylococcus aureus and Staphylococcus epidermidis to identify MnaA , a 2-epimerase that we demonstrate interconverts UDP-GlcNAc and UDP-ManNAc to modulate substrate levels of TarO and TarA wall teichoic acid ( WTA ) biosynthesis enzymes . Genetic inactivation of mnaA results in complete loss of WTA and dramatic in vitro β-lactam hypersensitivity in methicillin-resistant S . aureus ( MRSA ) and S . epidermidis ( MRSE ) . Likewise , the β-lactam antibiotic imipenem exhibits restored bactericidal activity against mnaA mutants in vitro and concomitant efficacy against 2-epimerase defective strains in a mouse thigh model of MRSA and MRSE infection . Interestingly , whereas MnaA serves as the sole 2-epimerase required for WTA biosynthesis in S . epidermidis , MnaA and Cap5P provide compensatory WTA functional roles in S . aureus . We also demonstrate that MnaA and other enzymes of WTA biosynthesis are required for biofilm formation in MRSA and MRSE . We further determine the 1 . 9Å crystal structure of S . aureus MnaA and identify critical residues for enzymatic dimerization , stability , and substrate binding . Finally , the natural product antibiotic tunicamycin is shown to physically bind MnaA and Cap5P and inhibit 2-epimerase activity , demonstrating that it inhibits a previously unanticipated step in WTA biosynthesis . In summary , MnaA serves as a new Staphylococcal antibiotic target with cognate inhibitors predicted to possess dual therapeutic benefit: as combination agents to restore β-lactam efficacy against MRSA and MRSE and as non-bioactive prophylactic agents to prevent Staphylococcal biofilm formation .
Staphylococcus aureus is a leading cause of hospital and community-acquired infections by Gram-positive bacteria [1–3] and Staphylococcus epidermidis has emerged as the most common cause of biofilm infections on medical implant devices [4] . In large part , the difficulty in treating these infections lies in their broad resistance to β-lactams , an otherwise powerful class of antibiotics that include methicillin , penicillin , cephalosporins and carbapenems such as imipenem [5] . Mechanistically , β-lactams are bactericidal agents that lyse cells by inhibiting penicillin binding proteins ( PBPs ) involved in peptidoglycan ( PG ) synthesis and cross-linking in the cell wall [5 , 6] . Methicillin-resistant strains of S . aureus ( MRSA ) and S . epidermidis ( MRSE ) , however , have acquired an exogenous PBP ( Pbp2a ) that exhibits low binding affinity to β-lactams , thus rendering such strains clinically resistant to nearly all β-lactams [5 , 7 , 8] . Staphylococcal drug resistance is further exacerbated by the pathogen’s propensity to form a biofilm , in which many bacterial cells display a “persister”-like state of low metabolic activity and which renders antibiotics inactive , such as β-lactams that target active metabolic processes including growth and cell division [9 , 10] . Biofilm formation also mediates antibiotic drug resistance by providing a complex and extensive polysaccharide extracellular matrix that serves as an effective physical barrier to antibiotic penetration into the cell [11–13] . Wall teichoic acid ( WTA ) is an anionic glycophosphate cell wall polymer in Gram-positive bacteria that is present in roughly equal amounts to PG [14] . Interestingly , WTA has important functional roles in both the tolerance of methicillin-resistant Staphylococci to β-lactams [15–19] as well as in biofilm formation [20–24] . WTA is synthesized using the lipid carrier bactoprenyl phosphate and a sequential series of cytosolic-exposed plasma membrane associated Tar ( teichoic acid ribitol ) enzymes , starting with TarO and TarA [19 , 22 , 25–27] ( Fig 1 ) . The polymer is subsequently translocated across the plasma membrane by an ABC transporter encoded by TarG and TarH [22 , 28 , 29] and ultimately cross-linked to the cell wall PG , upon which the liberated bactoprenyl carrier is recycled ( Fig 1 ) [22 , 25 , 30–32] . Interestingly , genetic studies in S . aureus and S . epidermidis reveal that whereas deletions of early WTA biosynthetic enzymes are nonlethal , but cause diverse attenuated virulence phenotypes [27 , 33 , 35 , 36] , deletions of later steps in WTA biosynthesis are not generally tolerated and the enzymes are normally essential for growth [28 , 37 , 38] . This is referred to as an ‘essential gene paradox’ , and may be explained either by 1 ) the accumulation of toxic WTA intermediates , or 2 ) sequestration of a non-recyclable pool of lipid carrier accumulating in late stage WTA deletion mutants such that bactoprenyl phosphate is unavailable to support PG biosynthesis ( Fig 1 ) [19 , 28 , 37–40] . While WTA is dispensable for growth amongst Gram-positive bacteria [28 , 35 , 37 , 38] , it buffers methicillin-resistant Staphylococci from the action of β-lactam antibiotics [16 , 17 , 18 , 33 , 41] by coordinating peptidoglycan cross-linking [42] and targeting the major autolysin Atl [43] . Accordingly , genetic or chemical inhibition of Tar enzymes restores the susceptibility of MRSA and MRSE to β-lactams . Inhibitors to early ( non-essential ) enzymes in WTA biosynthesis are particularly appealing as non-bioactive adjuvants or combination agents that , paired with β-lactams , provide a promising strategy to treat MRSA and MRSE infections [16 , 17 , 33 , 40 , 41 , 44 , 45] . A growing number of small molecules targeting Tar enzymes have also been identified [16 , 33 , 34 , 39 , 41 , 45 , 46] . Perhaps best known is tunicamycin , a natural product structurally related to UDP-N-acetylglucosamine ( UDP-GlcNAc ) , which inhibits TarO , the first enzyme in WTA biosynthesis [16 , 47] . Tunicamycin demonstrates strong synergistic activity in combination with β-lactam antibiotics , presumably by depleting the buffering capacity WTA provides in β-lactam resistance of MRSA and MRSE . A variety of additional WTA inhibitors have also been demonstrated to target TarG , the membrane-associated subunit of the WTA transporter [33 , 34 , 40 , 48] . WTA biosynthetic enzymes have been extensively characterized in S . aureus [19 , 22 , 25–27 , 49 , 50] . However , the identity and characterization of the 2-epimerase which interconverts UDP-GlcNAc and UDP-N-acetylmannosamine ( UDP-ManNAc ) , each a substrate of TarO and TarA respectively ( Fig 1 ) , has remained largely restricted to Bacillus subtilis [51] . Two proteins , Cap5P and MnaA , share homology to the B . subtilus 2-epimerase [51] and have been suggested to potentially perform this function in S . aureus [52] . Cap5P and MnaA are 59 . 6% identical and 77 . 2% similar in their amino acid sequence , and each has been demonstrated to complement the phenotype of an E . coli strain lacking a 2-epimerase [52] . In addition , S . aureus Cap5P epimerizes ~10% of UDP-GlcNAc to UDP-ManNAc in vitro , which is comparable to the conversion levels observed for the E . coli and B . subtilis homologs [51 , 52] . Disruption of S . aureus cap5P , however , did not yield an observable phenotype [52] , implying that Cap5P and MnaA may share redundant functions associated with capsule and/or WTA biogenesis . Herein we demonstrate that MnaA functions as the previously uncharacterized 2-epimerase that interconverts UDP-GlcNAc and UDP-ManNAc , thus providing the corresponding substrates of TarO and TarA in both S . aureus and S . epidermidis . Genetic evidence is provided demonstrating that MnaA is essential for WTA production and β-lactam resistance in MRSA and MRSE . Likewise , MnaA loss of function ( LOF ) mutants display restored susceptibility to β-lactam antibiotics in a mouse MRSA and MRSE thigh infection model . Whereas MnaA serves as the sole WTA 2-epimerase in S . epidermidis , MnaA and Cap5P provide overlapping roles in S . aureus WTA biosynthesis . We also demonstrate that MnaA is required for biofilm formation by methicillin-resistant Staphylococci , thus contributing to dual mechanisms of β-lactam resistance . We have determined the 1 . 9Å resolution crystal structure of S . aureus MnaA protein and describe critical residues for enzymatic dimerization , stability , and substrate binding . Finally , we demonstrate that tunicamycin , a known non-competitive inhibitor of TarO , also inhibits MnaA activity in vitro and discuss the potential therapeutic implications of WTA 2-epimerase inhibitors from the perspective of anti-Staphylococcal β-lactam combination agents .
Late steps of WTA biosynthesis are conditionally essential in S . aureus and S . epidermidis; genetic deletion or chemical inhibition of late WTA biosynthesis enzymes abolishes growth but can be tolerated provided early steps of WTA biosynthesis are also inactivated [28 , 33 , 37 , 38 , 40] . Accordingly , LOF mutations in early non-essential steps in WTA biosynthesis , such as TarO and TarA , act as bypass suppressors of late stage WTA inhibitors [16 , 33 , 34 , 41] . To explore whether additional yet previously uncharacterized genes participate in early aspects of WTA biosynthesis , we used the previously published TarG inhibitor , L638 [33] , as a chemical probe to screen for novel bypass suppressor mutations . Extensive L638-resistant ( L638R ) mutant selections were performed in both MRSA COL and MRSE CLB26329 strains . As expected , multiple independently derived missense mutations mapping either to tarG , or LOF mutations mapping to tarO and tarA were identified in both strain backgrounds following whole genome sequencing ( WGS ) as previously reported [33] . Interestingly , in S . epidermidis , multiple ( n = 9 ) independently derived L638R mutations specifically isolated in this subsequent screen mapped to mnaA , encoding a putative UDP-GlcNAc:UDP-ManNAc 2-epimerase [52] not previously implicated as a suppressor of defects in late stage WTA biosynthesis in Staphylococci ( Fig 2A ) . As WGS analysis indicates that each resistor isolate contains no additional non-synonymous mutations in their genome , we presumed mnaA mutations are causal for the L638R phenotype observed . Unlike L638R tarG mutations which are exclusively missense mutations conferring drug resistant amino acid substitutions to the target protein [33] , L638R mnaA mutations encompass nonsense , frameshift , and missense mutations ( Fig 2A ) , therefore implying drug resistance is likely achieved by LOF mutations that possibly impair WTA biosynthesis . Finally , as TarO and TarA , respectively , require UDP-GlcNAc and UDP-ManNAc as substrates for initiating WTA polymer synthesis and an ortholog of MnaA was described to participate in B . subtilis WTA polymer synthesis [51] we investigated the functional role of MnaA in methicillin-resistant Staphylococci . To directly evaluate the consequence of these mnaA mutations , WTA of the corresponding mutants was extracted and polymer levels visualized on an alcian blue-silver stained SDS PAGE gel ( Fig 3A and S1 Fig ) . As predicted , all MRSE mnaA mutants are completely depleted of WTA ( Fig 3A and S1 Fig , right ) , mirroring tarOSeG84* and tarASeG129R LOF mutants ( Fig 3A; [33] ) . Importantly , like previously described MRSE tarO and tarA LOF mutants [33] , all mnaA mutants display restored susceptibility to diverse β-lactams , with their minimal inhibitory concentration ( MIC ) below the clinical breakpoint defined for resistance to these agents ( Table 1 and Table A in S1 Text ) . MRSE mnaA LOF mutants are up to 1000-fold more sensitive to imipenem , 256- to 512-fold more sensitive to nafcillin and 512-fold more sensitive to dicloxacillin compared to the parental MRSE strain ( Table 1 and Table A in S1 Text ) . Notably , this dramatic antibiotic sensitization is specific to β-lactams ( Table A in S1 Text ) . Representative mnaA and Δcap5P mutations in MRSA and MRSE as well as complementation strains thereof are shown . Minimum inhibitory concentrations ( MIC; μg ml-1 ) of β-lactams imipenem ( IPM ) , nafcillin ( Naf ) , and dicloxacillin ( Dic ) are provided . L638 is included to quantify drug resistance of bypass mutations . Analogous L638R mutant selections performed in MRSA COL were unsuccessful in identifying mnaA LOF mutants . Unlike S . epidermidis , however , S . aureus maintains a second 2-epimerase involved in serotype 5 capsular polysaccharide ( CP5 ) synthesis , Cap5P ( S2A Fig ) [52] . To determine whether L638R mnaA LOF mutants were not identified in MRSA COL due to a functional redundancy between Cap5P and MnaA , a cap5P deletion mutant was constructed ( S3 Fig ) and the L638R studies were repeated . Under these conditions , in addition to identifying the expected tarG L638R mutations as well as tarO and tarA LOF mutations , multiple ( n = 11 ) independent resistor isolates obtained uniquely possess distinct mutations that map to mnaA and are predicted to inactivate gene function as well as directly confer L638R drug resistance based on the absence of additional non-synonymous mutations in their genome following WGS analysis ( Fig 2B ) . While MRSA COL Δcap5P exhibits no WTA depletion phenotype and remains resistant to β-lactams , MRSA COL mnaA , Δcap5P double mutants are completely devoid of WTA and are also highly sensitive to β-lactams ( Fig 3B and Table 1 and Table A in S1 Text ) , again mirroring the restored β-lactam susceptibility of tarO and tarA deletion mutants [16 , 33 , 41] . Indeed , MRSA COL mnaA Δcap5P double mutants are 32- to 64-fold more sensitive to imipenem , 8- to 16-fold more sensitive to nafcillin , and 16- to 32-fold more sensitive to dicloxacillin compared to either Δcap5P or the isogenic parental strain ( Table 1 and Table A in S1 Text ) . Consistent with the functional role of MnaA in WTA biogenesis , MRSA COL mnaA , cap5P double mutants and MRSE CLB26329 mnaA single mutants display related growth and morphological defects as observed for S . aureus ΔtarO and ΔtarA mutants . For example , in both MRSA and MRSE strains examined , genetic inactivation of MnaA /Cap5P function led to a slightly reduced growth rate within the first 6 h of growth in fresh medium but no apparent difference in cell density versus the wild-type control over a 24 h extended growth period ( S4 Fig ) . Similarly , super resolution microscopy analysis of MRSA COL mnaA , cap5P double mutants and MRSE mnaA single mutants revealed morphological phenotypes consistent with WTA depletion [16] , including increased cell size heterogeneity and septation defects ( S5 and S6 Figs ) . Genetic complementation studies further demonstrate the overlapping functional activity of MnaA and Cap5P in Staphylococci . Complementing Δcap5P mnaASaP12L and Δcap5P mnaASaY194* with either cap5P or mnaASa reintroduced on an inducible plasmid restored WTA polymer levels , resistance to each of the β-lactams tested , and wild-type sensitivity to L638 ( Fig 3B and Table 1 ) . Interestingly , cross complementation of these mutants with mnaASe also restored WTA production , albeit only partially restored wild-type drug susceptibilities ( Fig 3B and Table 1 ) . Similarly , mnaASeΔ151 and mnaASeG171D were also fully complemented for each of the above phenotypes by reintroducing a wild-type plasmid-based copy of mnaASe ( Fig 3A and Table 1 ) . Strikingly , introduction of either mnaASa or cap5P fully restored WTA production , β-lactam resistance , and L638 susceptibility of mnaASe LOF mutants ( Fig 3A and Table 1 ) . To further investigate β-lactam susceptibility phenotypes associated with mnaA inactivation , kill curve experiments were performed against MRSA and MRSE strains treated with the β-lactam imipenem . Whereas imipenem ( 4 μg ml-1 ) is ineffective in inhibiting growth of wild-type methicillin-resistant Staphylococci , imipenem displayed a dramatically restored bactericidal activity against MRSA Δcap5P mnaASaP12L as well as MRSE mnaASeΔ151 strains , leading to a 3 log reduction in viable cells within 7 hr of drug treatment ( S7 Fig ) . Similar to other phenotypes examined , full complementation as well as heterologous complementation between mnaA orthologs were again observed ( S7 Fig ) . Collectively , these data demonstrate that whereas MnaA seems to be one of two redundant UDP-GlcNAc:UDP-ManNAc 2-epimerases in MRSA COL , it is the sole 2-epimerase required for WTA biosynthesis in MRSE CLB26329 . To evaluate the significance of the observed in vitro hypersensitivity of mnaA , cap5P double mutants to β-lactam antibiotics in an in vivo context , MRSA COL Δcap5P mnaASaP12L , Δcap5P mnaASaY194* , and Δcap5P mnaASaD281Y strains were used to conduct imipenem efficacy studies in a previously described murine deep thigh model of infection [53] . Imipenem is ineffective at treating animals infected with wild-type MRSA COL or the Δcap5P mutant when dosed three times daily ( TID ) with 10 mg kg-1 imipenem over 24 hours ( Fig 4A ) [33] . Conversely , imipenem efficacy is significantly restored against MRSA in Δcap5P mnaASaP12L , Δcap5P mnaASaY194* , and Δcap5P mnaASaD281Y mutants , ranging between a 2–3 log reduction of bacterial burden versus control strains after imipenem treatment ( Fig 4A ) . As MRSE displays somewhat greater sensitivity to imipenem in our infection model , a lower dose ( 2 . 5 mg kg-1 ) was required to demonstrate restored efficacy of imipenem against the mnaASe mutant . Here again , mice administered imipenem ( TID ) at this non-efficacious dose and infected with the mnaASe mutant possessed a significantly reduced ( > 3 log ) bacterial burden versus the wild-type MRSE parent , similar to the effects of tarOSeG84* and tarASeG129R mutants ( Fig 4B ) . Since deletion of tarO has been shown to be important for biofilm formation and attachment [20–24] , we evaluated the role of mnaA , cap5P , and other WTA biosynthesis genes in this process . MRSA COL strains with LOF in early ( tarA , tarO ) and intermediate ( tarB , tarD , tarI’ ) steps in WTA biosynthesis [33] were all substantially defective in biofilm formation ( Fig 5A ) . Conversely , Δpbp3 and Δpbp4 single mutants as well as the Δpbp3 Δpbp4 double mutant control strains faithfully produced biofilms indistinguishable from the wild-type MRSA COL parent ( Fig 5A ) . MRSE strains deleted of tarOSeG84* or tarASeG129R also failed to form robust biofilms ( Fig 5B ) . Paralleling this WTA-mediated role in biofilm formation and attachment , mnaASe mutants and Δcap5P mnaASa mutants similarly displayed impaired biofilm formation . MRSA COL Δcap5P mutants , however , failed to impair biofilm formation ( Fig 5A ) , consistent with its lack of phenotypes related to WTA biogenesis , β-lactam susceptibility , and virulence . Fluorescence microscopy on stained , similarly grown , and treated biofilms confirmed these phenotypes ( Fig 5C and 5D and S8 and S9 Figs ) . Genetic complementation of the biofilm impairment observed in Δcap5P mnaASaP12L and Δcap5P mnaASaY194* is fully achieved by reintroducing either wild-type S . aureus gene and partially achieved by S . epidermidis mnaA ( Fig 5A and 5C and S8 Fig ) . Similarly , impaired biofilm formation of mnaASeΔY151 was faithfully complemented by reintroduction of mnaA as well as S . aureus mnaA or cap5P ( Fig 5B and 5D and S9 Fig ) , again reiterating a strong functional overlap between these 2-epimerases . To test whether known inhibitors of WTA biogenesis similarly disrupt biofilm formation , MRSA and MRSE strains were grown as above and treated with sub-MIC concentrations of tunicamycin or L638 . Tunicamycin treatment at levels shown to completely inhibit WTA production [17 , 35] decreased biofilm formation to amounts similar to those achieved by genetic inactivation of its target , TarO ( Fig 5B , 5C and 5E and S8 and S9 Figs ) . Similarly , L638 treatment at sub-MIC levels that do not dramatically affect growth produce a dose-dependent inhibition of biofilm formation ( Fig 5A , 5B and 5C and S8 and S9 Figs ) . Conversely , neither tunicamycin nor L638 similarly tested singly or in combination with a sub-MIC level of imipenem significantly disrupted the gross morphology , adherence , viability or antibiotic susceptibility of pre-existing biofilms ( S10 Fig ) . Therefore , inhibition of WTA synthesis can prevent the establishment of a biofilm growth state , presumably by disrupting the early attachment step in biofilm colonization , but does not significantly impair biofilm viability or disrupt the extracellular matrix of pre-existing biofilms . Lipoteichoic acid ( LTA ) , another cell surface teichoic acid common to Gram-positive bacteria , has also been reported to play a role in biofilm formation [54 , 55] and co-depletion of WTA and LTA demonstrate a synthetic lethal genetic interaction in both B . subtilis [56] and S . aureus [57] . Accordingly , we tested whether depletion of both WTA and LTA synergistically impair biofilm formation . Since LTA is essential [58] , a previously described partial LOF ltaSSa mutant that produces lower levels of LTA than the parental strain [59] was tested for biofilm formation both in the absence and presence of increasing tunicamycin concentrations . Whereas the ltaS defective strain exhibits a slight 2-fold reduction in biofilm formation , treatment with sub-MIC levels of tunicamycin produces a dose dependent further reduction in biofilm formation approaching that of tarSa mutants ( Fig 5E ) . Interestingly , this baseline level of residual biofilm formation in the ltaSSa mutant background was achieved with ~10 percent the normal concentration of tunicamcyin required to similarly impair biofilm in the wild-type parent strain ( Fig 5E ) . Such an apparent synergistic effect further suggests a functional interdependence between these teichoic acid biosynthetic pathways and biofilm formation . The MRSA COL MnaA crystal structure was solved at 1 . 9Å resolution ( Fig 6A ) . The protein crystallizes with a dimer in the asymmetric unit . The structure is closely related to that of other bacterial 2-epimerases ( root-mean-square ( RMS ) deviation differences with E . coli MnaA in Cα positions for all atoms in a monomer between 1 . 40 and 2 . 90Å depending on the chains being compared ) , and a similar dimerization interface and quaternary structure is observed ( Fig 6B ) . While there are significant structural differences between the E . coli and S . aureus models , they are located on the protein surface , away from either the substrate binding site or the oligomerization interface . Surprisingly , there are differences between the two monomers in the S . aureus MnaA structure: the RMS . deviation in positions for all Cα is about 1 . 0Å . We refer in the following the structure of the ternary complex between MnaA , UDP and UDP-GlcNAc as the “closed” form [60] , and other states of the protein , either in apo form or a binary complex with UDP , as the “opened” form , consistent with past structural characterization of the enzymes [60] . This difference between monomers is comparable to the differences with the structure of the E . coli enzyme ( PDB entry 1F6D ) in opened form , 1 . 0 to 1 . 9Å , depending on which chains are compared . However , further comparisons show that the differences with the closed form are less when comparing to one of the monomers in the S . aureus crystal structure rather than the other ( RMS deviation of 1 . 2Å for 354 atoms in the superposition with B . anthracis MnaA in complex with UDP and UDP-GlcNAc ( closed form ) , PDB entry 3BOV , versus 1 . 4Å for 340 atoms for the other monomer ( See also Fig 6 ) ) . In addition to differences in the quaternary structure , other significant local structural rearrangements distinguish the two monomers . Notably , His 205 to Gly 211 ( E . coli His 213 to Gly 219 , B . anthracis His 209 to Gly 215 ) differ significantly , but in the monomer nearer to the closed form adopts a conformer similar to the one observed in the closed form [61] . By contrast , the same loop in the other monomer has a local fold similar to the one found in the M . jannaschii epimerase in apo form ( PDB entry 3NEQ ) . Collectively , crystallographic results described here allow for a more refined understanding of the enzyme regulation at a structural level: a dynamic equilibrium between the opened form and an “intermediate closed” conformer of the enzyme is present in solution in apo form or in presence of UDP only . The equilibrium is moved and locked towards the “closed” form in the ternary complex with UDP—UDP-GlcNAc . L638R bypass mutants corresponding to MnaA LOF mutants were mapped to the MRSA COL MnaA crystal structure ( Fig 6C and 6D ) . Among eight MnaA LOF mutations isolated in MRSE ( Fig 2A ) , only the Gly283/Arg and Pro131/Leu mutations are located at the ligand ( UDP and UDP-GlcNAc ) binding sites . Gly283/Arg is positioned at an area across both the UDP and UDP-GlcNAc binding sites ( UDP-GlcNAc binding site was mapped from the structure of UDP-GlcNAc bound Bacillus anthracis 2-epimerase ( PDB ID 3BEO ) through structure overlay ) as shown in the X-ray structure of MnaA ( Fig 6B ) . The large side-chain of Arg residue may cause van der Waals ( VDW ) clashes with the ligands and surrounding residues , thus interfere with binding of substrate UDP-GlcNAc and intermediate UDP , and destabilize the protein . The Pro131/Leu mutation is adjacent to the UDP-GlcNAc binding site and close to the dimer interface; it could both affect substrate binding and dimer stability through VDW conflicts by the Leu side-chain . Mutations Pro129/Thr , Gly171/Asp , Asp281/Glu and Arg300/Ile introduce amino acids with bulkier side-chains , compromise the favorable hydrogen bonds and hydrophobic interactions around the wild-type residues , and thus decrease protein stability . Protein stability could also be dramatically reduced by the deletion at Lys151 and the insertion at His76 , the latter of which is located right in the middle of a helix , which is packed against the other monomer at the dimer interface . Mapping of MnaA LOF mutations isolated in MRSA COL Δcap5P ( Fig 2B ) into the S . aureus MnaA structure shows that all the sequence changes are in close proximity to the substrate binding site or the UDP-GlcNAc binding site ( Fig 6D ) , therefore rendering the enzyme inactive . Tunicamycin targets multiple UDP-GlcNAc binding enzymes . At higher drug concentrations tunicamycin binds MraY , a UDP-N-acetylmuramoyl-pentapeptide: undecaprenyl-phosphate phospho-N-acetylmuramoyl-pentapeptide transferase enzyme involved in peptidoglycan synthesis [62] . However , at low drug concentrations , tunicamycin selectively inhibits TarO [34 , 47] . Considering MnaA and Cap5P are epimerases responsible for interconverting UDP-GlcNAc and UDP-ManNAc and that TarO utilizes the same substrate , we tested whether tunicamycin may also bind MnaA and Cap5P . To test this possibility , we performed saturation transfer difference ( STD ) nuclear magnetic resonance ( NMR ) studies , which allow for the detection of transient binding of small molecules to proteins [63] . Such studies using 15 μM tunicamycin in the presence or absence of 5 μM of S . aureus MnaA or Cap5P protein revealed binding of tunicamycin to both MnaA and Cap5P , as evidenced by the tunicamycin specific peaks appearing only when run in the presence of 2-epimerases ( Fig 7A , 7C , 7D and 7F ) . Binding of tunicamycin indicated that MnaA may represent an additional target of the nucleoside antibiotic in WTA biosynthesis beyond TarO . Functional reconstitution of the MnaA-catalyzed reaction in vitro followed by capillary electrophoresis ( CE ) analysis with UV detection showed the interconversion of UDP-GlcNAc and UDP-ManNAc , confirming 2-epimerase activity ( S11 Fig ) . Enzyme kinetics ( Michaelis-Menten constant , Km , and maximal velocity , Vmax ) were determined for both , forward and reverse reaction . The conversion of UDP-GlcNAc to UDP-ManNAc ( forward reaction ) was in the linear range ( steady-state phase ) for up to 75 min ( S12A Fig ) , while the reverse reaction exhibited a lag-period of 50 min after reaction initiation and reached equilibrium after 180 min ( S12B Fig ) . MnaA displayed a Km value of 411 ± 57 μM for UDP-GlcNAc and a Vmax value of 0 . 171 ± 0 . 037 μmol/min/mg protein . A Km value of 131 ± 21 μM for UDP-ManNAc and a Vmax value of 0 . 159 ± 0 . 021 μmol/min/mg protein were determined for the reverse reaction ( S12B Fig ) . The reversible reaction attained an equilibrium ratio of 9:1 in favor of UDP-GlcNAc , in line with reported epimerization ratios ranging from 12:1 to 9:1 for homologous enzymes [52 , 64 , ] including the orthologous MnaA 2-epimerase required for B . subtilis WTA biosynthesis [51] . Testing tunicamycin in the in vitro system revealed a dose-dependent inhibition of MnaA ( Fig 8 ) , verifying that the 2-epimerase indeed represents a secondary target within the WTA biosynthesis pathway .
Here , we describe genetic , biochemical , and X-ray crystal structure studies revealing the functional role of MnaA and Cap5P , encoding 2-epimerases which interconvert UDP-GlcNAc and UDP-ManNAc and provide the requisite substrate for the two first enzymes involved in Staphylococcal WTA biosynthesis , TarO and TarA , respectively . Whereas most of the enzymes involved in WTA polymer synthesis have been extensively characterized , the role of 2-epimerases in this process has remained largely enigmatic amongst medically relevant Staphylococci . Presumably , this is due to the genetic redundancy between MnaA and Cap5P in S . aureus and the limited studies of WTA biogenesis performed in S . epidermidis , where its identification is uniquely amenable by genetic means . To identify the functional contribution of MnaA in Staphylococcal WTA biosynthesis , L638 , a recently discovered WTA inhibitor with potent S . epidermidis activity was used as a chemical probe to screen for novel bypass suppressor mutations able to reverse the drug’s bacteriostatic effect [33] . Extensive genetic and chemical biology evidence predict that in addition to target-based drug resistant mutations , additional bypass mutations may arise and reflect gene inactivation mutations in early non-essential steps in WTA biosynthesis [33 , 34] . Accordingly , bypass mutations in tarO and tarA as well as mnaA were uncovered by L638R suppressor analysis and WGS of resistor isolates . Indeed , an extensive characterization of mnaA and cap5P mutant phenotypes in both MRSA and MRSE described here reveal that WTA 2-epimerases serve as a new and highly unconventional class of antibiotic drug targets . Unlike traditional antibiotic drug targets , MnaA and other early stage WTA enzymes are not essential for cell growth or viability . In fact , genetic inactivation of MnaA 2-epimerase activity resulted in only a minimal effect on Staphylococcal growth rate . However , mnaA mutant phenotypes faithfully recapitulate those of tarO and tarA mutants and reveal multiple therapeutic contexts in which a cognate inhibitor to MnaA could provide broad efficacy against methicillin-resistant Staphylococci . Firstly , we demonstrate these 2-epimerases are essential for WTA synthesis in both MRSA and MRSE and depletion of WTA dramatically restores β-lactam susceptibility to these drug resistant pathogens both in vitro as well as in relevant mouse infection models . Therefore , as we and others have proposed [8 , 16 , 17 , 33 , 45 , 65–67] , inhibitors to such β-lactam potentiation targets could serve as novel adjuvants to partner with existing β-lactams to restore bactericidal therapeutic activity against β-lactam resistant Staphylococci . We also provide extensive evidence that abolishing WTA biosynthesis renders methicillin-resistant Staphylococci unable to effectively form robust biofilms . Accordingly , inhibitors of early stage WTA biosynthetic enzymes , including MnaA , may also serve as prophylactic agents to prevent Staphylococcal biofilm formation . As inhibitors of any of these targets are not expected to display antibacterial activity , such prophylactic agents are also predicted to be highly selective and spare the gut microbiota from antibiotic-mediated alterations . Finally , as ΔtarO strains exhibit dramatically attenuated virulence phenotypes across diverse animal infection models tested [35 , 55 , 68] , WTA 2-epimerase inhibitors may also provide prophylactic or therapeutic utility as novel anti-virulence agents . Surprisingly , S . epidermidis encodes only a single epimerase whereas S . aureus encodes two related enzymes amongst all published genomes we have examined . This likely reflects that S . aureus expresses a second epimerase involved in capsular biosynthesis , which is not produced by S . epidermidis . Indeed , S . aureus Cap5P epimerizes UDP-GlcNAc and UDP-ManNAc and participates in CP5 synthesis [52 , 69] . Therefore , whereas S . epidermidis MnaA appears solely responsible for WTA synthesis , S . aureus requires a second enzyme to fulfill the biosynthetic needs of two disparate cell wall polymers . Interestingly , B . anthracis also requires two highly related 2-epimerases to fulfill biogenesis of the S-Layer , a cell wall elaboration analogous to WTA in many ways , and one member ( GneZ ) is essential for vegetative growth [70] . Biochemical analysis of MnaA revealed that the 2-epimerase interconverts UDP-GlcNAc and UDP-ManNAc , demonstrating reversible epimerase activity . At equilibrium the conversion of UDP-GlcNAc to UDP-ManNAc attained ~10% , thereby limiting the available amount of UDP-ManNAc for WTA biosynthesis . This agrees well with the required prioritization of UDP-GlcNAc for the essential processes of peptidoglycan biosynthesis within the cell . MnaA may thus resemble a checkpoint that contributes to control the flux of UDP-GlcNAc and channel the shared soluble cell wall precursor into the different synthesis pathways . As TarO , MnaA and Cap5P all bind a common substrate , UDP-GlcNAc , we examined whether the 2-epimerases potentially serve as additional WTA targets of tunicamycin . Tunicamycin is a natural product-derived antibiotic that is structurally related to UDP-GlcNAc and can bind in the active sites of TarO [47] and MraY [16 , 62] . We demonstrate that tunicamycin also binds to purified MnaA and Cap5P by STD NMR and inhibits the MnaA catalyzed interconversion of UDP-GlcNAc and UDP-ManNAc in a dose-dependent fashion in vitro . These findings may explain the tremendous potency of tunicamycin as a WTA inhibitor versus a PG inhibitor . Whereas tunicamycin displays a relatively low activity against S . aureus ( MIC = 32 μg ml-1 ) , it is a highly potent WTA inhibitor ( IC50 = 50 ng ml-1 ) [16] . We speculate this preferential inhibitory activity of WTA over PG biosynthesis may in part reflect that tunicamycin inhibits both TarO and MnaA/Cap5P-mediated steps in WTA biosynthesis and that even partial inhibition of each enzyme in the pathway may be synergistic in a whole cell context , analogous to the synergistic mechanism of combining trimethoprim ( TMP ) and sulfonamide ( SULF ) antibiotics [5] . Unlike the chemical synergy achieved by TMP and SULF , however , tunicamycin may achieve this effect due to its polypharmacological effects as a single agent . Notably , a single agent antibiotic with multiple targets is predicted to possess a very low propensity for drug resistance , which is true for tunicamycin . Therefore , a non-toxic analog of tunicamycin , alternative active site inhibitor , or allosteric inhibitor of TarO and MnaA/Cap5P may all benefit by displaying potent WTA inhibitory activity as well as an extremely low frequency of resistance . Finally , due to the limited homology between MnaA and the closest human BLASTP homolog , GNE2 ( S2B Fig ) , it is unlikely that host 2-epimerases would be affected . Based on the crystal structure of the B . anthracis UDP-GlcNAc 2-epimerase , an in silico screen was recently performed to identify a UDP-GlcNAc 2-epimerase inhibitor named epimerox [71 , 72] . Although several chemotypes of epimerox are reported to recapitulate terminal phenotypes of the B . anthracis epimerase conditional mutant , no direct genetic , biochemical , biophysical , or structural data are provided to independently reinforce this conclusion . Surprisingly , epimerox displays potent S . aureus activity ( MIC = 8 μg ml-1 ) [71] and S . epidermidis activity ( MIC = 2 μM ) [72] despite our conclusion that 2-epimerase activity is dispensable for growth in MRSA COL , MRSE CLB26329 , and the routinely studied methicillin-sensitive S . aureus strain , RN4220 ( S13 Fig ) . Such a paradox between the bioactivity of epimerox and non-essentiality of its reported drug target in multiple different strain backgrounds suggest that additional mechanism of action studies seem warranted , including whether epimerox effectively inhibits WTA synthesis , synergizes in combination with β-lactams , and/or prevents biofilm formation or Staphylococcal virulence . Determining the S . aureus MnaA crystal structure as well as key residues essential for enzyme function offers important new resources to assist MnaA inhibitor discovery .
MRSA COL is a hospital-acquired penicillinase-negative strain extensively used in Staphylococcus aureus methicillin resistance and virulence studies [73 , 74] and from which its genome has been fully sequenced and annotated [75] . MRSE strain ( MB6255 ) is a previously described methicillin-resistant S . epidermidis clinical isolate ( CLB26329; [76] ) isolated from a New York ICU in 2004 . All strains were grown in trypticase soy broth ( TSB ) or cation-adjusted Mueller Hinton broth ( CAMHB ) ( Difco , BD , Franklin Lakes , New Jersey , USA ) at 37°C , 250rpm unless otherwise indicated . All compounds were prepared in DMSO . All strains are described in Table C in S1 Text . All subcloning methods are described in S1 Text . Approximately 1 x 109 cells of strains MRSE CLB26329 , MRSA COL or Δcap5P grown to stationary phase overnight were spread on CAMHA ( Difco ) containing 16 μg ml-1 L638 for MRSE ( 4-fold MIC ) and 8 μg ml-1 L638 ( 4-fold MIC ) for MRSA and Δcap5P . Plates were incubated for 48–96 hours for MRSE and 48–72 hours for MRSA and Δcap5P . L638 resistance was confirmed in a second round of growth on 16 μg ml-1 L638 , and colonies were counter screened against 8 μg ml-1 imipenem to differentiate mutations in TarG versus early and intermediate steps in WTA biosynthesis . Genomic DNA was prepared from imipenem sensitive mutants ( DNEasy Blood & Tissue Kit , Qiagen , Venlo , Netherlands ) and Sanger sequencing for mnaA was performed using mnaA-locus specific primers 1731 , 1732 , 1733 and 1734 for MRSE and primers 1525 and 1526 for MRSA ( Table B in S1 Text ) . Sequence analysis was performed using Sequencher 5 . 0 software . MnaA LOF mutations were independently confirmed by Illumina-based whole genome sequencing ( >100× genome coverage ) ( BGI Hong Kong ) . No additional non-synonymous mutations were found in MRSE . Only one MRSA COL LOF mutant carried an additional non-synonymous mutation ( Fig 2 ) . MICs were determined by the broth microdilution method in accordance with the recommendations of the Clinical and Laboratory Standards Institute in 96 well plates and assayed visually . MRSA strains were tested in CAMHB ( Difco ) . MRSE strains were tested in Luria Bertani broth ( Difco ) . Previously published [26] . Very briefly , stationary phase cells were used for extractions . Cells were washed and boiled for one hour , and pellets harvested for further processing . WTA was hydrolyzed and run on polyacrylamide gel electrophoresis . Performed as previously published [53] . Briefly , immune-suppressed CD-1 mice ( 5 per group ) were challenged intramuscularly in the right thigh with 1x106 CFUs of MRSA for imipenem efficacy or with indicated 10-fold dilutions for virulence studies . Mice were challenged with 2x106 CFUs of MRSE . For efficacy studies , mice were treated with indicated amounts of imipenem ( IPM ) . Thighs were harvested at 24hrs , homogenized and plated to determine CFU per thigh . All animal procedures were performed in accordance with the highest standards for the humane handling , care and treatment of research animals and were approved by the Merck Institutional Animal Care and Use Committee . The care and use of research animals at Merck meet or exceed all applicable regulations of the Animal Welfare Act as put forth by the United States Department of Agriculture . The protocol number is 2018-300643-Jan . It was approved in January of 2015 and will expire in January of 2018 . For total biofilm formation assays , wild-type MRSA COL or MRSE CLB26329 and their derived loss of function mutants or MRSA COL ltaS hypomorph were grown in TSB ( Difco ) with or without sub-MIC concentrations of drugs overnight at 37°C , 250rpm . Cultures were normalized to OD600 = 1 . 5 and diluted 1/50 in TSB + 0 . 2% glucose with or without indicated sub-MIC concentrations of drug . 200μl of culture were seeded in triplicate wells in duplicate 96-well plates pretreated overnight with bovine plasma ( Lampire , Pipersville , Pennsylvania , USA ) . Plates were incubated wrapped in parafilm at 37°C for 24 hours . One plate was shaken to resuspend biofilm and pellicles in liquid and OD600 taken to quantify total growth per well . The duplicate plate was processed for biofilm analysis . Supernatant was aspirated and wells washed gently three times with H2O . Biofilms were then fixed with Bouin’s fixative ( Electron Microscopy Sciences , Hatfield , Texas , USA ) for 15 minutes , supernatant removed and biofilms stained with 0 . 1% safranin ( Ricca Chemical Company , Arlington , Texas ) solution for 15 minutes . Plates were washed under running tap water to remove excess stain . Stained biofilms were dissolved in glacial acetic acid and OD564 measurements taken to quantify biofilm formation . Readings were normalized to corresponding total growth readings from the duplicate plate . For biofilm killing assays , biofilms were grown for 24 hours as above in the absence of compounds before addition of compounds at indicated drug concentrations . Biofilms were incubated another 24 hours , washed , fixed and stained with Syto 10 ( Life Technologies , Carlsbad , California , USA ) for total cell staining and DEAD Red ( Life Technologies ) for membrane-damaged , dead cell staining . Plates were excited at 492nm and emissions read at 505 and 615nm , respectively . For fluorescence microscopy , biofilms were grown for 24 hours as above in Cellcoat black μClear 96 well plates ( Greiner Bio-one , Monroe , North Carolina , USA ) in the absence or presence of compounds , supernatant was aspirated and wells washed gently three times with H2O . Biofilms were then stained with 0 . 1 μM BacLight Green bacterial stain ( Life Technologies ) in DPBS for 15 minutes , washed once and fixed with 4% formaldehyde for 30 minutes . Biofilms were examined at 60x magnification on a Nikon Eclipse Ti using a FITC filter . Z-stacks were acquired using NIS Elements AR software ( Nikon , Tokyo , Japan ) . Protein purification methods are described in Supplemental materials . The S . aureus COL MnaA protein sample concentrated at 31 mg ml-1 was screened for crystallization by free interface diffusion using Topaz nano-chips . The crystallization condition most readily transferable to a set-up by vapor diffusion contained 0 . 1M Na Cacodylate , pH 6 . 5 , 0 . 1M Li2SO4 , 30% PEG 400 . However the reproducibility of the experiments was poor and the crystals when they grew had often poor diffraction not exceeding 4–6 Å . Structure determination using the best crystal showed that UDP is present bound to the protein in the crystal , although no exogenous UDP was ever added to the protein sample at any step during protein purification or crystallization . Crystallization reproducibility and crystal diffraction were subsequently significantly improved by adding Na2UDP from a 200 mM stock solution to the protein sample right before crystallization set-up to a final ligand concentration of 4 mM . The final optimized conditions were 0 . 1 M Tris Cl pH 8 . 0 , 0 . 1M Na2SO4 , 52% PEG 400 at a temperature of 22°C , adding 1 . 5 μl precipitant to 1 . 5 μl protein and let equilibrate by vapor diffusion against precipitant in a hanging drop set-up . For diffraction experiments the crystals were harvested from the crystallization drop and directly frozen in a bath of liquid nitrogen . The crystals grow in space group P 21 21 21 , a = 55 . 5Å , b = 85 . 8Å , c = 168 . 8Å with two molecules per asymmetric unit and diffract up to 1 . 9Å . The data were collected at the Canadian Light Source 08ID-1 beam line on a Mar mosaic CCD300 CCD detector ( Canadian Light Source , Saskatoon , Saskatchewan , Canada ) . The diffraction data were processed , reduced and merged using the autoPROC automated pipeline with calls to the XDS software for indexing and integration , and the package AIMLESS for scaling and merging . The structure was solved by molecular replacement with the MOLREP program using PDB entry 1F6D as a starting point . The model was first refined with autoBUSTER , then rebuilt with the sequence switched to the S . aureus sequence using the COOT graphical suite . The structure was compared with one rebuilt at that point with the Phenix automated AutoBuild procedure and some results obtained with the latter incorporated in the model . The UDP ligand was added in a difference “omitmap” ( i . e . the ligand was always excluded from the model prior to generating this map ) . After several cycles of refinement using autoBUSTER and rebuilding with COOT the final models contains all residues except residues 38 to 43 , 60 to 67 , and from 376 to the C-terminal residue in one copy of the molecule in the asymmetric unit , and 60 to 67 , and from 376 to the C-terminal end in the other . The model also contains one molecule of UDP per chain , 185 waters , and 3 sulfate anions . It refines to a final crystallographic Rwork and Rfree values of 20 . 4% and 22 . 7% , respectively , and presents good stereochemistry according to the program Molprobity . The model and structure factors have been deposited in the Protein Data Bank with code 5ENZ . Tunicamycin binding was detected by saturation transfer difference ( STD ) NMR . 15 μM tunicamycin was added into 500 μl binding buffer [25 mM Tris ( d11 ) -DCl ( pD 8 . 0 ) , 50 mM NaCl and 25 μM TSP ( 2 , 2 , 3 , 3-Tetradeutero-3 ( trimethylsilyl ) propionic acid ) in 99 . 98% D2O] which contained either 5 μM 2-epimerase ( MnaA/Cap5P ) or no protein as a negative control . The binding mixture was incubated for 2 hours at 25°C before NMR data collection . STD NMR spectra were collected at 298 K on a Bruker 600 MHz Avance spectrometer ( Bruker , Billerica , Massachusetts , USA ) equipped with a 5 mm TXI cryogenic probe . Selective saturation of the protein was applied by switching the on-and off-resonance saturation frequency after each scan . A train of Gaussian shape pulses with 50 ms pulse length ( corresponding to an excitation width of 100 Hz ) separated by a delay of 1ms was used , with the total length of the selective saturation set to 3s , and the on-and off-resonance saturation frequencies set to -120 Hz and 20 , 000 Hz , respectively . A total time of 50 minutes was required to collect a single STD NMR spectrum including sample changing . The STD NMR experiment was repeated on a solution of tunicamycin in the absence of protein to exclude any artifacts and make sure the observed STD NMR signals are due to tunicamycin binding to 2-epimerase . MnaA-catalyzed interconversion of UDP-GlcNAc and UDP-ManNAc was carried out in a total volume of 50 μl containing either UDP-GlcNAc or UDP-ManNAc ( 0–3 mM as indicated ) in 10 mM NaPi , 50 mM NaCl , pH 8 . 0 . Reactions were initiated by the addition of 0 . 109 μg MnaA-His6 ( forward reaction , FW ) or 0 . 327 μg MnaA-His6 ( reverse reaction , RV ) and incubated for 10 min to 5 h at 30°C . All enzymatic reactions were quenched by heating ( 10 min , 100°C ) and analyzed by capillary electrophoresis . Tunicamycin ( Sigma Aldrich , Munich , Germany ) was added at concentrations ranging from 0 to 200 μM . Reactions were stopped by heating after 120 min ( 10 min , 100°C ) .
|
Staphylococcus aureus and Staphylococcus epidermidis cause life-threatening infections that are commonly acquired in hospitals as well as the community and remain difficult to treat with current antibiotics . In part , this is due to the emergence of methicillin-resistant S . aureus and S . epidermidis ( MRSA and MRSE ) , which exhibit broad resistance to β-lactams such as penicillin and other members of this important founding class of antibiotics . Compounding this problem , Staphylococci commonly colonize the surface of catheters and other medical devices , forming bacterial communities that are intrinsically resistant to antibiotics . Here we functionally characterize a family of 2-epimerases , named MnaA and Cap5P , that we demonstrate by genetic , biochemical , and X-ray crystallography means are essential for wall teichoic acid biosynthesis and that upon their genetic inactivation render methicillin-resistant Staphylococci unable to form biofilms as well as broadly hypersusceptible to β-lactam antibiotics both in vitro and in a host infection setting . WTA 2-epimerases therefore constitute a novel class of methicillin-resistant Staphylococcal drug targets .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"biofilms",
"bacteriology",
"antimicrobials",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"drugs",
"microbiology",
"staphylococcus",
"aureus",
"methicillin-resistant",
"staphylococcus",
"aureus",
"antibiotics",
"pharmacology",
"bacteria",
"bacterial",
"pathogens",
"staphylococcus",
"medical",
"microbiology",
"microbial",
"pathogens",
"biochemistry",
"bacterial",
"biofilms",
"microbial",
"control",
"biology",
"and",
"life",
"sciences",
"biosynthesis",
"organisms",
"staphylococcus",
"epidermidis"
] |
2016
|
Chemical Genetic Analysis and Functional Characterization of Staphylococcal Wall Teichoic Acid 2-Epimerases Reveals Unconventional Antibiotic Drug Targets
|
Large-scale sequencing efforts have captured a rapidly growing catalogue of genetic variations . However , the accurate establishment of gene variant pathogenicity remains a central challenge in translating personal genomics information to clinical decisions . Interferon Regulatory Factor 6 ( IRF6 ) gene variants are significant genetic contributors to orofacial clefts . Although approximately three hundred IRF6 gene variants have been documented , their effects on protein functions remain difficult to interpret . Here , we demonstrate the protein functions of human IRF6 missense gene variants could be rapidly assessed in detail by their abilities to rescue the irf6 -/- phenotype in zebrafish through variant mRNA microinjections at the one-cell stage . The results revealed many missense variants previously predicted by traditional statistical and computational tools to be loss-of-function and pathogenic retained partial or full protein function and rescued the zebrafish irf6 -/- periderm rupture phenotype . Through mRNA dosage titration and analysis of the Exome Aggregation Consortium ( ExAC ) database , IRF6 missense variants were grouped by their abilities to rescue at various dosages into three functional categories: wild type function , reduced function , and complete loss-of-function . This sensitive and specific biological assay was able to address the nuanced functional significances of IRF6 missense gene variants and overcome many limitations faced by current statistical and computational tools in assigning variant protein function and pathogenicity . Furthermore , it unlocked the possibility for characterizing yet undiscovered human IRF6 missense gene variants from orofacial cleft patients , and illustrated a generalizable functional genomics paradigm in personalized medicine .
The rapid development of next-generation sequencing technologies has ushered in a new era of personalized medicine for a myriad of diseases [1] . Large-scale consortia sequencing efforts have documented thousands of whole exomes and genomes from both disease patients and the general population , and captured a growing catalogue of genetic variations for statistical comparisons and analyses [2] . However , a frequent challenge in the analysis of human gene variants is the establishment of pathogenicity for disease , distinguishing disease-causing variants from the background of variants present across human populations that are rare and undetermined in function , but not actually pathogenic . Statistical methods based on the relative enrichment of certain gene variants in disease populations [3–6] , and computational methods based on sequence conservation or structural information with limited biological data are often inadequate and provide conflicting results [7 , 8] . Indeed , false assignment of pathogenicity for gene variants is a critical challenge in translating knowledge gained from genome sequencing to clinical diagnoses and treatments . One focused resequencing study recently revealed that as many as 27% of previously published disease-causing variants were benign or lacked sufficient evidence for pathogenicity and therefore should be categorized as variants of unknown significance [9] . In addition , the Exome Aggregation Consortium ( ExAC ) recently published a study utilizing the largest aggregation of human exomes to reveal that while each person has an average of 54 variants in their genome that are currently annotated as pathogenic , as many as 41 of them are now observed to occur frequently in the human population and thus are unlikely to cause disease [10] . As the amount of exome/genome sequencing data continues to increase exponentially , it is crucial for candidate gene variants to undergo rigorous , multi-pronged evaluation before pathogenicity assignments . In addition to statistical ( case-control association , familial segregation , population frequency , and etc . ) and bioinformatic ( evolutionary conservation , protein energetics , and etc . ) methods , experimental approaches utilizing biological assays that directly test the protein functions of gene variants should be implemented to provide functional evidence that directly links gene variants to the pathogenesis of disease [11] . One well-studied example of a common congenital malformation associated with gene variants is that of orofacial clefts associated with the transcription factor Interferon Regulatory Factor 6 ( IRF6 , ENSG00000117595 ) [12–15] . Orofacial clefts are among the most common congenital malformations with an estimated incidence of approximately 1 in 700 births [16] . Pathogenic variants in IRF6 are among the most common genetic causes of cleft lip and/or palate ( CL/P ) [17] and are associated with both Van der Woude syndrome ( VWS , OMIM 119300 ) and Popliteal Pterygium syndrome ( PPS , OMIM 119500 ) , autosomal dominant Mendelian disorders with variable penetrance and expressivity that are characterized by CL/P and skin abnormalities [12] . The IRF6 gene sequence is highly conserved across vertebrates and contains two functional domains , a helix-turn-helix DNA-binding domain and a SMIR/IAD protein-binding domain [12 , 18] . From murine studies , disruption of Irf6 led to a CL/P phenotype in addition to oral epithelial adhesions , poor epithelial barrier functions , and improper skin stratification , revealing a potentially important role for the oral epithelium in directing palate development [19 , 20] . Approximately 300 human IRF6 gene variants have been catalogued [15] , and despite the wealth of structural and biological data for this well-described transcription factor , accurate determinations of variant protein functions and pathogenicity assignments associated with IRF6 gene variants remain significant challenges . Moreover , different computational programs use algorithms that weight various aspects of amino acid change differently and often provide conflicting predictions on protein function for the same missense mutation [21] . The large number of human IRF6 gene variants and its well-understood biology make IRF6 an ideal model to examine the challenges of assigning variant protein function and pathogenicity . In order to develop a rapid functional assay to determine the protein functions of a large number of IRF6 gene variants , we utilized a novel irf6 null zebrafish model by taking advantage of the finding that maternally-deposited irf6 transcripts are required for the proper development of the embryonic epithelium ( periderm ) during epiboly [22] . By using CRISPR-Cas9 to generate a zebrafish irf6 null model , we were able to establish maternal-null irf6 -/- homozygotes where 100% of the embryos lacked Irf6 and ruptured unless rescued by functional Irf6 protein . This irf6 rescue assay was used to test the protein functions of human IRF6 missense gene variants , and provide an additional line of biological evidence to help bridge the gap between gene variant identification and pathogenicity assignment .
To establish a zebrafish irf6 null model and investigate the role of irf6 in periderm and craniofacial development , CRISPR-Cas9 was used for mutagenesis targeted at exon 6 of irf6 at the start of the SMIR/IAD protein-binding domain ( Fig 1A ) . CRISPR-injected P0 embryos were raised to adulthood , out-crossed with wild type Tü adults , and genotyped at the gRNA target site . A F1 line was identified containing an 8 bp deletion in exon 6 of the irf6 coding region ( NC_007133 . 7 ( NM_200598 . 2 ) :c . 772_779del ) , here forward referred to as “Δ8bp” ( Fig 1A ) . Heterozygous F1 embryos ( irf6 +/Δ8bp ) were in-crossed to produce wild type , heterozygous and homozygous progeny at the expected Mendelian ratios with normal embryonic and craniofacial development ( Fig 1B ) . It has been previously reported through gene expression analysis and experiments involving dominant-negative Irf6 that maternal transcripts of irf6 , deposited in the oocyte cytoplasm during gametogenesis , were critical for proper periderm differentiation and epiboly progression [22 , 23] . The zebrafish periderm is an embryonic epithelium with many morphologic and molecular features similar to the mammalian embryonic oral epithelium that surrounds the facial prominences [24 , 25] . Based on these previous reports of irf6 maternal contributions , homozygous irf6 Δ8bp/Δ8bp F2 females were crossed with males of any genotype to produce heterozygous ( +/Δ8bp ) and homozygous ( Δ8bp/Δ8bp ) irf6 embryos that developed normally up to the sphere stage ( 4 hours post fertilization = hpf ) . However , shortly thereafter these maternal irf6 Δ8bp/Δ8bp embryos , regardless of genotype , failed to appropriately initiate epiboly compared to wild type embryos , resulting in the separation of the animal pole structures from the underlying yolk , periderm rupture , and embryonic lethality in 100% of embryos between 5–6 hpf ( Fig 1D , 1D’ , 1E and 1E’ ) . Conversely , when homozygous F2 males were crossed with wild type or heterozygous females , the resulting embryos underwent normal epiboly progression and embryonic development , confirming that maternal contributions of irf6 , even from heterozygous irf6 +/Δ8bp females , were sufficient for epiboly progression ( Fig 1B ) . Another zebrafish irf6 line harboring a 5 bp insertion in exon 6 of the irf6 coding region was also identified from the same CRISPR-injected P0 embryos and produced the same set of phenotypes as the Δ8bp deletion line ( Fig 1A ) . The 8bp deletion in the irf6 coding region was predicted by in silico translation to cause a frameshift and nonsense mutation downstream of the CRISPR gRNA target site , truncating the resulting protein to 264 amino acids ( 29 kD ) compared to the complete wild type 492 amino acids ( 55 kD ) ( Fig 1A ) . It has been previously reported that some missense and nonsense mutations in exon 6 of irf6 resulted in Irf6 proteins with dominant-negative activities [12] . Thus , in order to characterize the irf6 Δ8bp zebrafish model at the molecular level , RT-qPCR using primers specific to the irf6 5’ UTR overlapping the protein N-terminus was performed on zebrafish embryos at the sphere stage ( 4 hpf ) from permutations of wild type and maternal/paternal homozygous irf6 Δ8bp/Δ8bp crosses . Transcript levels of irf6 were undetectable in all maternal irf6 Δ8bp/Δ8bp crosses at 4 hpf , regardless of whether the embryos were homozygous or heterozygous for the Δ8bp irf6 allele ( Fig 1C ) . Conversely , the relative levels of irf6 transcripts in embryos from wild type and paternal irf6 Δ8bp/Δ8bp parents were comparable ( Fig 1C ) . Western blot analysis using a polyclonal antibody specific for zebrafish Irf6 was performed and confirmed that Irf6 protein was undetectable in embryos from maternal irf6 Δ8bp/Δ8bp crosses but comparable between embryos from wild type and paternal irf6 Δ8bp/Δ8bp crosses ( Fig 1C ) . Furthermore , all embryos from irf6 +/Δ8bp heterozygous females crossed with irf6 Δ8bp/Δ8bp homozygous males developed normally , when one would have expected them to rupture if the truncated Irf6 protein from the Δ8bp irf6 allele possessed dominant-negative activity because the maternal deposition of both wild type and Δ8bp irf6 transcripts in irf6 Δ8bp/Δ8bp embryos would have led to the manifestation of the periderm rupture phenotype ( Fig 1B ) . Taken together , the results suggest that the Δ8bp deletion caused irf6 transcript destabilization and degradation in the oocyte cytoplasm , and significantly decreased Irf6 protein production , thus indicating that the Δ8bp irf6 allele generated by CRISPR is a null allele ( from here forward “Δ8bp” is referred to as “–“ ) . Although the CRISPR gRNA used to generate the irf6 Δ8bp allele was predicted in silico to have no off-target sites in coding regions of the zebrafish genome ( crispr . mit . edu ) , we sought to determine the specificity of the embryonic periderm rupture phenotype from CRISPR irf6 gene disruption by injecting wild type zebrafish irf6 mRNA into maternal-null irf6 -/- embryos at the one-cell stage , and assessing whether the periderm rupture phenotype can be rescued . Injection of zebrafish irf6 mRNA reliably rescued the rupture phenotype in maternal-null irf6 -/- embryos , and the rescued embryos were able to initiate epiboly and undergo normal embryonic development indistinguishable from wild type embryos ( Fig 2A–2H ) . Because of the significant sequence conservation of IRF6 across species ranging from human to zebrafish [18] , we hypothesized that it may be possible for human IRF6 to retain its function in zebrafish . Wild type human IRF6 cDNA ( NM_006147 . 3 ) was isolated , in vitro transcribed into mRNA , and injected into maternal-null irf6 -/- zebrafish embryos to determine whether the periderm rupture phenotype could be rescued . The ability of human IRF6 mRNA to fully rescue the maternal-null irf6 -/- periderm rupture phenotype was indistinguishable from that of zebrafish irf6 mRNA . In addition , maternal-null irf6 -/- embryos rescued by either zebrafish or human IRF6 mRNA appeared indistinguishable from wild type embryos throughout embryonic development with normal craniofacial morphologies ( Fig 2M and 2N ) . It was previously demonstrated that inhibition of Irf6 protein function in zebrafish causes decreased expression of several known downstream target genes such as krt4 , klf2a , and grhl3 , many of which are important for maintaining proper periderm integrity and developmental signaling pathways [23 , 24] . Using the irf6 -/- zebrafish model , we sought to determine the gene expression changes that result from the depletion of irf6 maternal transcripts . Whole-mount in situ hybridization for genes in the irf6 gene regulatory network in maternal-null irf6 -/- embryos at the sphere stage revealed significant down-regulation of gene expression in several of the genes tested , demonstrating a more complete ablation of gene expression in irf6 downstream gene regulatory targets compared to the previously reported dominant-negative Irf6 model [22] ( Fig 2A–2H ) . To further validate the zebrafish irf6 -/- model , we sought to molecularly characterize the functional rescue of maternal-null irf6 -/- periderm rupture by microinjection of zebrafish or human IRF6 mRNA by examining the rescue of gene expression within the irf6 gene regulatory network . RT-qPCR was performed for several genes previously known to be down-regulated in the absence of functional irf6 such as tfap2a [26 , 27] , grhl1 [22] , and grhl3 [23 , 24] . The results revealed significant down-regulation of these genes in maternal-null irf6 -/- embryos at 4 hpf compared to wild type , but statistically indistinguishable levels compared to wild type in maternal-null irf6 -/- embryos microinjected with either zebrafish or human IRF6 mRNA ( Fig 3I ) . Furthermore , the qPCR results revealed significant down-regulation of genes important for epithelial and craniofacial development in maternal-null irf6 -/- embryos compared to wild type ( Fig 3I ) , suggesting that irf6 maternal transcripts might play key roles in activating the molecular pathways required for zebrafish periderm maturation and epiboly , and potentially later for signaling pathways related to craniofacial development . The conservation of human IRF6 protein function in zebrafish provided an opportunity to assess the protein functions of human IRF6 missense gene variants . Over 300 IRF6 gene variants have been identified from human CL/P patients and approximately 50% are missense variants [15] . Two of the more commonly used computational prediction programs , PolyPhen-2 [28] and SIFT [29] , were used to predict the effects of amino acid substitutions on the protein functions of IRF6 missense variants and segregate them into three categories: 1 ) both programs agree the variant disrupts protein function , resulting in a loss-of-function protein , 2 ) the programs disagree on the effects of the variant on protein function , and 3 ) both programs agree the variant does not disrupt protein function . Human IRF6 missense gene variants were then mapped to their corresponding nucleotides in the zebrafish irf6 cDNA by sequence conservation , in vitro transcribed into mRNA , and microinjected into maternal-null irf6 -/- zebrafish embryos to assess their ability to rescue the periderm rupture phenotype ( Fig 4A ) . The results demonstrated that the computational programs did not offer a significant advantage in predicting the biological functions of variant Irf6 proteins ( Fig 4D–4F ) . Variants that received conflicting predictions from PolyPhen-2 and SIFT also provided mixed results in their abilities to rescue ( Fig 4D ) . Moreover , variants that were predicted by both computational programs to result in loss-of-function proteins were often able to experimentally rescue the rupture phenotype ( Fig 4E ) . Lastly , missense gene variants that were predicted by both programs to be non-deleterious to protein function were also mixed in their abilities to rescue , further demonstrating the limitations of these computational programs at predicting protein function ( Fig 4F ) . For example , the missense gene variant p . F252L that was predicted by both programs to be non-deleterious to protein function was unable to rescue the maternal-null irf6 -/- rupture phenotype ( Fig 4F ) . When the experimentally tested IRF6 gene variants were grouped according to their abilities to rescue periderm rupture and mapped to the predicted human IRF6 protein structure ( protein-binding domain and C-terminus ) generated by ExPASy , the distribution of the amino acid residues suggested variants that could not rescue mostly resided in protein secondary structures and thus likely to disrupt protein conformation and function ( Fig 4C ) . Conversely , amino acid residues for human IRF6 variants that retained protein function and rescued periderm rupture mostly mapped to regions without secondary structures ( Fig 4B ) and thus less likely to disrupt protein conformations critical for IRF6 function . In addition , the variants tested were re-examined for their genetic backgrounds and the numbers of individuals previously identified with the variant and this information was used to classify the variants according to the five-category system for variant pathogenicity established by the American College of Medical Genetics guidelines [30] ( Fig 4D–4F ) . While no variants in the highest pathogenicity certainty category of “Pathogenic” were able to rescue the rupture phenotype , other variants with varying degrees of uncertainty in pathogenicity were mixed in their abilities to rescue ( Fig 4D–4F ) . The only variant identified to be classified as “Benign” , p . V274I , was able to rescue the rupture phenotype ( Fig 4F ) . The human IRF6 missense gene variants functionally tested in this study could result in reduced function rather than complete loss-of-function proteins , thereby leaving open the possibility that while they can rescue the zebrafish maternal-null irf6 -/- rupture phenotype in this assay , their reduced function in vivo is sufficient to cause disease in humans . To address this possibility , the mRNA of wild type zebrafish irf6 , human IRF6 , and several missense gene variants were individually microinjected into maternal-null irf6 -/- embryos at the one-cell stage through a range of concentrations to establish a titration response curve for periderm rupture rescue ( Fig 5B ) . To assist in the functional analysis of these IRF6 missense gene variants with potentially more nuanced functional changes , we utilized the significantly increased statistical power of rare human gene variant detection provided by the Exome Aggregation Consortium database of over 60 , 000 individuals [10 , 15] . In addition , we also examined the gnomAD database , a more recent effort from the same group which now extended to over 125 , 000 exomes and 15 , 000 whole genomes . From these databases , the IRF6 missense gene variants p . R45Q , p . R45W , p . G70R , p . V274I , p . D354N , and p . F369S were identified . Through various lines of evidence , the variant p . V274I was already considered benign by ACMG standards ( Fig 5A ) . Indeed , 9 , 280 alleles of this gene variant ( allele frequency 0 . 077 ) was found in ExAC distributed across all populations including Europeans , Africans , and Asians . p . D354N and p . F369S were found to be non-conserved in zebrafish irf6 and illustrates the possibility that the identities of these residues are not essential for IRF6 protein function . In addition , although p . D354N was previously identified in four VWS pedigrees , 37 alleles were also found in the ExAC/gnomAD database . Three other missense gene variants , p . R45Q , p . R45W , and p . G70R were identified as single alleles in the ExAC/gnomAD database and were able to rescue the periderm rupture phenotype of maternal-null irf6 -/- embryos . The small number of these alleles do not provide as strong of support for non-pathogenicity as p . V274I . But , their presence in ExAC/gnomAD does raise the possibility that these gene variants could retain protein function and potentially be non-pathogenic , and therefore should be functionally tested . The aforementioned variants , in addition to p . V274I and several others that were mixed in their ability to rescue were used in a mRNA microinjection dosage titration assay . Interestingly , the tested missense gene variants naturally segregated into three functional categories upon dosage titration . The variants identified in ExAC/gnomAD rescued to the same degree as wild type zebrafish and human IRF6 mRNA ( Fig 5B Green ) . Variants that were able to rescue in the periderm rupture assay but not found in ExAC/gnomAD were found to be reduced in protein function compared to wild type , rescuing a smaller percentage of embryos at each of the mRNA dosages tested ( Fig 5B Blue ) . These results are in contrast to the missense gene variants that could not rescue the rupture phenotype at any dosage tested , which were not found in ExAC/gnomAD or other public exome/genome databases , suggesting that they are complete loss-of-function variant proteins and likely pathogenic in human orofacial cleft patients . Although the IRF6 missense gene variants that were not found in ExAC/gnomAD had reduced protein activities , they otherwise retained their biological functions and were not only able to rescue the zebrafish periderm rupture phenotype at high mRNA injection dosages , but also permitted normal embryonic development ( Fig 6 ) . The variant mRNA-rescued maternal-null irf6 -/- embryos that otherwise would have ruptured were grown under standard conditions and resulted in not only phenotypically wild type craniofacial development ( Fig 6 ) , but also in viable and fertile adults .
While it is tremendously useful to document human genetic variations in genes associated with disease from a wide range of populations , as in the case of IRF6 for orofacial clefts , a key challenge in human genetics remains how to functionally ascertain whether coding sequence variations result in harmful alterations in protein function , and whether these functional changes are pathogenic for disease [5 , 6] . Statistical methods can provide support for the pathogenicity of gene variants by examining the segregation of pathogenic variants with disease status within affected families , or by examining the frequencies of the variants in the population or disease cases versus control [8] . However , such statistical methods do not test biological protein functions and are prone to biases , especially for rare gene variants . Distinct but unobserved pathogenic variants may be located on the same haplotype as the candidate rare variant and thus segregation analysis alone cannot unambiguously assign pathogenicity , especially in smaller pedigrees [1] . In addition , the incidence of CL/P is higher in regions of the world where whole exome and genome sequencing has not yet captured a large cross section of the normal population for use as controls . Due to this geographical clustering , many cases of newly discovered IRF6 gene variants cannot be statistically compared to public exome/genome databases because the population mismatch could over emphasize the relative rarity of certain gene variants and thereby their potential pathogenicity . Computational programs such as PolyPhen-2 and SIFT that predict the effects of missense mutations on variant protein functions use complex algorithms that take into consideration a multitude of parameters to predict the thermodynamic stability and functions of variant proteins after amino acid substitutions . However , there are still many unaccounted factors that go into translating amino acid changes to changes in protein function and thus such computational programs often provide results that conflict with biological evidence . Since many computational prediction programs depend on machine-learning algorithms [31] , the direct biological assessment of variant protein functions could be reiterated through the same algorithms to improve their predictive powers for both IRF6 and other proteins with similar sequences and motifs . According to the recent ACMG guidelines , the assessment of gene variant pathogenicity should be multi-pronged with various independent lines of supporting evidence from different approaches , including statistical , computational and experimental [30] . While pathogenicity assignments typically cannot be made from any line of evidence alone , using experimental models to directly interrogate variant protein functions provide valuable insights into the biological effects of missense variant amino acid substitutions on protein functions and can greatly aid in the interpretation of variant protein function and pathogenicity . The zebrafish irf6 -/- model revealed the importance of irf6 maternal transcripts for embryonic periderm development , corroborating previous reports that characterized irf6 function in the zebrafish and Xenopus laevis models [22 , 23] . However , in contrast to previous studies of irf6 performed using mRNA injections of dominant-negative Irf6 which could result in delayed and incomplete knockdown of Irf6 function , the genetic disruption of irf6 reported in this study more completely ablated Irf6 function in the early embryo with near undetectable expression of downstream transcriptional targets such as grhl3 and krt4 [24 , 25 , 32 , 33] . This irf6 null model exhibited developmental arrest at the sphere stage and suggests that irf6 is a potential epiboly initiation factor necessary for regulating cell signaling pathways that orchestrate this complex morphological event during zebrafish embryogenesis . Interestingly , paternal-null irf6 -/- embryos developed normally into adulthood , suggesting that zygotic transcription of irf6 may not be necessary in zebrafish for normal embryonic development , possibly due to the persistence of maternal Irf6 protein throughout early embryogenesis . Irf6 -/- mice exhibited cleft palates with oral epithelial adhesions and a number of other epithelial abnormalities , suggesting that IRF6 plays an important role in regulating epithelial proliferation and differentiation [19 , 20] . The zebrafish embryonic periderm has emerged as a model of the mammalian oral epithelium because many of the gene regulatory networks and cellular behaviors during zebrafish epiboly , such as convergence-extension and epithelial-to-mesenchymal transition , are conserved in the mammalian oral epithelium during palate development [24 , 25 , 34] . The observation that maternal-null irf6 -/- embryos failed to initiate epiboly cellular movements suggests that many of the downstream pathways and cellular behaviors are dependent upon the proper establishment of the periderm , and emphasizes the potential importance of the oral epithelium in initiating and orchestrating both epithelial and mesenchymal tissue behaviors during palate development in the mammalian system . Because of the early embryonic lethal periderm rupture phenotype of maternal-null irf6 -/- embryos , we are in the process of elucidating the functional requirements of Irf6 during craniofacial development in zebrafish , separated from its functions in the establishment of the periderm during epiboly . The experimental finding that human IRF6 mRNA could rescue not only periderm rupture in zebrafish maternal-null irf6 -/- embryos but also normal embryonic development and IRF6 gene regulatory network gene expression suggests that there is significant cross-species conservation in IRF6 protein structure and function . Taken together , our finding demonstrated that the zebrafish maternal-null irf6 -/- model could serve as a sensitive and specific platform for the rapid assessment of human IRF6 missense variant protein functions in a relevant in vivo context . This biological assay can complement the statistical analyses to form a more comprehensive picture in the process of assigning pathogenicity to IRF6 missense gene variants . This complementary approach is especially important for rare IRF6 gene variants identified in a small number of individuals often in a single pedigree . In the case of p . R45W , this gene variant was detected in a single VWS affected proband but also in his unaffected sibling and mother , leaving in question whether this missense variant is truly pathogenic or simply a rare benign variant that was annotated as pathogenic despite the imperfect variant co-segregation with disease which was previously attributed to incomplete penetrance and variable expressivity [35] . Although this interpretation is possible due to the variable expressivity of phenotypes exhibited by VWS patients in the same pedigree , other interpretations are possible such as a case where an unobserved pathogenic variant in a separate gene adjacent to IRF6 is on the same haplotype co-segregating with the p . R45W variant . This interpretation is further supported by the biological analysis of p . R45W variant protein function in the zebrafish model , which revealed that the p . R45W variant protein was able to rescue maternal-null irf6 -/- periderm rupture and function quantitatively to the same degree as wild type IRF6 ( Fig 5B ) . While this experimental validation of p . R45W variant protein function does not conclusively exclude its potential pathogenicity in human patients , it does suggest that further biological evidence is needed before p . R45W can be annotated as pathogenic in public databases and used in the clinical diagnosis of VWS patients . The functional genomics validation of IRF6 missense gene variant protein function presented in this study is complemented by rapid increases in statistical power for rare gene variant identification and pathogenicity assignment through expansions in large public exome/genome databases . More and more individuals are sequenced daily with advances in sequencing technologies and concomitantly decreasing sequencing costs . Several IRF6 missense gene variants previously unobserved in the general population and therefore thought to be pathogenic were identified in the ExAC/gnomAD databases [10] , potentially weakening the statistical power of their pathogenicity assignments by ACMG standards . In addition , their discovery in the ExAC/gnomAD databases streamlined the variant protein functional validation process by identifying variants with the highest probability of ambiguous pathogenicity assignments for testing in our model . The zebrafish maternal-null irf6 -/- model allowed for the detection of subtle changes in IRF6 protein function such as reduced function variants through mRNA microinjection dosage titrations . The stability of variant mRNA and proteins could also be readily assessed through molecular biology techniques after mRNA microinjections . The missense variants p . R45Q , p . R45W , p . G70R , and p . V274I were able to not only rescue the periderm rupture phenotype of maternal-null irf6 -/- zebrafish embryos , but were also functionally indistinguishable from wild type zebrafish and human IRF6 in dosage titration , suggesting that the proteins produced from these missense variants retained full function . However , the pathogenicity of these variants cannot be conclusively determined with this IRF6 functional model due to the possibility that only a subset of IRF6 functions are conserved between human and zebrafish . Interestingly , the variants tested that were able to rescue in the embryo rupture assay and yet not found in the ExAC/gnomAD databases were reduced in protein function when compared to wild type zebrafish or human IRF6 . This result suggests that these missense gene variants are not complete loss-of-function variants but rather reduced in function , and thus still potentially pathogenic in humans because their reduced functions might be insufficient to prevent phenotype and disease onset . Lastly , the variants that were not able to rescue the maternal-null irf6 -/- periderm rupture at any of the dosages tested likely represent complete loss-of-function missense variants . These variants were not found in the ExAC/gnomAD databases , and their complete loss-of-function provide an additional line of biological evidence in support for their pathogenic status in human VWS patients . Because of the relatively small number of missense gene variants functionally tested through dosage titration in this study , no variant was discovered to be functionally wild type and yet not identified in the ExAC/gnomAD databases . Although the collection of sequenced exomes and genomes is continuously increasing , a larger control population still does not guarantee the discovery of rare gene variants previously thought to be pathogenic in human disease . In these situations , experimental validation of rare gene variant protein functions could bridge a gap in knowledge and provide additional biological insights to assist in variant pathogenicity assignments and improve their accuracies . While it is possible that only a subset of IRF6 protein functions are conserved in the zebrafish model and tested through this functional rescue assay , this model serves to provide novel insights into the effects of human IRF6 missense gene variants on IRF6 protein function , and complement current statistical and bioinformatics results . Overall , the zebrafish maternal-null irf6 -/- model not only offers a method to rapidly assess the protein functions of current and yet undiscovered human IRF6 missense gene variants , but also illustrates a generalizable functional genomics paradigm where novel human gene variants can be biologically tested for protein function using corresponding zebrafish mutants to provide another line of biological evidence to assist with pathogenicity assignments . With advances in CRISPR-Cas9 targeted mutagenesis in zebrafish , it will be increasingly efficient to develop zebrafish gene disruption and rescue assays to test the functions of human gene variants for genes in other contexts of disease .
All work with zebrafish ( adult , larval , and embryonic ) was performed in strict accordance with protocols approved by Massachusetts General Hospital IACUC ( Protocol 2010N000106 ) . Zebrafish Danio rerio were maintained in accordance with approved institutional protocols at Massachusetts General Hospital , as described [36] . Embryos were kept at 28 . 5°C in E3 media containing 0 . 0001% methylene blue and staged according to [37] by hours or days post fertilization ( hpf or dpf ) . All wild type and mutant zebrafish lines used in these experiments were generated from the Tübingen ( Tü ) strain . Potential CRISPR gRNA target sites were identified using the CRISPR design program at ( zifit . partners . org/ZiFiT/ and crispr . mit . edu ) . The irf6 exon 6 gRNA was generated by in vitro transcription from a T7 promoter as described [38] . Zebrafish optimized cas9 template DNA pT3TS-nls-zCas9-nls [39] was linearized using XbaI and purified using the QIAquick PCR purification kit ( Qiagen ) . Capped cas9 mRNA was synthesized from a T3 promoter using the mMESSAGE mMACHINE T3 transcription kit ( Ambion ) and purified using the RNeasy mini kit ( Qiagen ) . One-cell staged zebrafish embryos were injected directly in the cytoplasm with 2 nl of a solution containing 25 ng/μl of gRNA and 100 ng/μl of cas9 mRNA . Genomic DNA for genotyping was isolated from either whole 24 hpf embryos or tail fin clips using the HotSHOT method [40] . Genotyping primers flanking CRISPR target site were designed with a forward primer modified with 5’-FAM and submitted for microsatellite analysis to determine indel mutation size and frequencies . Sanger sequencing of the CRISPR target site was performed by cloning the genotyping PCR amplicon into pGEM-T easy ( Promega ) to validate the exact sequence changes from CRISPR mutagenesis . Stage-matched zebrafish embryos were flash-frozen in liquid nitrogen and homogenized with a micropestle in TRIzol ( Invitrogen ) . Total RNA was isolated using phenol-chloroform exaction and digested using DNase I ( Ambion ) to remove genomic DNA contamination . Total RNA was quantified using a NanoDrop spectrophotometer and 5 μg was used for reverse transcription using the SuperScript III cDNA synthesis kit ( Invitrogen ) . Quantitative RT-PCR was performed using PowerUP SYBR Green qPCR master mix ( Invitrogen ) on the StepOne Plus RT-PCR platform ( Applied Biosystems ) . Elongation factor 2α or β-actin were used as internal controls for expression normalization . Amplification specificity was checked using melt-curve analysis . Control amplifications were performed on samples either without reverse transcription or template . Zebrafish embryos were enzymatically dechorionated with pronase ( Sigma ) and deyolked according to [41] supplemented with HALT protease inhibitor cocktail ( Thermo Scientific ) . Cell pellets were flash frozen in liquid nitrogen and homogenized using a micropestle in RIPA buffer . Protein lysate concentrations were quantified using the DC Protein Assay kit ( Bio-Rad ) and subjected to electrophoresis ( 20 μg/lane ) in Novex Bis-Tris 10% protein gels ( Invitrogen ) . Gels were subsequently blotted onto a 0 . 22 μm PVDF membrane ( Novex ) , blocked for two hours at room temperature with StartingBlock in TBST ( Thermo Scientific ) and incubated with a rabbit polyclonal antibody for zebrafish Irf6 ( GeneTex ) ( 1:1500 dilution ) and a rabbit monoclonal antibody for zebrafish ß-actin ( Cell Signaling Technology ) ( 1:1500 dilution ) in blocking buffer at 4°C overnight . The membrane was washed 3x 10 min at room temperature and incubated with a HRP-conjugated anti-rabbit antibody ( Abcam ) ( 1:2000 dilution ) in blocking buffer at room temperature for one hour . Bands were visualized using Novex ECL chemiluminescence reagent ( Invitrogen ) . Full-length zebrafish irf6 ( NM_200598 . 2 ) and human IRF6 ( NM_006147 . 3 ) cDNA was synthesized using GeneArt ( Invitrogen ) and sub-cloned into the pCS2+8 vector ( Addgene #34931 ) at the EcoRV site in the MCS . IRF6 gene variants were identified in previously published literature and mapped to their corresponding nucleotides in the zebrafish irf6 cDNA . PCR-based site-directed mutagenesis was performed to generate pCS2+8 vectors containing irf6 missense gene variants using the Q5 site-directed mutagenesis kit ( New England Biolabs ) with amplification primers designed by the NEBaseChanger online tool ( http://nebasechanger . neb . com ) . Plasmids containing irf6 missense gene variants were isolated using the QIAprep spin miniprep kit ( Qiagen ) and sequence verified with either complete plasmid sequencing or Sanger sequencing through the entire cDNA region . Plasmids ( pCS2+8 backbone ) containing individual IRF6 missense gene variant cDNAs were linearized by NotI-HF and purified using QIAquick PCR purification columns ( Qiagen ) . Variant mRNA was synthesized by in vitro transcription using the SP6 mMESSAGE mMACHINE transcription kit ( Ambion ) and linearized plasmids as template . cDNA template was digested using DNase I and variant mRNA was purified using the RNeasy mini kit ( Qiagen ) and quantified with either a NanoDrop spectrophotometer or Qubit fluorometer ( Invitrogen ) . Variant mRNAs were diluted to 800 ng/μl and stored as aliquots in -80°C . For embryo microinjections , 2 nl of the injection mix was delivered directly into the cytoplasm of one-cell staged embryos with variant mRNA diluted to a final concentration of 50 ng/μl . Zebrafish embryos were fixed in 4% paraformaldehyde ( PFA ) at 4°C overnight and subsequently transferred into 100% methanol prior to whole-mount in situ hybridization ( WISH ) . WISH and DIG-labeled riboprobe synthesis were performed essentially as described in [42] . All riboprobes were amplified from mix-staged embryonic zebrafish cDNA , cloned into pGEM-T easy ( Promega ) , and direction/sequence verified with Sanger sequencing . Riboprobes were synthesized from either the T7 or SP6 promoter using a digoxigenin ( DIG ) labeling kit ( Roche ) . Detection was performed with an alkaline phosphatase conjugated anti-DIG antibody ( Roche ) and BCIP/NBT colorimetric substrates ( Sigma ) . Zebrafish embryos were fixed in 4% PFA at 4°C overnight and bleached ( 0 . 8% W/V KOH , 0 . 1% Tween20 , 0 . 9% H2O2 ) until pigmentation of cells were no longer present . Acid-free alcian blue staining was performed essentially as described [42] overnight on a rotating platform at room temperature . Whole or dissected stained embryos were mounted in 3% methylcellulose on a depression slide and imaged using a Nikon Eclipse 80i compound microscope with a Nikon DS Ri1 camera . Z-stacks were taken to increase the depth-of-field using the NIS Element BR 3 . 2 software and processed by ImageJ to provide a composite maximum intensity projection image . Human IRF6 missense gene variants were identified from previously published literature [15] and entered into PolyPhen-2 ( http://genetics . bwh . harvard . edu/pph2 ) and SIFT ( http://sift . jcvi . org ) for computational predictions of missense variant protein function . The ExPASy SWISS-MODEL online tool ( https://swissmodel . expasy . org ) was used to align the IRF6 amino acid sequence with the known crystalline structure of IRF1 to model the IRF6 SMIR/IAD protein-binding domain and C-terminus . The resulting protein structures were visualized using PyMOL . For statistical analysis of experimental data , error bars represent ±2x standard mean error ( SEM ) , and statistical significance was interrogated using two-tailed Student’s T-tests with <0 . 05 as the P-value cut-off . For the identification of IRF6 missense variants from the ExAC database ( exac . broadinstitute . org ) and gnomAD database ( gnomad . broadinstitute . org ) , IRF6 was queried and the results were sorted for missense gene variants .
|
Advances in sequencing technologies have led to rapid increases in personalized genetics information . Millions of differences exist when comparing the genomes of two individuals , accounting for the diversity of humans and occasionally disease pathogenicity . Accurate determination of the functional consequences of these differences is critical for translating personal genetic information into clinical decision . Various methods have been devised to meet this challenge . However , they are imperfect , especially in their abilities to interpret rare variants . Rare variants must be evaluated by multi-pronged approaches to establish disease pathogenicity , one crucial approach being functional testing through experimental models . Here , we utilized a zebrafish model to rapidly evaluate the protein functions of rare gene variants of IRF6 , a gene of established importance in orofacial cleft pathogenesis . IRF6 functions are well-conserved across species , and allowed us to test the functions of human IRF6 variants by their abilities to rescue zebrafish embryos depleted of irf6 . Many variants previously labeled pathogenic and loss-of-function retained full wild type-level protein activity , suggesting they could potentially represent benign , rather than disease-causing genetic variations . This paradigm is applicable to other genes and will allow researchers to rapidly triage gene variants for further study and clinicians to provide more accurate genetic diagnoses .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"biotechnology",
"genome",
"engineering",
"fish",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"microinjection",
"engineering",
"and",
"technology",
"synthetic",
"biology",
"messenger",
"rna",
"vertebrates",
"animals",
"synthetic",
"bioengineering",
"crispr",
"animal",
"models",
"osteichthyes",
"developmental",
"biology",
"model",
"organisms",
"experimental",
"organism",
"systems",
"molecular",
"biology",
"techniques",
"embryos",
"synthetic",
"genomics",
"research",
"and",
"analysis",
"methods",
"bioengineering",
"synthetic",
"genome",
"editing",
"embryology",
"gene",
"expression",
"pathogenesis",
"molecular",
"biology",
"biochemistry",
"zebrafish",
"rna",
"eukaryota",
"nucleic",
"acids",
"phenotypes",
"genetics",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2017
|
Rapid functional analysis of computationally complex rare human IRF6 gene variants using a novel zebrafish model
|
Complex diseases result from molecular changes induced by multiple genetic factors and the environment . To derive a systems view of how genetic loci interact in the context of tissue-specific molecular networks , we constructed an F2 intercross comprised of >500 mice from diabetes-resistant ( B6 ) and diabetes-susceptible ( BTBR ) mouse strains made genetically obese by the Leptinob/ob mutation ( Lepob ) . High-density genotypes , diabetes-related clinical traits , and whole-transcriptome expression profiling in five tissues ( white adipose , liver , pancreatic islets , hypothalamus , and gastrocnemius muscle ) were determined for all mice . We performed an integrative analysis to investigate the inter-relationship among genetic factors , expression traits , and plasma insulin , a hallmark diabetes trait . Among five tissues under study , there are extensive protein–protein interactions between genes responding to different loci in adipose and pancreatic islets that potentially jointly participated in the regulation of plasma insulin . We developed a novel ranking scheme based on cross-loci protein-protein network topology and gene expression to assess each gene's potential to regulate plasma insulin . Unique candidate genes were identified in adipose tissue and islets . In islets , the Alzheimer's gene App was identified as a top candidate regulator . Islets from 17-week-old , but not 10-week-old , App knockout mice showed increased insulin secretion in response to glucose or a membrane-permeant cAMP analog , in agreement with the predictions of the network model . Our result provides a novel hypothesis on the mechanism for the connection between two aging-related diseases: Alzheimer's disease and type 2 diabetes .
Complex diseases , such as diabetes and obesity , result from the interaction of genetic and environmental factors [1]–[3] . Approximately 170 gene loci have been robustly implicated in diabetes through genome-wide association studies [4] . Studies with knockout mouse models have identified hundreds of genes that can act autonomously to regulate insulin levels ( MP:0001560 ) [5] . However , it is still elusive to understand the underlying mechanisms of how these loci or genes contribute to diseases . Network modeling methods have been developed based on the premise that complex diseases are often caused by perturbation to a sub-network of genes [1] , [6]–[14] . We have applied these methods to identify causal genes for diabetes-related traits in multiple experimental mouse crosses [13]–[14] and human populations [1] . These analyses suggest that potentially many thousands of genes , under the right circumstances , can affect metabolic states . With the advancement of high-throughput technologies , such as DNA and RNA sequencing , methods that integrate various high-volume data sources are providing for more comprehensive characterizations of biological systems [15]–[18] . New methods have been developed to utilize high-dimensional data sets to infer unknown pathways , untangle gene-based regulatory networks , and identify novel disease-causing genes [13] , [19]–[23] . However , studying complex diseases at a systems level is still in its infancy . New technologies for data collection and novel methodologies of data interpretation are needed for a better resolution view of the system . In this study , we developed a network-based model to identify key genes that regulate plasma insulin levels in a B6XBTBR obese F2 cross . By applying a causality test for genes whose expression trait is linked to two loci that overlap insulin QTLs ( quantitative trait loci ) and integrating protein-protein interactions , we constructed a network for each of five tissues under study . We predicted that multiple genes in the pancreatic islet network may be involved in modulating plasma insulin levels in the B6XBTBR F2 cross , including App , Gria3 , Grb10 , Calca , and Ins1 . In particular , our pancreatic islet network predicts that the Alzheimer's disease gene , amyloid precursor protein App is a negative regulator of insulin abundance in the plasma . We therefore studied insulin secretion from islets of App knockout mice . Islets from 17-wk-old , but not 10-wk-old App −/− mice showed an increase in glucose and cAMP-stimulated insulin secretion , confirming that App acts as a negative regulator of insulin secretion . This result elucidates a possible mechanism connecting two common age-related diseases , Alzheimer's disease and type 2 diabetes .
Treating gene expression as a phenotypic trait , we computed eQTLs , for all genes expressed in pancreatic islets , white adipose tissue , liver , hypothalamus , and gastroc muscle of each male F2 mouse . We hypothesized that genes with eQTLs that co-localized with insulin QTLs are co-regulated by common genetic factors [13] . We identified eQTLs within each tissue that had LOD profile peaks on chromosomes 2 and 19 , the same genomic regions containing the peak insulin linkages . Among genes physically located within the insulin QTLs on chromosome 2 or 19 ( Table S1 ) , 89 genes have cis-eQTLs ( gene expression QTLs are mapped to within 10 Mb of the genomic location of the genes ) in islet , 66 in white adipose tissue , 52 in liver , 51 in hypothalamus , and 5 in gastroc have cis-eQTLs . Clearly , genes with cis-eQTLs may play significant roles in modulating insulin and methods have been developed to identify the causal genes with cis-eQTLs for various phenotypic traits [14] , [26]–[27] . However , each gene with a cis-eQTL can only explain the variance in the trait linked to its location . Here we considered a complementary strategy where we focused on genes with trans-eQTLs and interactions among them that integrate perturbations from multiple loci . As a greater number of genes showed trans-linkage , it is worth studying the potential mechanisms by which these genes jointly mediate the phenotypic variation . The expression traits that overlapped with the insulin QTLs were tissue-specific , and are enriched in different GO biological pathways ( Table S2 ) . The largest number of these traits was from pancreatic islets ( Table 2 , Figure 2 ) . In addition , islets contained the largest proportion of eQTLs that showed linkage to both loci on chromosomes 2 and 19 , indicating that similar to traditional complex traits ( e . g . insulin ) , gene expression is also regulated by multiple genetic loci [28] . Co-localization of gene eQTLs and plasma insulin QTLs does not imply the gene is related to plasma insulin regulation . To filter out genes that , while linked to the same QTL region as insulin QTLs , are likely independent of plasma insulin regulation ( described in Methods and Text S1 ) , we applied a genetic causality test developed by Schadt et al . [13] to further narrow our list of candidate regulatory genes . Given the known feedback loop ( shown in Figure 3 ) ( islets – insulin levels – peripheral tissues , and – glucose levels ) , genes supported as either causal ( QTL <$>\scale 80% \raster="rg1"<$> gene <$>\scale 80% \raster="rg1"<$> insulin ) , or reactive with respect to insulin levels ( QTL <$>\scale 80% \raster="rg1"<$> insulin <$>\scale 80% \raster="rg1"<$> gene ) were identified for consideration as insulin regulation genes . A model considering the two loci , on chromosomes 2 and 19 , accounts for a greater part of the variation in plasma insulin than a single locus model . Several models of various degrees of complexity could explain the joint regulation of a common trait by multiple loci . As plotted in Figure 4A , the simplest case ( M1 ) would be that the two loci directly regulate the same gene and that such a gene is responsible for modulating the trait . A slightly more complex case ( M2 ) would be each locus regulates a different gene , which could collaborate directly through protein-protein physical interaction to influence the trait . In model M3 , genes regulated by different loci interact indirectly and multiple steps exist before the perturbation signals merge on the common trait . In model M4 , multiple tissues and their interactions are involved in regulating the trait . Here , we developed an approach that seeks to combine the first two models to identify those components of the network underlying insulin regulation that are modulated by the chromosome 2 and 19 genetic loci and that may be physically interacting . The more complex models is not considered here but left for future work . As shown in Figure 4B , genetic variation at a single locus results in perturbations of biological functions that are reflected in the transcripts , or nodes , linked to that locus ( orange or green nodes ) . Genetic variation at two loci could result in a larger functional influence on nodes showing linkage to both loci ( yellow nodes ) or nodes interacting across the two sub-networks ( nodes connected by red edges ) . Here we only consider the situation in which single genetic variations are synergistic , although antagonistic relationships may certainly occur . We hypothesize that nodes mediating the interaction between the two sub-networks are critical points for integrating the effects of multiple loci . To identify and rank these critical nodes , we developed an algorithm for assessing a gene's potential for being such an integrator in each of five tissues . We first collected a mouse protein-protein interactome by combining information from various databases as previously described [20] , where most interactions were experimentally derived and manually curated . We then extract tissue-specific networks by mapping genes with eQTLs overlapping insulin QTLs on chromosome 2 or 19 onto this interactome and considering only interactions across the two eQTL gene groups . Islets contain the greatest number of genes involved in a cross group interaction ( listed in Table S3 ) between the sub-networks showing linkage to chromosomes 2 or 19 ( Table 3 ) . Figure 5 illustrates the islet protein-protein interaction network constructed from gene transcripts showing linkage to either chromosome 2 , 19 or both loci . Genes contained within the islet network are significantly enriched for several gene ontology ( GO ) categories ( Table S4 ) , such as “neuron projection” ( p = 5 . 6×10−8 ) , “extracellular space” ( p = ) , and “hormone activity” ( p = 3 . 2×10−6 ) . These results demonstrate that our network identifies gene sets with common biological functions and some of these functions appear to be related to insulin secretion . As shown in Figure 5 , many genes are supported as being involved in cross group interactions and could conceivable play a critical role in regulating plasma insulin levels . To assess their potential in mediating cross group interactions and regulating clinical traits , we designed a novel ranking algorithm that integrates Trait , protein-protein Interaction , and gene Expression ( referred to as , TIE score ) to identify those genes most likely to play a critical role in insulin regulation . Instead of focusing on the property of individual genes , the TIE score incorporates an Interaction Potential ( IP ) between a protein pair . Because PPI data are not assayed in a relevant physiological context , we leverage the expression data , which is assayed in a relevant context , to weigh whether a protein interaction pair is relevant to our context of interest ( i . e . , interactions between diabetes-relevant tissues in the cross population ) . There are many post-transcriptional and post-translational modifications that may impact protein-protein interactions and their resulting functional activities . However , these modifications cannot be inferred by gene expression profiling . For a pair of proteins known to interact , we make the simplifying assumption that their protein activities and binding affinity do not change . We further assume a strong interaction potential ( IP ) if both genes are highly expressed in a mouse relative to the two genes in other F2 mice ( since protein-protein association rates depend on protein concentrations in a simple diffusion model ) ; if one or the other gene has relatively low expression , the IP will be weaker . We then calculate the correlation between the IP and trait , yielding a Trait – IP Correlation ( TIPC ) . The TIPC represents the potential of an interaction ( instead of individual genes ) in regulating a clinical trait . Using TIPC as a weight for each interaction in the protein-protein interaction network , the TIE score is then computed for each node based on the small sub-network formed by the node and its direct neighbors ( see Method for details ) . A node receives a high TIE score if it has numerous interactions with large TIPC values . TIE scores are context dependent so that a given gene may have different TIE scores in different tissues , given that the levels of its own expression and that of its interaction partners' may be different between tissues . In contrast to other network-based approaches [29]–[31] , the TIE score enables us to identify key nodes within a network that have multiple protein interactions with neighboring nodes , where such interactions are supported as exerting a strong influence on the particular clinical trait under investigation ( in our case , plasma insulin levels ) . For ranking purposes , we limit our calculation to genes with 5 or more interactions and set the TIE score to zero for the rest of the genes , given genes with too few interactions may spuriously influence the TIE score and thus lead to unreliable results . We also permutated gene expression for the nodes in the network a thousand times and calculated the empirical distribution of TIE scores to assess the significance of derived TIE scores ( see supplementary methods in Text S1 ) . Pancreatic islets had the greatest number of genes with a significant non-zero TIE score ( Table 4 , and listed in Table S5 ) . Adipose tissue had many fewer genes with a non-zero TIE-score ( listed in Table S6 ) than islets , and the other three tissues had no genes with non-zero TIE scores , suggesting that protein-protein interactions in these tissues may not be the mechanism underlying the cross-locus regulation of plasma insulin involving chromosomes 2 or 19 . The top 5 ranked genes for both adipose and islet tissues are given in Table 4 . Our result suggests that pancreatic islets contribute the most to variation in plasma insulin in our F2 cross between B6 and BTBR mice . Circulating levels of plasma triglyceride , an indicator of insulin resistance , showed no significant genetic linkage ( Figure S2 ) , suggesting that factors controlling insulin resistance may be distinct from that controlling plasma insulin in our B6XBTBR-ob/ob F2 cross . Cdkn1a ( cyclin-dependent kinase inhibitor 1 A ) , the top ranking gene in adipose tissue , is a potent cyclin-dependent kinase inhibitor ( p-value = ) . The protein binds to and inhibits the activity of cyclin-Cdk2 or –Cdk4 complexes , and thus functions as an inhibitor of cell cycle progression at G1 . The expression of this gene is tightly controlled by the tumor suppressor protein p53 , in response to a variety of stress stimuli [32] . Previous reports demonstrate that p53 expression in adipose tissue is crucially involved in the development of insulin resistance [33] . The p21 KO mouse ( Cdkn1a−/− ) showed 90% increase in fat pad weights , 70% increase of adipocyte numbers , and insulin resistance [34] . Compared to the top scores in adipose tissue , the top scores for genes in the islet network are much higher ( Table 4 ) , suggesting a greater number of genes in islet make a larger contribution to insulin variation . The top ranked gene in the islet network is App ( Amyloid β Precursor Protein ) . Successive proteolytic processing of APP by β- and γ- secretase enzymes generates the amyloid-β peptide , a primary component of amyloid plaques , which are thought to be central to the etiology of Alzheimer's disease ( AD ) [35] . Although this gene has been heavily studied by AD researchers , its relevance in type 2 diabetes is much less known . In addition to App , several other genes with high TIE scores could potentially be involved in regulating plasma insulin . Wang et al . showed that peripheral-tissue-specific knockout of Grb10 results in enhanced insulin sensitivity in vivo [36] that could be due to the loss of Grb10-mediated degradation of the insulin receptor [37] . More recently , disrupting Grb10 is shown to increase pancreatic beta cell mass and reduce beta cell apoptosis in mice [38] . The enrichment of genes previously shown to participate in the regulation of circulating insulin in the top islet gene list , such as Grb10 and insulin Ins1 , supports our use of TIE to identify novel regulators of insulin . The expression of App in islets strongly negatively correlates with plasma insulin levels in the F2 cross ( Pearson correlation R = −0 . 68 , p-value≪0 . 01 , Figure S3 ) . We have previously characterized the difference in diabetes susceptibility between the two parental mouse strains [24] . At 10 weeks of age , BTBR-ob/ob mice are diabetic , whereas B6-ob/ob remain euglycemic . BTBR-ob/ob mice have lower plasma insulin and a higher level of App in islets than B6-ob/ob mice ( difference in App expression p-value∼0 . 05 , Figure S4 ) , which is consistent with the negative correlation that we observed in the F2 cross . The distributions of the islet App expression levels as a function of genotype at the Chr . 2 and 19 loci ( Figure S5 ) indicate that F2 mice with BTBR genotypes at the two loci have higher islet App expression , consistent with App gene expression negatively correlating with plasma insulin levels . Previous work has shown that compared to wild-type mice , whole-body App knockout mice ( App−/− ) have reduced plasma glucose and elevated insulin secretion in response to an intravenous glucose injection [39] . Given that App is expressed in multiple tissues , including the brain where it may regulate neurogenesis [40] , we sought to determine if the changes observed in App−/− mice reflect direct or indirect effects of App on islet function and/or health . To assess whether the loss of App has a direct impact on islet function , we monitored insulin secretion ex vivo from pancreatic islets collected from either wild-type or App−/− mice ( Figure 6 ) . At 10 weeks of age , insulin secretion from wild-type versus App−/− mice was not different ( Figure 6A ) . However , at 17 weeks of age insulin secretion was elevated ∼2–3-fold from App−/− islets in response to glucose ( p-value<0 . 05 ) , a depolarizing concentration of KCl ( p-value<0 . 01 ) , or a membrane-permeant analogue of cAMP ( p<0 . 01 ) ( Figure 6B ) . The amount of insulin per islet ( insulin content ) was not significantly different between wild-type and App−/− mice ( Figure S6 ) . Further , insulin secretion in response to basal glucose concentration ( 1 . 7 mM ) at either 10 or 17 weeks of age was not significantly different between wild-type and App−/− mice ( Figure 6 , inserts ) . These results suggest that App directly functions as a negative regulator of insulin secretion in islets , and this only occurs in older mice .
We developed a novel network model that integrates genetic , transcription and protein-protein interaction information to pinpoint App as a key insulin regulatory molecule in pancreatic islet tissue . The computational model we developed has several unique features . Instead of pursuing cis-regulating genetic factors , it focused on networks of genes that were trans-regulated . The goal was not to identify the genetic factors whose variation at DNA level would lead to changes in circulating insulin . Instead , the model identifies networks of genes showing transcriptional changes as result of variation in the genetic factors . This is based on the assumption that the disease phenotype is at least partially mediated by these transcriptional changes . Genes identified by this approach could also have a more direct link to the disease phenotype compared to the upstream genetic factors . The model also simultaneously considered multiple loci , which enabled the study of the interactions between trans-regulated gene modules . As it is extremely common for complex disease phenotype traits to map to multiple loci , it is clear that we need models considering the joint effects of multiple loci . Ideally such models should not only be meaningful in the mathematical terms , but also provide biological insight to the possible mechanisms . Although the linear regression model indicated a joint regulation of the insulin trait , it did not generate any hypotheses on how the joint regulation occurred biologically . Compared to other network models , such as co-expression network [41] , ARACNE [9] , and Bayesian network [22] , [42] , which focus on grouping co-expression of individual genes , our method focuses on dissecting potential mechanisms of integrating information from multiple co-expression modules . By considering the protein-protein interactions across the two groups of genes , it is possible to actually identify potential molecular mechanisms involved in joint regulation . Although currently the protein-protein interaction dataset we compiled may be rather incomplete , hundreds of genes were connected by these interactions . This makes prioritizing genes for experimental validation a more important task compared to finding out what could have been missed due to incomplete protein interaction information . To prioritize the key nodes in the disease network , we developed the novel scoring system in the context of the protein interaction network . As we posit that proteins their function by interacting with their neighbors in the network , the TIE score gives a weighted estimation on how strongly the intensity of these interactions correlates with the phenotype . A gene with high TIE score suggests that the intensity of its interactions strongly correlates with the phenotype based on large numbers of interactions . Therefore , the gene is likely to regulate the trait . By integrating genetic , gene expression , and phenotypic trait information , the ranking algorithm identified biologically meaningful candidate insulin regulators . A previous publication has shown that , compared to wild-type mice , whole-body App knockout mice ( App−/− ) have elevated insulin secretion in response to an intravenous glucose injection [39] . A recent study of the cross of App transgenic mice and T2D predisposition mice shows that increased Aβ production impairs insulin signaling and accelerates insulin resistance [43] . To our knowledge , however , no other studies have demonstrated a direct effect of APP on islet function . Given that App is highly expressed in pancreatic islets [44]–[45] , we sought to determine if the changes observed in App−/− mice reflect direct or indirect effects of App on islet function . Our measurements of glucose stimulated insulin secretion in isolated islets from App KO mice confirms our network analysis and is also consistent with the causality test [13] which also indicates App as a causal gene in pancreatic islet tissue ( Figure S7 ) . The model demonstrates that App is under the regulation of multiple genetic loci , and may function as an integrator for these perturbation signals , mediating interactions between two distinct gene sets that share a common genetic architecture with plasma insulin . We have previously shown that Lepob/ob mutation exposes a strain-dependent difference in diabetes susceptibility between BTBR and B6 mice [25] . In the current study we exploited this difference and used it as a “sensitized screen” to genetically map genes and diabetes-related clinical traits that may underlie this difference . This approach allowed us to identify App as a key negative regulator of insulin secretion from pancreatic islets . In this study , we compared wild-type and App−/− mice to test for a direct role of App in insulin secretion in mice not expressing the Lepob/ob mutation . In these studies , the loss of App resulted in enhanced insulin secretion , consistent with the strong negative relationship between islet App and circulating insulin across the F2 samples . These results suggest that while leptin deficiency was critical in revealing the islet network involving App and circulating insulin , it was not required to demonstrate the direct role of App in insulin secretion . Our results , which demonstrate a difference in insulin secretion between islets collected from wild-type and those collected from App−/− mice at 17 weeks , but not 10 weeks of age , implies an age-dependence for the role that App plays in the islet . However , studies in mouse [46] and human islets [47] have not reported an age-dependent change in App expression . It is possible that proteolytic processing of App mediated by the beta- and gamma-secretase enzymes , or other forms of post-translational modification , are necessary for App to regulate insulin secretion . Mouse and rat beta cells are more sensitive to oxidative stress than human beta cells [48]–[49] , due to the relatively higher expression of antioxidant enzymes in human beta cells [47] , [50] . We showed that the sub-network regulating plasma insulin level variation ( Figure 5 ) is enriched for GO categories “neuron projection” ( p = 5 . 6×10−8 ) , “extracellular space” ( p = 4 . 03×10−7 ) , and “hormone activity” ( p = 3 . 2×10−6 ) . Genes involved in the stress response process are not enriched in the subnetwork . Recent RNAseq data [47] suggests that APP robustly expresses in human islet cells . In addition , it has been shown that aggregated amyloid-β peptide as well as other proteins have been detected at higher levels in pancreatic islets of T2D patients comparing to healthy control people [51] . These suggest that the subnetwork and key regulators in mouse islet we identified in the F2 cross are expected to be relevant in human islets . Our findings support the hypothesis that APP contributes to the common pathogenesis of AD and T2D [52] . For the future development , ( 1 ) a generalized multi-way interaction model is needed to capture complex interaction networks underlying complex traits such as plasma insulin; ( 2 ) additional experiments are needed to systematically validate candidate genes ( such as genes in Table S5 and genes connected to App in Figure 5 ) for their roles in affecting β-cell function which in turn affect insulin production and insulin secretion; ( 3 ) the molecular mechanism of age-dependent App regulating insulin secretion is warrant further study . In conclusion , using an integrative analysis of gene expression , genotypes , and phenotypic traits of the B6xBTBR ob/ob F2 cross , we showed that plasma insulin is modulated by the variation of multiple genetic factors , presumably through expression changes of hundreds of genes in multiple tissues . Our approach focused on revealing the underlying disease network across loci and tissues . The model predicted that App acts in pancreatic islets to affect plasma insulin . This prediction was tested in isolated islets where the knockout of App was associated with increased insulin secretion . Considering App is known for Alzheimer's disease development and a strong association between T2D and AD , our findings point to a potential mechanism through which these two diseases are linked .
All animal studies were conducted at the University of Wisconsin in the Biochemistry Department in accordance with NIH guidelines and the University of Wisconsin Research Animals Resource Center . App−/− mice were purchased from the Jackson Labs ( stock number 004133 ) . C57BL/6J ( B6 ) ob/+ male and BTBR ob/+ female mice were bred to obtain F1 ob/ob mice [24] . Leptin deficiency causes infertility [53]–[54] . To restore fertility to F1 ob/ob mice , at approximately 4 weeks of age the F1 ob/ob male and female mice each received subcutaneous transplants of white-adipose tissue from lean ( leptin-competent ) litter mate donor mice , resulting in the restoration of fertility in >90% of the F1 ob/ob mice . The F1 ob/ob mice were then bred to produce a panel of ∼550 F2 ob/ob mice . At 10 weeks of age , the F2 ob/ob mice were sacrificed and tissues collected ( islet , white adipose , liver , gastroc muscle , and hypothalamus ) . Gene expression was profiled on an Agilent custom murine gene expression microarray consisting of 4 , 732 control probes and 39 , 558 non-control oligonucleotides extracted from mouse Unigene clusters and combined with RefSeq sequences and RIKEN full-length cDNA clones . All F2 mice were genotyped with the Affymetrix 5 K SNP array , which identified ∼2 , 000 SNPs that were polymorphic between B6 and BTBR mice that spread uniformly across genome . Various clinical traits were measured for each mouse just prior to sacrifice . See Supplementary materials ( Text S1 ) for additional description of methods . Intact pancreatic islets were isolated from mice using a collagenase digestion procedure [55] . Briefly , the mice were sacrificed and the pancreases immediately inflated with 5 ml Hanks Buffered Salt Solution ( HBSS ) supplemented with 0 . 02% BSA and collagenase ( 0 . 5 mg/ml ) . After inflation , the pancreata were carefully dissected from the mice , placed in 25 ml of HBSS/BSA/collagenase , and incubated for 16 min at 37°C , with intermittent agitation . A ficoll gradient was used to partially purify islets from the digested pancreata , and further purified by hand-picking the islets viewed under a stereo-microscope . Media used for isolation and insulin secretion studies was a Krebs-Ringer Bicarbonate Buffer ( KRB ) containing ( in mM ) : 118 . 41 NaCl , 4 . 69 KCl , 2 . 52 CaCl2 , 1 . 18 MgSO4 , 1 . 18 KH2PO4 , 25 NaHCO3 , and 5 HEPES . For the measurement of plasma insulin , all mice were fasted beginning at 8 AM , by transfer to a clean cage and provided water ad libitum . Approximately 4 hours later , ∼0 . 1 ml of blood was collected via retro-orbital draw , transferred to a tube containing 3 µl of 0 . 5 mM EDTA as the anti-coagulant , and then centrifuged ( 5 mins , 10 , 000×g , 4°C ) to isolate plasma . The level of plasma insulin was measured as described previously [24] . Briefly , high-binding plates ( Corning ) were coated overnight with 3 µg/mL of an anti-insulin antibody ( D6C4 , Research Diagnostics ) , blocked with PBS containing 4% RIA-grade BSA ( Sigma ) for 1 h and then incubated for 1 h with insulin standards ( Linco Research , 0 . 1–10 ng/mL ) or 25 µl whole plasma . An anti-proinsulin antibody ( 1 µg/ml of D3E7-BT , Research Diagnostics ) was added and incubated for an additional hour . After extensive washing ( 50 mM Tris , 0 . 2% Tween-20 , pH 8 . 0 ) , 1 µg/mL of streptavidin-HRP ( Pierce ) in PBS/0 . 1% BSA was added and incubated for 30 min . Following additional washes , 16 µmol/ml of o-phenylenediamine ( Sigma ) , dissolved in citrate buffer ( 0 . 1 M citrate-phosphate , 0 . 03% H2O2 at pH 5 . 0 ) , was added and incubated for 30 min; 0 . 18 M sulfuric acid was used to quench the reaction . Absorbance at 492 nm was determined by a plate reader ( Ultra 384 TECAN ) . Insulin contents in plasma were calculated by comparison to known standards . Three islets of equivalent size were placed in 12×75-mm glass tubes , where the bottom of the tube was formed by a 62-µm mesh ( Elko Filtering Co . ) . The 12×75-mm tubes were transferred to 16×100-mm tubes containing 1 ml of KRB with 1 . 7 mM glucose and 0 . 5% BSA and pre-incubated at 37°C for 45 min . Following the pre-incubation , the 12×75-mm tubes were transferred to a fresh 16×100-mm tube containing 1 ml KRB supplemented with 1 . 7 , 11 . 1 or 16 . 7 mM glucose , with or without additional KCl or 8-Br-cAMP as indicated . For studies where 40 mM KCl was added to the secretion medium , NaCl was reduced to 78 . 41 mM to maintain osmolarity . Following a 45-min incubation period at 37°C , the 12×75-mm tubes were transferred to a fresh tube containing 1 ml of HCl-ethanol-water ( 1∶50∶14 ) to extract cellular insulin from the islets . The incubation media was collected and frozen for insulin determination by ELISA . Insulin trait cQTL and gene eQTL mapping were performed using scanone function in R package R/qtl [56] . The causality test was described previously [13] and a Bayesian network version was used to conduct the test . The global mouse protein-protein interaction network was collected as described previously [20] . For two interacting proteins and , we define an Interaction Potential ( IP ) for the protein pair in an individual mouse as is the gene expression in mouse , so is for gene . and are the maximum and minimum tissue-specific expression observed across the entire F2 panel for gene . The calculation assumes the variation of protein abundance is approximated by its gene expression , and IP is proportional to the relative levels of the two proteins . Thus , a reduction in gene expression would lead to reduced protein-protein interaction for a given pair and vice versa . The predictive power of for a specific trait value ( such as plasma insulin in our case ) can be calculated as the Trait Interaction Potential Correlation ( TIPC ) . We define , where is the trait in consideration , is the function for calculating the correlation coefficient , and is the union of all the mice . We consider both network topology and TIPC scores to rank genes . Assume protein interacts with a set of proteins , the TIE score is computed aswhere and is the number of proteins contained in . Gene with a high TIE score indicates it has large number of interactions and for its direct neighbors , the average correlation between interaction potential and trait is also high .
|
Alzheimer's disease and type 2 diabetes are two common aging-related diseases . Numerous studies have shown that the two diseases are associated . However , the mechanisms of such connection are not clear . Both diseases are complex diseases that are induced by multiple genetic factors and the environment . To understand the molecular network regulated by complex genetic factors causing type 2 diabetes , we constructed an F2 intercross comprised of >500 mice from diabetes-resistant and diabetic mouse strains . We measured genotypes , clinical traits , and expression profiling in five tissues for each mouse . We then performed an integrative analysis to investigate the inter-relationship among genetic factors , expression traits , and plasma insulin , a hallmark diabetes trait , and developed a novel method for inferring key regulators for regulating plasma insulin . In islets , the Alzheimer's gene App was identified as a top candidate regulator . Islets from 17-week-old , but not 10-week-old , App knockout mice showed increased insulin secretion in response to glucose , in agreement with the predictions of the network model . Our result provides a novel hypothesis on the mechanism for the connection between two aging-related diseases: Alzheimer's disease and type 2 diabetes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"microarrays",
"systems",
"biology",
"genetics",
"of",
"disease",
"genomics",
"computational",
"neuroscience",
"gene",
"expression",
"genetics",
"regulatory",
"networks",
"molecular",
"genetics",
"biology",
"computational",
"biology",
"gene",
"networks",
"genetics",
"and",
"genomics",
"gene",
"function"
] |
2012
|
Integrative Analysis of a Cross-Loci Regulation Network Identifies App as a Gene Regulating Insulin Secretion from Pancreatic Islets
|
Microorganisms are capable of communication and cooperation to perform social activities . Cooperation can be enforced using kind discrimination mechanisms in which individuals preferentially help or punish others , depending on genetic relatedness only at certain loci . In the filamentous fungus Neurospora crassa , genetically identical asexual spores ( germlings ) communicate and fuse in a highly regulated process , which is associated with fitness benefits during colony establishment . Recognition and chemotropic interactions between isogenic germlings requires oscillation of the mitogen-activated protein kinase ( MAPK ) signal transduction protein complex ( NRC-1 , MEK-2 , MAK-2 , and the scaffold protein HAM-5 ) to specialized cell fusion structures termed conidial anastomosis tubes . Using a population of 110 wild N . crassa isolates , we investigated germling fusion between genetically unrelated individuals and discovered that chemotropic interactions are regulated by kind discrimination . Distinct communication groups were identified , in which germlings within one communication group interacted at high frequency , while germlings from different communication groups avoided each other . Bulk segregant analysis followed by whole genome resequencing identified three linked genes ( doc-1 , doc-2 , and doc-3 ) , which were associated with communication group phenotype . Alleles at doc-1 , doc-2 , and doc-3 fell into five haplotypes that showed transspecies polymorphism . Swapping doc-1 and doc-2 alleles from different communication group strains was necessary and sufficient to confer communication group affiliation . During chemotropic interactions , DOC-1 oscillated with MAK-2 to the tips of conidial anastomosis tubes , while DOC-2 was statically localized to the plasma membrane . Our data indicate that doc-1 , doc-2 , and doc-3 function as “greenbeard” genes , involved in mediating long-distance kind recognition that involves actively searching for one’s own type , resulting in cooperation between non-genealogical relatives . Our findings serve as a basis for investigations into the mechanisms associated with attraction , fusion , and kind recognition in other eukaryotic species .
Microbes engage in a wide variety of cooperative interactions to perform complex , multicellular , coordinated activities such as dispersal , foraging , nutrient acquisition ( including virulence ) , organismal defense , and production of multicellular structures such as biofilms , networks , or fruiting bodies [1–4] . Unlike larger organisms , many of the processes involved in microbial cooperation take place extracellularly in the public space , so that public goods produced by cooperative cells are particularly vulnerable to exploitation by cheaters ( which benefit from the availability of public goods without producing them ) [3 , 5] . Microbes have evolved multiple mechanisms for enforcing cooperation , by performing differential actions to others ( i . e . , rewarding cooperators and/or penalizing cheaters ) according to kinship ( i . e . , genome-wide relatedness ) or kind ( i . e . , phenotypic similarity caused by genetic relatedness at certain loci ) [1 , 6 , 7] . In fact , much discrimination in microbes appears to be based on kind rather than kin [3 , 7] , and many of the frequency-dependent processes commonly observed in microbes can be interpreted as kind discrimination , as they depend on expressing a trait that has differential effects on bearers and non-bearers [8–11] . Under this model , cooperation can involve kin or non-kin individuals as long as they share a single cooperative gene or set of genes; such genes are termed “greenbeard” genes . Individuals with a given greenbeard gene can identify the presence of that greenbeard gene in other individuals , resulting in a change in activity or interaction [12] . Kind discrimination can be divided into “harming” and “helping” types [11] . “Harming” kind discrimination includes the poison—antidote system , which is widespread among bacteria and some archaea , and involves releasing a bacteriocin that can be rendered ineffective by related strains expressing an antidote protein , but which kills strains lacking it ( reviewed in [13] ) . “Helping” kind discrimination is exemplified by the slime mold Dictyostelium discoideum or the yeast Saccharomyces cerevisiae , for which discrimination involves cell adhesion proteins that are important for adherence of amoeba in aggregation streams or for flocculation , respectively [7 , 14 , 15] . In many microbial eukaryotes , somatic growth is a form of cooperation: all somatic cells are “hopeful reproductives" ( i . e . , they retain the potential to sexually reproduce ) , but most of them will never engage in sexual reproduction and instead help other cells to reproduce , which provides a direct benefit to the individual , and an indirect benefit to related individuals upon somatic fusion [16] . However , with somatic fusion , soma becomes a public good that is vulnerable to exploitation by cheaters [17–19] . In filamentous fungi , somatic fusion can occur within or between clonemates: an interconnected mycelia network can be formed via cell fusion between germinated asexual spores ( germlings ) [20–22] and/or between hyphae in a mature colony . The benefits of fungal somatic fusion have been associated with the sharing of cytoplasm , organelles ( including nuclei ) , nutrients , and other resources to ensure rapid spatial expansion [23–26] , intra-organismal communication , mitotic recombination ( especially for highly clonal species [27] ) , redistribution of water and nutrients , and general homeostasis within the mycelium [16 , 25 , 28–32] . Somatic fusion can also occur in filamentous fungi via hyphal fusion between different colonies , potentially leading to the presence of genetically different nuclei in a common cytoplasm ( heterokaryon ) . The cost of somatic fusion has been associated with the transmission of infectious cytoplasmic elements and mycoviruses , which are widespread among fungi [33–35] . Within a heterokaryon , allorecognition processes determine the fate of the fused cells: compatible genotypes lead to a heterokaryon indistinguishable from a homokaryotic colony , while heterokaryotic cells resulting from the fusion of incompatible genotypes are rapidly compartmentalized and undergo programmed cell death , termed vegetative ( or heterokaryon ) incompatibility [36–39] . Empirically , vegetative incompatibility in filamentous fungi has been shown to prevent somatic parasitism and reduce the risk of transmission of selfish nuclei and cytoplasmic elements [40–42] . In Neurospora crassa , the molecular basis of chemotropic interactions and cell fusion between genetically identical germlings has been studied extensively , but processes involved in recognition between genetically non-identical germlings have not been investigated . In genetically identical germlings , chemotropic interactions are initiated when germlings are in close proximity ( ~15 μm ) , and are associated with redirected growth and cell fusion via specialized structures termed conidial anastomosis tubes ( CATs ) ( Fig 1A ) [21 , 43] . During chemotropic interactions , the mitogen-activated protein kinase ( MAPK ) signal transduction protein complex ( NRC-1 , MEK-2 , MAK-2 , and the scaffold protein HAM-5 ) assembles/disassembles at the CAT tip of communicating germlings with perfectly out of phase dynamics to SOFT , a scaffold protein for components of the MAPK cell wall integrity pathway [44–48] . For example , if SOFT is at the CAT tip of one germling , NRC-1/MEK-2/MAK-2/HAM-5 complex is at the CAT tip of its partner germling; switching between MAK-2 complex and SOFT at a single CAT tip occurs approximately every 4–5 min ( Fig 1A ) . The spatiotemporal coordination of the MAK-2 signal transduction complex versus SOFT at CATs during chemotropic interactions is postulated to allow genetically and developmentally identical cells to coordinate their behavior and achieve mutual attraction and fusion , while avoiding self-stimulation [44 , 49] . Here , we investigate recognition interactions between genetically different germlings using a population of N . crassa . Within this population , we defined distinct communication groups and show that genetically different germlings can distinguish each other without physical contact , in a process that involves actively searching for one’s own type . Communication groups were associated with haplotypes at three linked loci , doc-1 , doc-2 , and doc-3 . Alleles at doc-1 , doc-2 , and doc-3 were highly divergent between haplotypes , and they showed transspecies polymorphisms consistent with long-term balancing selection caused by negative frequency-dependent selection ( i . e . , rare allele advantage ) . Live cell imaging showed that DOC-1 oscillates with the conserved MAK-2 signal transduction pathway . Thus , here we describe the identification and characterization of a form of assortative kind recognition that involves multiple alleles at the greenbeard genes doc-1 , doc-2 , and doc-3 and that acts at a distance by preventing chemotropic interactions between non-kind germlings from different communication groups . Our findings reveal a heretofore underappreciated complexity in fungal communication and serve as a basis for investigations into mechanisms associated with long-distance kind recognition in other eukaryotic species .
Kind discrimination was neglected in previous studies on germling communication in N . crassa , as strains were used whose genetic background was identical to the commonly used laboratory strain ( FGSC 2489 ) [43 , 44 , 50] . To assess whether germlings of different genetic backgrounds can undergo productive chemotropic interactions , we took advantage of a N . crassa population isolated from Louisiana , United States; the laboratory strain ( FGSC 2489 ) is a member of this population [51] . RNAseq data showed a substantial level of polymorphism ( on order of two single nucleotide polymorphisms [SNPs] per kbp ) , while analyses of population structure revealed no subdivision [51–53] . We randomly picked 14 isolates from this population and analyzed chemotropic interactions ( defined as reoriented growth of germlings toward each other; Fig 1A ) between genetically identical germlings from the same isolate ( self-communication ) versus chemotropic interactions between germlings from the 14 isolates and FGSC 2489 ( non-self-communication ) using differential fluorescence labeling ( see Materials and Methods ) . Self-communication frequencies among the wild isolates varied between approximately 50% and 95% ( Figs 1A , 1B and S1 ) . However , when germlings of these 14 wild isolates were paired with FGSC 2489 ( non-self-communication ) , some pairings showed very low communication frequencies , while others showed communication frequencies with FGSC 2489 that were similar to self-communication frequencies ( Figs 1A , 1B and S1 ) . Importantly , this communication phenotype was not linked to the mating type of the strains ( S1 Table ) . We therefore assessed self and non-self germling communication phenotype of the remaining members of the Louisiana population ( 95 strains ) ( see Materials and Methods; S1 Table ) . From these analyses , three communication groups were defined . While genetically identical and non-identical germlings within a communication group showed robust chemotropic interactions , germlings from different communication groups , even when in close proximity , grew past each other to find a germling of their own communication group ( S1 and S2 Movies ) . The first communication group ( CG1 ) contained 29 strains , which showed similar communication frequencies between and within strains and which included FGSC 2489 ( Fig 1C , orange ) . The second communication group ( CG2 ) contained 51 strains ( Fig 1C , green ) , while the third communication group ( CG3 ) contained 21 strains ( Fig 1C , blue; S1 Table ) . The remaining nine strains ( Fig 1C , black ) did not produce sufficient asexual spores to determine CG affiliation . These observations indicated that the germling communication trait in N . crassa functioned in assortative kind recognition and occurs at a distance . To determine whether communication groups are unique to the Louisiana population , we used tester strains for each communication group ( FGSC 2489 , CG1; JW262 , CG2; P4483 , CG3; framed in Fig 1C ) and evaluated communication frequencies with other N . crassa population samples ( isolates from Haiti , Panama , Costa Rica , Puerto Rico , Texas , Florida , Venezuela , Guyana ) . All of the wild isolates from these different N . crassa populations communicated with one of the three Louisiana communication group tester strains ( CG1 , CG2 , or CG3; S1 Table ) . Thus , communication groups were not unique to the Louisiana population , but also occurred in other wild populations of N . crassa . Based on the distribution of communication groups in the Louisiana population , we reasoned that genes that conferred kind recognition in N . crassa functioned as a Mendelian trait . To test this hypothesis , we used crossings to determine the number of loci mediating CG affiliation , making use of the fact that the affiliation of strains in the different CGs does not affect sexual compatibility [54] . We crossed a CG1 strain ( FGSC 2489 ) with a CG2 strain ( JW258 ) , a CG1 strain ( FGSC 2489 ) with a CG3 strain ( D113 ) , and a CG2 strain ( JW242 ) with a CG3 strain ( D113 ) . In all crosses , the CG phenotype of the progeny segregated approximately 1:1 , with approximately one-half of the progeny communicating with one parental strain and the second half of the progeny communicating with the other parent , consistent with our prediction that a single locus or closely linked loci were involved in kind recognition and determined CG affiliation . To identify the CG locus , we performed a bulk segregant analysis followed by whole genome resequencing of progeny from a cross between a CG1 strain ( FGSC 2489 ) and a CG2 strain ( JW258 ) . Genomic DNA from 46 CG1 progeny or 46 CG2 progeny was isolated , pooled , and sequenced , revealing a ~100 kbp region on the right arm of linkage group V that showed segregation of SNPs between CG1 versus CG2 progeny at ~100% frequency , which was embedded within a larger divergent region of ~450 kbp ( Fig 2A ) . A random SNP distribution of ~50% was observed for the remaining six linkage groups . We used resequencing data from 26 wild isolates from the Louisiana population [55] to define allelic sequences at the CG locus within the ~100 kbp region . Among these 26 strains , seven isolates were members of CG1 , 15 isolates were in CG2 , and three isolates were members of CG3 , while one isolate did not produce sufficient asexual spores to determine CG affiliation ( asterisks in Fig 1C; S1 Table ) . Analysis of the ~100 kbp interval in the genomes of these 26 isolates revealed a 14 kbp region that showed five different genomic rearrangements that spanned four loci , NCU07191 to NCU07194 ( gene nomenclature based on the reference genome from FGSC 2489 [56] ) , referred to as communication group haplotype 1 through 5 ( CGH1–5 ) ( Fig 2B ) . Of the 26 isolates , seven showed a CGH1 organization , five showed a CGH2 organization , three had a CGH3 organization , four had a CGH4 organization , and seven had a CGH5 organization . The CGH2 and CGH4 strains contained a duplication of NCU07192 ( doc-3 ) , while CGH5 strains did not contain NCU07192 ( Fig 2B ) . Inversions of NCU07192 ( CGH2 and CGH3 isolates ) or the entire genetic interval between NCU07191 and NCU07194 were also observed ( CGH4 and CGH5 isolates ) . Of the genes within this region , NCU07191 and NCU07192 showed ~43% DNA sequence identity , suggesting that they are paralogous . Paralogy was also supported by analyses using OrthoMCL ( OG5_241519 ) [57] . To determine whether structural differences between CGHs were also associated with nucleotide differences , we used sequence alignments to characterize the nature and level of variability at genes within the haplogroups . Among five loci in the genetic interval associated with CGH , NCU07190 , NCU07193 , and NCU07194 displayed a high level of conservation among all 26 isolates ( >90% DNA sequence identity with few nucleotide substitutions; Fig 2B; S2 Table ) . In contrast , NCU07191 and NCU07192 displayed high levels of allelic variability among the 26 isolates , with alleles falling into five main groups , which correlated with the genomic rearrangements among the five CGHs . Alleles at NCU07191 and NCU07192 showed only ~50% DNA sequence identity , with members of the different CGHs being highly divergent ( 0 . 22 to 0 . 74 differences/bp between NCU07191 alleles and 0 . 26 to 0 . 87 differences/bp between NCU07192 alleles; S2 Table ) . The predicted proteins encoded by NCU07191 and NCU07192 among members of the different CGH groups were also highly variable , with only ~35% amino acid identity ( CGH1 versus all other CGH isolates ) , with few regions showing high conservation in all of the predicted NCU07191 or NCU07192 proteins ( S2 Fig ) . In contrast , isolates within a single CGH showed DNA and amino acid identity at NCU07191 and NCU07192 that were comparable with the rest of the genome ( up to 99% DNA and over 95% amino acid sequence identity ) ( S2 Fig ) , with the exception of the CGH1 isolates . Within the CGH1 isolates , alleles at NCU07191 and NCU07192 fell into two different subgroups ( CGH1A and CGH1B ) , with ~70% DNA and ~60% amino acid sequence identity between members of the two subgroups ( S2 Table ) . DNA sequence alignments of the genetic interval between NCU07191 and NCU07194 of the CGH1 isolates indicated that there were CGH1A- and CGH1B-specific indels and SNPs in the intergenic region between NCU07190 and NCU07193 ( S3 Fig ) . Similarly , both CGH-specific SNPs resulting in amino acid substitutions and CG-specific indels differentiated isolates between the different CGH groups ( S2 Fig ) . The presence of five genomic CGHs with only three phenotypically distinguishable communication groups within the Louisiana population prompted us to reevaluate the communication phenotype of the 26 sequenced strains . For members of CG1 and CG3 , the germling communication phenotype was completely correlated with CGH ( S1 Table; Fig 2B ) ; no difference in germling communication frequency was observed in isolates between the subgroups CGH1A and CGH1B . Unlike the CG1/CGH1 and CG3/CGH3 strains , the 15 strains defined as CG2 displayed multiple different genomic arrangements in this region ( CGH2 , CGH4 , and CGH5; Fig 2B ) . However , using an isolate from each defined communication group ( FGSC 2489 , CG1; JW262 , CG2; P4483 , CG3 ) , we identified a fourth phenotypic communication group ( CG4 ) . Germlings within this communication group ( D111 , JW179 , P4479 , P4489; S1 Table ) showed very low communication frequencies with CG1 germlings , but underwent robust chemotropic interactions with both CG2 and CG3 germlings ( S1 Table ) . These CG4 strains all showed the same genomic organization in the NCU07191 to NCU07194 genetic interval ( CGH4 ) , and with high DNA and amino acid sequence identity ( >99% ) . Thus , of the 26 sequenced strains , seven isolates fell into CG1 , with a CGH1 genomic organization; 11 isolates fell into CG2 , with a CGH2 or CGH5 genomic organization; three isolates fell into CG3 , with a CGH3 genomic organization; and four isolates were CG4 with a CGH4 genomic organization . An additional isolate ( JW246 ) did not produce asexual spores , but had a CGH2 genomic organization . We refer to NCU07191 , NCU07192 , and the duplicated version of NCU07192 as determinant of communication 1 , 2 , and 3 ( doc-1 , doc-2 , and doc-3 , respectively ) . doc-1 , doc-2 , and doc-3 encode predicted hypothetical proteins ( DOC-1 , 828 aa; DOC-2 , 839 aa; DOC-3 , 920 aa ) . All of the DOC-2 proteins from the different CGH groups have a predicted OmpH-like outer membrane protein domain , although with some variability in conservation ( S2 Fig ) , while DOC-1 and DOC-3 lack any identifiable functional domains . DOC-1 and DOC-2 contain one or two predicted transmembrane domains , while no transmembrane domain was predicted for DOC-3 . Conserved homologs of doc-1 , doc-2 , and doc-3 were identifiable by BLAST in the Sordariales ( order within the class Sordariomycetes in the division Ascomycota ) , but were not obvious in more distantly related fungal species . The association of communication group phenotype with CGH supported the hypothesis that the doc genes confer communication group specificity . To evaluate this hypothesis , we examined strains carrying deletions of doc-1 or doc-2 for communication phenotype . Strains carrying deletions of doc-1 ( Δdoc-1 ) or doc-2 ( Δdoc-2 ) in the CG1 background ( FGSC 2489 ) [58] were macroscopically indistinguishable from FGSC 2489 . To determine germling communication group phenotype in these deletion strains , we constructed Δdoc-1 and Δdoc-2 strains carrying a gene encoding cytoplasmic green fluorescent protein ( GFP ) . Conidia of the communication group tester strains ( FGSC 2489 , JW262 , or P4483 ) were stained with the membrane-selective endocytic dye FM4-64 , mixed with the Δdoc-1 ( GFP ) or Δdoc-2 ( GFP ) strains , and subsequently analyzed for germling communication frequencies ( Fig 3A ) . These analyses revealed that both the Δdoc-1 and Δdoc-2 germlings were impaired in self-communication: communication between isogenic Δdoc-1 germlings was reduced to 48 ± 14% , as compared to the parental strain frequency of 84 ± 7% , while self-communication frequency in Δdoc-2 germlings was reduced to 37 ± 16% . Although self-communication frequency was reduced in Δdoc-1 and Δdoc-2 germlings , non-self-communication frequencies with germlings from the different communication group tester strains were similar to each other and to the self-communication frequencies of each deletion strain . For example , when Δdoc-1 germlings were paired with either CG2 or CG3 germlings , non-self-communication frequencies were similar or even slightly higher than self-communication frequencies of Δdoc-1 germlings ( 65 ± 7% for Δdoc-1 + CG2 germlings; 71 ± 7% for Δdoc-1 + CG3 germlings; Fig 3A ) . Similarly , non-self-communication frequencies of the Δdoc-2 germlings with CG2 and CG3 tester strains were not significantly different from self-communication frequencies ( Fig 3A ) , although communication frequencies of Δdoc-2 germlings with the parental CG1 strain were higher ( 48 ± 6% ) than with the CG2 ( 18 ± 9% ) or CG3 tester strains ( 12 ± 5% ) . These data indicate that doc-1 and doc-2 are essential for mediating communication group discrimination in N . crassa . Since both the Δdoc-1 and Δdoc-2 mutants showed reduced germling communication frequencies , but no significant difference in the frequency of communication with members of the three communication groups , we hypothesized that a Δdoc-1 Δdoc-2 double mutant would be completely deficient in communication and cell fusion . To test this hypothesis , we created a Δdoc-1 Δdoc-2 mutant by homologous recombination ( see Materials and Methods ) . As with the Δdoc-1 or Δdoc-2 single mutants , the Δdoc-1 Δdoc-2 mutant was morphologically indistinguishable from its parental strain ( FGSC 2489 ) , a phenotype different than other fusion mutants , which display a “flat” phenotype [50] . To our surprise , unlike the single Δdoc-1 or Δdoc-2 mutants , the Δdoc-1 Δdoc-2 mutant showed a self-communication frequency that was indistinguishable from its parental strain ( ~85%; Fig 3A ) . Even more surprisingly , the communication frequency between Δdoc-1 Δdoc-2 germlings and their otherwise isogenic parental CG1 strain ( FGSC 2489 ) was extremely low ( ~2% ) . The Δdoc-1 Δdoc-2 germlings also showed very low communication frequency with the CG3 tester strain ( P4483 , ~8%; Fig 3A ) . In contrast , the Δdoc-1 Δdoc-2 germlings communicated fairly well with the CG2 tester strain ( JW262 , ~50% ) , although still significantly less than self-communication frequencies of Δdoc-1 Δdoc-2 germlings . The communication phenotype of the Δdoc-1 Δdoc-2 germlings suggested that DOC-1 /DOC-2 negatively regulate germling communication behavior and that removal of DOC-1/DOC-2 resulted in the generation of a new communication group . We further tested this hypothesis by evaluating the communication phenotype of Δdoc-1 Δdoc-2 germlings with additional members of CGH2 , CGH3 , CGH4 , and CGH5 . A CGH3 ( D113 ) strain and CGH4 strains ( D111 , P4489 ) showed very low communication frequencies with Δdoc-1 Δdoc-2 germlings ( ~10% ) , while CGH2 strains ( JW258 , P4463 ) showed a reduction in communication frequency with Δdoc-1 Δdoc-2 germlings , as was observed for the CG2 tester strain JW262 ( ~50%; S4 Fig ) . However , in contrast to members of the other CGH groups , germling communication frequencies of the Δdoc-1 Δdoc-2 mutant with members of CGH5 ( JW75 , JW220 , and JW242 ) were identical to self-communication frequencies ( Fig 3B; S4 Fig ) . These data showed that the Δdoc-1 Δdoc-2 mutant behaved exactly like members of CGH5 , which contained one copy of doc-1 but no copy of doc-2 . Thus , the phenotype of the Δdoc-1 Δdoc-2 mutant with the CGH5 isolates defined a new communication group , termed CG5 . The availability of isogenic strains that only differed in communication behavior ( FGSC 2489 , CG1; Δdoc-1 Δdoc-2 , CG5 ) allowed us to assess whether just the genetic difference at doc-1 and doc-2 was sufficient to affect the formation of heterokaryons ( i . e . , a syncytium of two or more genetically different nuclei ) , which is mediated by both germling and hyphal fusion . We introduced auxotrophic markers ( his-3 or pyr-4 ) into a Δdoc-1 Δdoc-2 strain and first evaluated its ability to form ( Δdoc-1 Δdoc-2; his-3 + Δdoc-1 Δdoc-2; pyr-4 ) heterokaryotic colonies as compared to strains isogenic to their parent ( FGSC 2489 ) but that carry complementary auxotrophic markers ( his-3 or ad-3B ) ( S4 Table ) . As shown in Fig 3D , heterokaryon formation was indistinguishable between ( Δdoc-1 Δdoc-2; his-3 + Δdoc-1 Δdoc-2; pyr-4 ) strains and ( his-3 + ad-3B ) strains . In contrast , heterokaryon formation was drastically reduced when Δdoc-1 Δdoc-2; his-3 conidia were mixed with ad-3B conidia ( Figs 3D and S4E ) . These data indicated that even if communication and cell fusion were essential for survival of N . crassa , differences in communication group affiliation almost completely prevented cooperation via heterokaryon formation . To determine if doc-1 and doc-2 were sufficient for communication group affiliation , we cloned the doc-1 and doc-2 alleles from a CG3 strain ( P4471; doc-1CG3 and doc-2CG3 ) and targeted them to the his-3 locus in the Δdoc-1 Δdoc-2 mutant . The Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) strain was macroscopically indistinguishable from the laboratory strain from which it was derived ( FGSC 2489 ) . Germlings from Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) showed high self-communication frequencies ( ~80%; Fig 3A and 3C ) . However , Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) germlings showed greatly reduced communication frequency with their parental Δdoc-1 Δdoc-2 strain , showing that doc-1CG3 and doc-2CG3 were functional in this strain ( Fig 3C ) . Germlings from the Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) strain also showed low communication frequency with the CG1 tester strain ( FGSC 2489 ) , but showed some communication with the CG2 tester strain ( JW262 ) , although germling communication frequencies were reduced ( ~50%; Fig 3A ) . However , the Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) germlings communicated well with the CG3 tester strain ( P4483 ) and the donor for doc-1CG3 and doc-2CG3 alleles ( P4471 ) ( Figs 3A and S4D ) . These data indicated that addition of doc-1CG3 and doc-2CG3 to the Δdoc-1 Δdoc-2 mutant was sufficient to switch communication group from CG5 to CG3 . To confirm that the Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) germlings no longer belonged to CG5 , we tested communication frequencies of Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) germlings with a CG5 strain ( JW242 ) . In contrast to Δdoc-1 Δdoc-2 germlings , Δdoc-1 Δdoc-2 ( his-3::doc-1CG3 doc-2CG3 ) germlings showed low communication frequency with JW242 germlings ( ~15%; S4C Fig versus S4D Fig ) . Thus , swapping doc-1 and doc-2 alleles from a member of CGH1 with doc-1 and doc-2 alleles from a member of CGH3 was sufficient to switch communication group . Self-recognition between isogenic germlings requires the MAK-2 MAP kinase complex ( NRC-1 , MEK-2 , MAK-2 , and the scaffold protein HAM-5 ) , which oscillates to the tips of conidial anastomosis tubes and completely out-of-phase with SOFT , a scaffold protein for the cell wall integrity MAPK pathway ( Fig 1A ) [45–47 , 59 , 60] . To determine the cellular location of DOC-1 and DOC-2 during germling communication , and its relationship with temporal patterns of signaling , we constructed strains bearing doc-1-gfp and doc-2-gfp alleles that were fully functional in restoring communication frequencies in Δdoc-1 or Δdoc-2 germlings , respectively ( S4 and S5 Figs ) . In self pairings between Δdoc-1 ( doc-1-gfp ) germlings , DOC-1-GFP localized to intracellular punctae , which oscillated to the tips of conidial anastomosis tubes during chemotropic interactions ( Fig 4A ) , with an interval of 8–10 min ( S3 Movie; S5 Fig ) , an identical oscillation pattern to that of the MAK-2 complex and SOFT [44] . To investigate whether DOC-1 oscillates with MAK-2 or with SOFT during chemotropic interactions , we analyzed DOC-1-GFP oscillation in germlings undergoing chemotropic interactions with germlings bearing MAK-2-mCherry or SOFT-mCherry; DOC-1 oscillated with MAK-2 , but completely out of phase to SOFT ( Figs 5A , 5B and S5; S3 Movie ) . A heterokaryotic strain bearing both DOC-1-GFP and MAK-2-mCherry confirmed these observations ( Fig 5C ) . In hyphae , DOC-1-GFP localized to puncta that also oscillated at the hyphal tip during chemotropic interactions prior to hyphal fusion ( similar to MAK-2 and HAM-5; [45] ) , as well as to septa ( Fig 4B; S4 Movie ) . To determine whether DOC-1-GFP localization was dependent on MAK-2 , we expressed DOC-1-GFP in a strain deleted for mak-2 . The Δmak-2 ( doc-1-gfp ) strain showed a typical Δmak-2 phenotype , including lack of chemotropic interactions and cell fusion . Additionally , DOC-1-GFP showed cytoplasmic and vacuolar localization and never localized to puncta , although localization to septa was retained in this strain . Thus , DOC-1 is a component of the MAK-2 signaling complex and co-oscillates with this complex during chemotropic interactions . A doc-2-gfp allele regulated by the ccg-1 promoter localized to the plasma membrane and septa in mature colonies ( Fig 4C ) . When a fully functional gfp-doc-2 allele was placed under the regulation of the tef-1 promoter ( ccg-1 has low expression levels in germlings compared to tef-1; [47] ) , GFP-DOC-2 localized to puncta in germlings ( Fig 4D ) . However , oscillation of DOC-2 during chemotropic interactions in either germlings or in fusion hyphae was never observed . The oscillation of signaling components is necessary to maintain chemotropic interactions; inhibition of MAK-2 kinase activity obliterated oscillation and chemotropic interactions in communicating germlings [44] . The function of DOC-1/DOC-2 in communication group discrimination could be to prevent initiation of signaling , or to prevent reinforcement of signaling , which is hypothesized to be required for sustained chemotropic interactions [49 , 61] . To differentiate between these two hypotheses , we analyzed localization of MAK-2-GFP in Δdoc-1 Δdoc-2 germlings when they were in close proximity to CG2 germlings ( JW262 ) , where communication frequency is ~50% ( Fig 3A ) . Prior to chemotropic interactions , localization of MAK-2-GFP in Δdoc-1 Δdoc-2 germlings was observed at the tip when in close proximity to CG2 germlings ( Fig 6A ) . However , chemotropic interactions were only established if oscillation of MAK-2-GFP occurred ( Fig 6B; S5 Movie ) . If oscillation of MAK-2 was not maintained , chemotropic interactions between Δdoc-1 Δdoc-2 and CG2 germlings were abolished . These observations suggest that DOC-1/DOC-2 do not function at the recognition stage of germling interactions , but instead function to mediate communication group discrimination at the point where robust oscillation of the MAK-2 signaling complex to the tips of conidial anastomosis tubes is reinforced , and which is essential for further chemotropic interactions . The necessity of coordinated and out-of-phase oscillation of MAK-2 complex with SOFT for successful communication [44] suggested that the DOC proteins might influence the oscillation interval ( 8–10 min in CG1 strains ) and that differences in oscillation timing might determine communication group affiliation . To test this hypothesis , we compared the oscillation timing of SOFT-GFP during chemotropic interactions in germlings from a CG1 strain ( FGSC 2489 [so-gfp] ) [44] as compared to the CG5 strain ( Δdoc-1 Δdoc-2 [so-gfp] ) . We quantified fluorescence intensities at the tip of communicating germlings and measured the interval between two fluorescence maxima . However , no significant differences between the oscillation intervals of SOFT could be detected for CG1 versus CG5 germlings , suggesting that alteration of oscillation timing was not the basis of communication group phenotype ( S5 Fig ) . Our data indicate that doc-1 and doc-2 function as helping greenbeard genes , with multiple alleles mediating assortative kind recognition by changing chemotropic behavior by negatively regulating interactions during germling fusion . The finding of five communication groups mediated by five highly divergent haplotypes suggested a relatively ancient origin of the communication locus controlling germling fusion . To test this hypothesis , we first performed phylogenetic analyses of alleles at NCU07190 through NCU07193 in the 26 sequenced wild isolates , as well as alleles at these same loci from a population sample from the distantly related species Neurospora discreta . For NCU07190 and NCU07193 , allelic lines from within species were reciprocally monophyletic ( Fig 7A ) , as predicted by theory [62] , given the estimated divergence time between N . crassa and N . discreta ( 7–10 million years ago [63] ) and their effective population size ( circa 106 and 104 individuals , respectively [51 , 64] ) . However , for the three doc genes , no reciprocal monophyly was observed , and N . crassa alleles from the same CGH-associated clade were closer to N . discreta alleles than to N . crassa alleles from another clade , indicating that the age of allelic lines exceeds the age of speciation events—a phenomenon referred to as transspecies polymorphism ( Fig 7B ) . Inferred genealogical histories of doc genes were in fact concordant with differences in patterns of genomic arrangements among communication group haplotypes: alleles from CGH1 to CGH5 ( including CGH1A and CGH1B ) were in distinct clades for doc-1 , and similar topologies were also inferred at doc-2 ( although CGH5 strains lack doc-2 ) and doc-3 ( found only in CGH2 and CGH4 strains; Figs 2 and 7B ) . Transspecies polymorphism is a signature of long-term balancing selection [65] , and evidence for balancing selection was also provided by tests of the standard neutral model using Tajima’s D , which measures skewness of the allele frequency spectrum ( S3 Table ) . Tajima’s D values at NCU17048 , NCU07190 , NCU07193 , or NCU07194 did not deviate from neutral expectations ( Tajima’s D < 1; p > 0 . 1 ) , while values of Tajima’s D were high , positive , and significant for doc-1 , doc-2 , and doc-3 ( Tajima’s D > 2; p < 0 . 01 for doc-1 and doc-2; p < 0 . 05 for doc-3 ) . These data indicate that balancing selection is acting to maintain polymorphisms at doc-1 , doc-2 , and doc-3 , but its signature is not detectable on surrounding genes . The finding of long divergent haplotypes under balancing selection at the doc communication locus suggested that recombination rates might be reduced across the region , thereby preventing the migration of variants between allelic lines [65 , 66] . To test for recombination within the doc region between isolates of different CGHs , we analyzed concordance among genealogies of all genes within the region [67] . These analyses of genealogical concordance within the doc region revealed congruent branching of sequence groups from different CGHs for doc-1 , doc-2 , and doc-3 over the entire length of the genes , consistent with a lack of recombination between haplotypes from different communication groups ( S6 Fig ) . In contrast , an analysis of genealogical concordance within the doc region among haplotypes defining the same communication group was consistent with multiple recombination events ( S7 Fig ) , except for CGH1 isolates , in which recombination was not observed within the doc-1/2 region between the CGH1A and CGH1B isolates ( S6 and S7 Figs ) . These data suggest that the recombination rate between haplotypes from different communication groups was reduced , probably because of strong selection against recombinants .
Frequency-dependent effects , involving the expression of traits with differential effects on bearers and nonbearers , are common in microbes and can be interpreted as kind discrimination via greenbeard genes [3 , 9] . Previously , microbial kind discrimination has been described as a post-contact process; for example , cell adhesion proteins in D . discoideum [7 , 14] or in S . cerevisiae [15] . Here , we show that the filamentous fungus N . crassa uses kind discrimination that acts at a distance to differentiate communication groups in wild populations . We show that this kind discrimination system is controlled by the paralogous greenbeard genes doc-1 , doc-2 , and doc-3 , which together determine communication group affiliation . In genetically identical cells , chemotropic interactions are associated with the out-of-phase oscillation of MAK-2 and SOFT complexes [44–47] , which is postulated to allow genetically identical and developmentally equivalent cells to coordinate their behavior while avoiding self-stimulation [22 , 44 , 49 , 68]; DOC-1 oscillates with MAK-2 during chemotropic interactions . Thus , kind discrimination mediated by the DOC proteins adds another layer of complexity to germling communication , because cells must not only avoid self-stimulation but also stimulation by non-kind individuals . Although ligand ( s ) and receptor ( s ) must exist to account for chemotropic interactions between fungal germlings and hyphae , screens of the N . crassa near full genome deletion set have failed to identify genes encoding these components [50 , 61 , 68] . Our working model incorporates a communication ligand/receptor , which serves as a universal signal for chemotropic interactions in this species ( Fig 8 ) . The communication receptor is activated in the receiving cell upon interaction with the ligand . This signal is transmitted intracellularly to DOC-1/DOC-2 , which together function in quality control , an element commonly required for self-/non-self-discrimination [69] . Since Δdoc-1 Δdoc-2 germlings undergo self-communication and chemotropic interactions , DOC-1 and DOC-2 must function to repress MAK-2 oscillation reinforcement if non-kind germlings are in close proximity , rather than being required for the activation of signaling . Chemotropic interactions between two germlings is established if quality control allows the reinforcement of the MAK-2 and SOFT oscillation rhythm in communicating germlings , even if they are of different communication groups . If the signal does not pass the quality control , reinforcement of MAK-2 oscillation is suppressed and chemotropic growth fails to occur . The model for long-distance kind recognition in N . crassa is reminiscent of the “missing-self” theory for vertebrate natural killer cells and for non-self-recognition in the basal chordate , Botryllus schlosseri [70 , 71] . Instead of directly recognizing different non-self signals , anything that is not recognized as self by default is considered non-self . For natural killer cells and for the self-ligand fuhc with its effector system fester in B . schlosseri , a self-education process is predicted to occur that helps cells to adapt to the correct combination of cell surface receptors [72 , 73] . We predict that a similar process is mediated by DOC-1/DOC-2 to “educate” unknown recognition components involved in the reinforcement of MAK-2 signaling complex oscillation to the tips of conidial anastomosis tubes and fusion hyphae during chemotropic interactions . For example , membrane-bound protein DOC-2 may mediate kind signaling ( “self”-coding ) , perhaps via modification/interaction with the receptor or ligand or other components involved in recognition ( Fig 8 ) . In filamentous fungi , multiple loci confer self-/non-self-discriminations that act post-fusion and are typically among the most polymorphic loci in fungal genomes , but with a limited number of compatible allelic classes per locus [42 , 55 , 66 , 74–76] . Our studies revealed the existence of a single region of linked paralogous loci that confers at least five communication groups that function in self-/non-self-recognition during chemotropic interactions , prior to cell contact . CGH1 isolates additionally fell into two subgroups that showed divergent doc-1 and doc-2 alleles , which suggests that members of CGH1A and CGH1B groups may represent an additional sixth communication group . Although the presence of multiple long-diverged allelic lines is observed at fungal self-/non-self-discrimination loci , assortative kind recognition is not theoretically expected from kin selection theory . Indeed , if fusion between individuals is mutually beneficial ( and/or rejection costly ) , individuals carrying a common recognition allele will more readily fuse and , hence , have a higher fitness than individuals carrying less frequent alleles . Hence , as its frequency increases , the recognition should be turned into a “greenbeard gene” that recognizes copies of itself and is being recognized by copies of itself , and it should reach fixation through positive-frequency-dependent selection , thereby removing the variation necessary to allow discrimination in the first place [3 , 12 , 16 , 77] . However , the finding of long-diverged alleles and transspecies polymorphism consistent with long-term balancing selection at germling fusion loci ( this study ) and previously characterized self-/non-self-discrimination loci acting post-fusion [36–38] , suggests that additional extrinsic selective forces may promote the establishment and maintenance of assortative kind discrimination . For instance , self-/non-self-discrimination genes may directly experience balancing selection if kind recognition genes are maintained polymorphic by pathogen selection pressures causing rare allele advantage [18 , 39 , 78] . Bioinformatics and comparative genomics should help determine whether self-/non-self-discrimination genes such as doc genes have secondary functions that keep them variable . Although kind discrimination mediated during cell contact has been described in organisms ranging from bacteria to colonial ascidians [38 , 71 , 79 , 80] , the proteins and signals involved here are quite different . For invertebrates , it has been postulated that proteins controlling non-self-recognition are unique to each phylum [81] . It is possible that the core of non-self-recognition resides in intracellular conserved processes that integrate and respond to polymorphic external stimuli . We believe that kind discrimination mechanisms function in many filamentous fungi that are capable of undergoing cell/hyphal fusion . Hyphal avoidance has been described in a number of fungal species that are very distantly related to N . crassa [82 , 83] , making filamentous fungi excellent models for investigating kind recognition mechanisms . Our study provides the basis for research in self-/non-self-recognition that will be applicable to attraction , fusion , and kind discrimination in other eukaryotic species .
Standard protocols for N . crassa can be found on the Neurospora homepage at the Fungal Genetics Stock Center ( FGSC , http://www . fgsc . net/Neurospora/NeurosporaProtocolGuide . htm ) . Strains were grown on Vogel’s minimal medium ( VMM [84] , with supplements as required ) or on Westergaard’s synthetic cross medium for mating [85] . The wild N . crassa strains used in this study ( S1 Table ) were isolated from Louisiana , US and are available at the FGSC [52 , 86 , 87] . Manipulated strains are listed in S4 Table . FGSC 2489 served as parental strain for gene deletions and as a WT-control for all experiments , unless stated otherwise . Single deletion mutants are available at the FGSC [58 , 88] . The Δdoc-2 mutant deposited at the Neurospora knockout collection showed a flat phenotype , and its conidia had slow germination rates . In a back cross with FGSC 2489 , none of the phenotypes co-segregated with hygromycin resistance , indicating that it was due to a secondary mutation . To create the Δdoc-1 Δdoc-2 mutant , a deletion construct was created using the method of fusion PCR [89] . Briefly , ~1 kb of the 5′ regions of doc-1 and doc-2 was amplified by PCR from genomic DNA ( S5 Table ) , and the hygromycin cassette was amplified from the vector pCSN44 [58] . The three fragments were fused in a fusion PCR reaction to create the deletion construct , which was used to transform the Δmus-51 strain of N . crassa [58] . Hygromycin-resistant transformants were analyzed by PCR , and positive strains were backcrossed to FGSC 2489 . Histidine auxotrophic strains for complementation experiments were obtained by crossing the doc deletion strains with FGSC 6103 . The plasmid pMF272 ( AY598428 ) was modified to create gfp-fusions to doc-1 and doc-2 , which were targeted to the his-3 locus [90] . A 300 bp fragment of the 3′ region of ccg-1 was cloned 3′ of gfp open reading frame as a termination signal using the EcoRI restriction site . Plasmid derivatives with the tef-1 promoter or native promoters were obtained by swapping out the ccg-1 promoter using the restriction enzymes NotI and XbaI . For CG switch experiments , doc-1 and doc-2 , including their native promoter and terminator sequences , were amplified from genomic DNA of the isolate P4471 ( S5 Table ) . Using the Gibson assembly , both fragments were cloned into the EcoRI/NotI digested vector pMF272 [91] . All constructs were transformed into FGSC 6103 with selection for His+ prototrophy and then crossed into the doc double or single deletion mutants . Each strain was grown on VMM in slant tubes for 4–6 d or until significant conidiation occurred . Conidia were prepared by filtering 600 μl of conidial suspension through cheesecloth . An aliquot of 180 μl of a conidial suspension from one strain was mixed with 20 μl of FM4-64 solution ( 16 μM ) , incubated for 15 min , and subsequently washed with 1 ml of ddH2O . The conidial titer was adjusted to 3×107 conidia/ml . An aliquot of 45 μl of conidial suspension from both strains was mixed , and 80 μl of this final mixture were spread on VMM agar plates ( 60 x 15 mm ) . Plates were incubated for 4 . 5 h at 25°C or 3 . 5 h at 30°C . Agar squares of 1 cm2 were excised and observed with a Zeiss Axioskop 2 equipped with a Q Imaging Retiga-2000R camera ( Surrey ) using a 40x/1 . 30 Plan-Neofluar oil immersion objective and the iVision Mac4 . 5 software . Different strains were either discriminated by GFP or FM4-64 fluorescence , or FM4-64 fluorescence versus no fluorescence , if two wild isolates were analyzed . Communication frequencies were determined for at least 15 fields , depicting a total of at least 100 interactions with three biological replicates . Conidia of strains bearing different auxotrophic markers ( his-3 , ad-3B , or pyr-4; S4 Table ) were harvested as described above . The conidial titer of one strain was adjusted to 3 x 106 conidia/ml , and the conidial titer of the forced communication partner ( bearing a different auxotrophic marker ) was adjusted to 3 x 105 conidia/ml . A 150 μl spore suspension of both strains was mixed and spread on modified VMM agar that promotes colonial growth ( FGSC , http://www . fgsc . net/Neurospora/NeurosporaProtocolGuide . htm ) . Due to the complementing auxotrophic markers , only heterokaryotic , prototrophic fusion products were able to grow on VMM . Plates were incubated at 30°C for 4 d , when cell-forming units/plate were documented . Cellular localization studies were performed with a Leica SD6000 microscope with a 100×1 . 4 NA oil-immersion objective equipped with a Yokogawa CSU-X1 spinning disk head , a 488 nm laser for GFP fluorescence , and a 563 nm laser for mCherry fluorescence controlled by the Metamorph software ( Molecular Devices , Sunnyvale , CA ) . Conidia from strains expressing fluorophore-tagged proteins were prepared for microscopy as described above . For time-lapse studies , images were taken at 30 s intervals . The software ImageJ ( http://imagej . nih . gov/ij/ ) was used for image processing . For co-localization studies , heterokaryons were created by inoculating the center of a plate with a mixture of conidia of a strain expressing DOC-1-GFP and a strain expressing MAK-2-mCherry , or a strain expressing DOC-2-GFP and a strain expressing MAK-2-mCherry , respectively ( S4 Table ) . Conidia bearing both GFP and mCherry fluorescent proteins were prepared and imaged as explained above . For DNA isolation , strains were grown on VMM agar plates covered with a disk of sterile cellophane at 30°C for 24 h . DNA was purified using the DNeasy Blood & Tissue kit ( Qiagen Inc . ) . Equal amounts of DNA from 46 segregants ( 66 ng/segregant ) were combined and used for library preparations using the TruSeq DNA LT Kit ( Illumina ) . All paired end libraries were sequenced on a HiSeq2000 sequencing platform using standard Illumina operating procedures ( Vincent J . Coates Genomics Sequencing Laboratory , Berkeley ) to a read length of 100 nucleotides and a minimum mean depth of genome coverage of 71 for the sequenced libraries after filtering for low-quality reads , using the DepthofCoverage program from GATKv2 . 3–9 [92] . Low-quality reads were removed from the sequencing data using the Fastx toolkit ( http://hannonlab . cshl . edu/fastx_toolkit/index . html ) . The filtered paired ends were regrouped by a custom perl script and mapped to the N . crassa genome FGSC 2489 v12 with the short read aligner Bowtie2 . 00 [93] . Read groups were added to sorted BAM files with Picard-tools v1 . 85 ( http:broadinstitute . github . io/picard/ ) and SNP analysis performed with The Genome Analysis Tool Kit v2 . 3–9 after indel realignment with the RealignerTargetCreator and IndelRealigner programs from the GATK [92] . SNPs were confirmed by viewing the mapped polymorphisms on the Integrative Genomics Viewer v2 . 3 [94] . The mapped reads for the two parental strains ( FGSC 2489 and JW258 ) plus the mapped reads for the 46 pooled segregants ( FGSC 2489 communicators , CG1 or JW258 communicators , CG2 ) are available at the Sequence Read Archive ( SRA ) ( http://www . ncbi . nlm . nih . gov/sra ) ( SRA311058 ) . The doc-1 and doc-2 sequences of N . crassa and N . discreta wild isolates were obtained by a BLAST search [95] using NCU07191 and NCU07192 from FGSC 2489 as a query against de novo sequence assemblies from 26 wild isolates [55] . For DNA sequence comparisons , the pairwise sequence alignment tool EMBOSS Needle from EMBL-EBI was used [96] . Codon alignments were carried out using Macse [97] and visualized and processed using JalView ( http://www . jalview . org/ ) . Modified multiple alignments were trimmed using Trimal [98] . Phylogenetic trees were inferred from trimmed alignments using the default pipeline from Phylogeny . fr ( Muscle , Gblocks , Phyml [100 bootstraps] ) [99] and visualized using FigTree1 . 4 ( http://tree . bio . ed . ac . uk/software/figtree/ ) . To obtain DNA divergence statistics , the trimmed codon alignments of doc-1 , doc-2 , and doc-3 sequences were sorted based on CGH groups . DnaSP5 was used to compute polymorphism and divergence and to test the standard neutral model using Tajima’s D [100] . Partitioned alignments for each locus were created using RAxML [101] . To detect recombination within CGHs in the region around doc-1/2 , the program Rdp [67] was used , applying the “all methods mode” with default setting . Sequences of about 30 kbp surrounding doc-1/2 were extracted from de novo genome sequence assemblies . Alignments were made using the program Mafft for each CGH group [102]; gaps were trimmed using Trimal [98] .
|
Microorganisms undergo social activities that benefit the species , but for social microbes , the ability to discriminate between genetically similar and genetically dissimilar individuals is instrumental in preventing cheaters from taking advantage of altruistic behavior . While kin recognition is important in animals , microbes often use kind recognition , in which cells are genetically related only at certain loci—so-called “greenbeard” genes . Genomic and genetic analyses of a wild population of the filamentous fungus Neurospora crassa showed that greenbeard genes mediate long-distance kind discrimination that regulates communication and chemotropic interactions of cells prior to somatic cell fusion; N . crassa cells actively search for fusion partners with similar greenbeard genes . Kind discrimination was regulated by a set of highly divergent paralogous genes ( doc-1 , doc-2 , and doc-3 ) that were necessary and sufficient to confer communication identity . Alleles that confer the interaction phenotype at the doc loci have been maintained through multiple speciation events , suggesting that selection is acting to maintain different communication groups in fungi .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"Methods"
] |
[
"sequencing",
"techniques",
"cell",
"physiology",
"fungal",
"genetics",
"fungi",
"model",
"organisms",
"molecular",
"biology",
"techniques",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"mycology",
"sequence",
"alignment",
"cell",
"communication",
"molecular",
"biology",
"genetic",
"loci",
"genetic",
"oscillators",
"fungal",
"genomics",
"neurospora",
"crassa",
"cell",
"biology",
"genetics",
"biology",
"and",
"life",
"sciences",
"yeast",
"and",
"fungal",
"models",
"genomics",
"neurospora",
"organisms",
"cell",
"fusion"
] |
2016
|
Characterization of Greenbeard Genes Involved in Long-Distance Kind Discrimination in a Microbial Eukaryote
|
Cell size is thought to play an important role in choosing between temporal and spatial sensing in chemotaxis . Large cells are thought to use spatial sensing due to large chemical difference at its ends whereas small cells are incapable of spatial sensing due to rapid homogenization of proteins within the cell . However , small cells have been found to polarize and large cells like sperm cells undergo temporal sensing . Thus , it remains an open question what exactly governs spatial versus temporal sensing . Here , we identify the factors that determines sensing choices through mathematical modeling of chemotactic circuits . Comprehensive computational search of three-node signaling circuits has identified the negative integral feedback ( NFB ) and incoherent feedforward ( IFF ) circuits as capable of adaptation , an important property for chemotaxis . Cells are modeled as one-dimensional circular system consisting of diffusible activator , inactivator and output proteins , traveling across a chemical gradient . From our simulations , we find that sensing outcomes are similar for NFB or IFF circuits . Rather than cell size , the relevant parameters are the 1 ) ratio of cell speed to the product of cell diameter and rate of signaling , 2 ) diffusivity of the output protein and 3 ) ratio of the diffusivities of the activator to inactivator protein . Spatial sensing is favored when all three parameters are low . This corresponds to a cell moving slower than the time it takes for signaling to propagate across the cell diameter , has an output protein that is polarizable and has a local-excitation global-inhibition system to amplify the chemical gradient . Temporal sensing is favored otherwise . We also find that temporal sensing is more robust to noise . By performing extensive literature search , we find that our prediction agrees with observation in a wide range of species and cell types ranging from E . coli to human Fibroblast cells and propose that our result is universally applicable .
Chemotaxis is the process whereby cells move towards a region of higher chemical stimulus concentration . Cellular movements towards the favorable direction enables , for example , prokaryotic unicellular organisms such as Escherichia coli ( E . coli ) to move towards food and eukaryotic cells such as neutrophils and macrophages to move towards the site of infection to phagocytize external parasites . Information about the external chemical gradient is transduced into the cell by binding of chemoattractant and chemorepellant molecules to specific receptors at the cell surface . These binding events then trigger downstream intracellular signaling to modulate the cell’s motility . To move up or down the gradient , cells can adopt two distinct strategies: temporal sensing or spatial sensing ( Fig 1 ) . In temporal or sequential sensing , cells compare the intensity of receptor stimulation at different times ( Fig 1 , left ) and modulate their probability of moving in the same direction or switching directions . In E . coli , an organism exhibiting temporal sensing , rotation of its flagella in the counter-clockwise direction results in directed motion whereas rotation in the clockwise direction results in tumbling and a random change in direction [1 , 2] . Binding of chemoattractant decreases the switching probability from counter-clockwise to clockwise rotation , thus reducing tumbling and increasing the run length when the cell is moving in the favorable direction . In spatial sensing , cells simultaneously measure the intensity of receptor stimulation at its two ends ( Fig 1 , right ) . The different receptor stimulation leads to cell polarization and motility in the preferred direction . In neutrophils , G protein-coupled receptors ( GPCRs ) are originally evenly distributed along the plasma membrane . Binding of chemoattractant results in activation of signaling pathways involving small Rho guanosine triphosphatases ( Rho GTPases ) and phosphoinositide 3 kinases ( PI3Ks ) and asymmetric polymerization of actin at the up-gradient edge of the cell , facilitating motion up the gradient [3] . The decision whether to employ temporal or spatial sensing has largely been attributed to cell size . It is thought that large cells have an advantage for spatial sensing as the intensities of receptor stimulation are expected to be very different at its two ends . In contrast , small cells of around or less than a micron in diameter are unable to exhibit spatial sensing as chemical gradients are rapidly homogenized by fast diffusion . For example , in an E . coli of 2um , the cytoplasmic CheY chemotaxis signal transduction protein with a diffusion constant of 4 . 6±0 . 8um2s−1[4] will take only 0 . 9s to transerve the cell . However , spatial localization of MinC , MinD and MinE proteins to bring about proper cell division [5] and polar localization of the chemoreceptor complex of cytoplasmic CheA and CheW proteins [6] suggest that spatial segregration of proteins can be established at the micron scale in small cells . Berg and Purcell also showed theoretically that , in principle , an immobile E . coli cell is able to perform spatial sensing [7] . Dusenbery , based on arguments of signal-to-noise ratio , also found that the cell size limit for spatial sensing ( < 1um ) is close to that for temporal sensing and is actually smaller than the size of many prokaryotes [8] . These works cast doubts on previous arguments for the inability of small cells of around a micron in diameter to perform spatial sensing and suggests that most cells , whether big and small , are able to perform both spatial and temporal sensing . Here , we use computational model to show that the decision to perform either temporal or spatial sensing is instead determined by the performance of each type of sensing . To determine the performance of temporal and spatial sensing , we need to integrate the sensing mechanism with the network circuits use for chemotaxis . A key goal in systems biology is to identify network motifs capable of achieving certain biological function . For chemotaxis to be effective , cells need to exhibit adaptation . Adaptation refers to a cell’s ability to respond to a change in the input stimulus and then return to its original level , even when the input stimulus remains high . This property allows cells to respond to a high range of chemoattractant concentration . Extensive efforts to understand the ability of E . coli to remain sensitive to a wide range of chemoattractant has led to the identification of the negative integral feedback ( NFB ) circuit for chemotaxis [9 , 10] ( Fig 2 , step 1 , left ) . In NFB , following stimulation of the output protein ( protein C ) by the activator ( protein A ) , a buffering component/inactivator ( protein B ) integrates the difference between the response and the baseline level and feeds this difference back into the response , enabling the output protein to return to the basal level after each pulse of chemoattractant . On the other hand , modeling efforts in eukaryotic gradient sensing have identified the incoherent feedforward ( IFF ) circuit ( Fig 2 , step 1 , right ) for amplification of the signaling response to shallow gradients [11–13] . In IFF , two nodes , an activator ( protein A ) and a repressor ( protein B ) , are activated proportionally to the stimulus but act with opposite effects on the output protein ( protein C ) . Like the NFB , the IFF circuit also has the adaptive property needed for sensing a wide range of chemoattractant . A comprehensive survey of all possible three-node network topologies had been carried out to search for networks that yield biochemical adaptation response [14] . They found that minimal circuits containing NFB and IFF motifs yield adaptation and that more complicated circuits that yield adaptation contain at least one of these two motifs . Hence we will use the NFB and IFF circuits to study cells’ chemotaxic response as they are the basic building blocks for three-node circuits that can yield adaptative property , an essential property for chemotaxis . We compare the performance of temporal and spatial sensing when a cell uses the NFB and IFF circuits by determining the conditions that favor one mode of sensing over the other . In temporal sensing , the cell compares the level of C with the steady state level of C ( area highlighted in green ) ( Fig 2 , step 4 , temporal ) whereas in spatial sensing , the cell compares the level of C at the front half and back half of the cell ( area highlighted in red ) ( Fig 2 , step 4 , spatial ) . We identify five dimensionless terms , namely the diffusivities of the activator ( protein A ) , repressor ( protein B ) and output ( protein C ) proteins , all normalised to the deactivation rate of the output protein , the effective chemoattractant gradient experienced by the moving cell , and the ratio of cell speed to the product of diameter and signaling rate that characterize the response of the negative integral feedback and incoherent feedforward circuits . By varying these five terms and comparing the performance of temporal and spatial sensing on the negative integral feedback and incoherent feedforward circuits , we find that spatial sensing performs better than temporal sensing in the regime where the cell velocity is small relative to the product of cell diameter and the circuit reaction rate , and when the repressor protein ( protein B ) diffuses faster than protein A ) and the diffusibility of the output protein ( protein C ) is low . In all other cases , temporal sensing performs better . By incorporating noise into our analysis , we also found that temporal sensing is more robust to noise than spatial sensing .
Here , we want to determine whether cell size is the determining factor or there are other factors contributing to the choice between temporal and spatial sensing . We assume that the mode of sensing that yields higher signaling output will be adopted by cells . In general , the signaling output will depend on both the signaling ( i . e . , molecular ) and physical properties of the chemotactic cell as well as the properties of the chemoattractant . Hence , we need to identify these important variables and determine how they affect the signaling outputs . However , one obstacle is that , very often , the values of these variables have not been measured experimentally . Thus , we will adopt a network motifs approach where the exact parameter values are not so critical as long as the parameter values lie within certain regimes , since the same behavior is typically observed over a range of parameter vaues . Hence , we will identify all possible behaviors of the networks by sweeping through parameter space . This approach has been widely adopted in modeling papers ( e . g . , Ma et al . , 2014 ) . In our analysis , we have swept through 4 to 5 orders of magnitude of parameter values and obtained the perfect adaptive behavior expected for the network motifs . Our analysis consists of four steps ( Fig 2 ) . We first set up the equations for the negative integral ( NFB ) and incoherent feedforward ( IFF ) circuits ( Fig 2 , step 1 ) and identify 100 sets of parameters that lead to high sensitivity and adaptation precision for these circuits ( Fig 2 , step 2 ) . High sensitivity is responsible for signal amplication in shallow gradients whereas high adaptation precision is required for signal adaptation . These are properties that enables a cell to perform chemotaxis effectively . Next , we determine the protein dynamics as the cell moves through a linear gradient for the sets of parameters that we have identified in step 2 for the NFB and IFF circuits ( Fig 2 , step 3 ) . The cell is modeled as a one dimensional ring , of diameter d , with an activator protein ( A ) , inactivator protein ( B ) and output protein ( C ) that can diffuse freely on the cell membrane . At time τ = 0 , the cell moves with velocity , v , into a linear chemoattractant gradient with slope , k . The cell experiences the gradient for a fixed time , Ts , before moving into a region with constant I = IH . The cell uses both the NFB and IFF circuits to process the chemoattractant input and interprets the results using temporal or spatial sensing ( Fig 2 , step 4 ) . More details can be found in S1 Text . Extending the equations for incoherent feedforward and negative integral feedback circuits to account for spatial differences of protein levels on the cell’s membrane and a changing external chemoattractant gradient ( see S1 Text ) , we find that the equations are fully described by the following variables: cell diameter d , cell velocity v , chemoattractant gradient k ( which has the unit of inverse length ) , signaling rates , of which we choose lBC , the deactivation rate of C , to be representative ( i . e . , other signaling rate can be expressed as ratios of it ) , and the activator ( A ) , inactivator ( B ) , and output protein ( C ) diffusivities , DA , DB , and DC , respectively . These variables can be grouped into the following five dimensionless variables below . The five dimensionless variables are as follows: First , we consider the effect of D C ′ on the choice between temporal versus temporal sensing . For spatial sensing , the cell compares the level of protein C at different parts of the cell . Therefore , C has to diffuse slowly to allow for spatial sensing . When D C ′ is big , any spatial information will be rapidly homogenized and temporal sensing will be favored . We use D C ′ = 0 for our analysis to study the effects of other parameters on the sensing choice . We hypothesize that α would not affect signaling outcome as a steeper or more gentle external gradient would affect the output from both sensing choice equally . To test this hypothesis , we vary α = 0 . 00001 , 0 . 0001 , 0 . 001 , 0 . 01 for D A ′ = 1 . 0 , D B ′ = 100 . 0 , D C ′ = 0 and β = 0 . 125 , 0 . 5 , 2 . 0 , 8 . 0 . For each set of parameters , we systematically simulate the dynamics for the selected sets of parameters and determine the output for spatial and temporal sensing . The strategy yielding the higher output will be selected . As shown in S1 ( a ) and S1 ( b ) Fig , α does not affect the choice of temporal and spatial sensing . We also plot the output using temporal sensing ( green ) versus spatial sensing ( red ) at β = 0 . 125 ( S1c Fig ) and β = 8 . 0 ( S1d Fig ) for different values of α . We observe that both the outputs scale linearly with the increase in α . Since α affects both outputs equally , it does not affect the sensing choice . Since α does not affect the sensing choice , we have fixed α = 0 . 001 and focus on the effects of D A ′ , D B ′ and β . We simulate the protein dynamics for β = 0 . 125 , 0 . 25 , 0 . 5 , 1 . 0 , 2 . 0 , 4 . 0 , 8 . 0 , D A ′ = 0 . 1 , 1 . 0 , 100 , 1000 and D B ′ = 0 . 1 , 1 . 0 , 100 , 1000 . We plot the percentage of runs that yield higher signaling output adopting the temporal ( green ) and spatial ( red ) strategy for different values of β in Fig 3 ( a ) and 3 ( b ) . Although the negative integral feedback ( NFB ) and incoherent feedforward circuits ( IFF ) have been associated with temporal [9 , 10] and spatial sensing [11 , 12] respectively , we find that the two circuits yield similar results . This shows that NFB can be used for spatial sensing and that the IFF can be used for temporal sensing . We find that when β is high ( cell velocity is high or cell diameter is small ) , temporal sensing yields higher output than spatial sensing independent of the value of D A ′ and D B ′ . To examine the effect of β , we plot the protein dynamics of the output protein for one set of parameter for the incoherent feedforward circuit at various values of beta for D A ′ = 1 . 0 and D B ′ = 1 . 0 . When β is small ( cell velocity is small or cell diameter is large ) , the front and back halves of the cell experience a big delay in the time that they observe the chemoattractant and the levels of the output protein , C , at the rear end ( blue curve ) of the cell only increase after the level of C at the front end ( green curve ) starts to decrease ( Fig 3c , left ) . The average level of C ( red curve ) , which sums over the two halves , shows a net increase at all times when the cell is moving through the gradient . As β increases , the time difference in which the front and back halves of the cell experiences the chemoattractant decreases and their dynamics began to converge ( Fig 3c , right ) . The signaling output for temporal sensing ( green ) and spatial sensing ( red ) were plotted in Fig 3d . When the cell uses the temporal sensing mechanism by comparing the average output C with the baseline level , it observes a net increase in output ( area shaded in green ) as the cell moves through the gradient . When spatial sensing is used to comparing the ratio of output at the front and back of the cell , the cell observes an increase in output ( area shaded in red above the x-axis ) as the cell entered the gradient ( entering phase ) followed by a decrease in output ( area shaded in red below the x-axis ) as the cell exits the gradient ( exit phase ) ( Fig 3d ) . However the area above the x-axis is always larger than the area below the x-axis indicating an overall positive response . As the difference between the level of the output protein at the front and back of the cell decreases with increasing β , so is the signal obtained from spatial sensing . This explains why at high β , temporal sensing is favored . Rather than size cell , we show that the relevant parameter for sensing is the ratio of cell velocity to the product of signaling rate and cell diameter . This suggests that cells moving faster than its cell diameter in the time it takes for signaling to propagate across the cell diameter should adopt temporal sensing , whereas cells moving slower than its cell diameter in that time should adopt spatial sensing . This can be reasoned as follows: a fast-moving , small cell performs better comparing the chemoattractant at different times in its trajectory; whereas , a slow-moving , big cell that is not travelling much performs better by comparing the chemoattractant concentration at its two ends . As shown in Fig 3a and 3b , both temporal and spatial sensing can occur when β is small and we will next focus on the effects of D A ′ and D B ′ on this sensing choice . As shown in Fig 4 ( a ) and 4 ( c ) , temporal sensing is favored when D A ′ > D B ′ and spatial sensing is favored when D A ′ < D B ′ . The dynamics of protein C are plotted for different values of D A ′ and D B ′ ( Fig 4b and 4d ) . At low diffusion ( D A ′ = 1 and D B ′ = 1 ) , the front and back halves behave like separate uncommunicating entities as discussed before and temporal signaling yields slightly higher output than spatial sensing ( Fig 4b and 4d , bottom row , left ) . When diffusion of the activator is slow and diffusion of the inactivator is fast ( D A ′ = 1 and D B ′ = 100 ) , coupling between the front and back of the cell occurred . Once the cell entered the gradient , inactivator B is produced and diffuses to the back of the cell to suppress the output level of protein C , amplifying the difference in levels of protein C between the front and the back . This amplification led to a reduction of C from its basal level at the back of the cell ( Fig 4b and 4d , bottom row , right ) . Hence spatial sensing yields much higher output signal than temporal sensing . Furthermore , this coupling ensured that the levels of protein C at the back of the cell is lower than that at the front even during the exit phase . This is consistent with previous models adopting a local acting activator and a globally acting inactivator for spatial sensing [11] . On the other hand when diffusion of the activator is fast and diffusion of the inactivator is slow ( D A ′ = 100 and D B ′ = 1 ) , the global activation and local inhibition happens with activator diffusing to the back of the cell . In IFF , this leads to higher level of protein C at the back than the front during the entering phase ( Fig 4b , top row , left ) . This occurs as protein A produces at the front end of the cell rapidly diffused to the back , homogenizing level of protein A throughout the cell . The higher level of inactivator , protein B , leads to greater repression and lower level of protein C at the front . In this case , the level of protein C becomes higher at the back and the signaling output from spatial sensing becomes negative , making spatial sensing an inviable option . This effect is not observed in NFB circuits as , protein B was activated by protein A rather than the external chemoattractant ( Fig 4d , top row , left ) . Hence level of protein B is always be proportional to that of protein A . However in this case , spatial sensing is also not favored as the rapid diffusion of protein A led to loss of information about the external chemoattractant gradient . Finally when both activator and inactivator diffuse fast ( D A ′ = 100 and D B ′ = 100 ) , the amplification effect observed for local excitation and global inhibition is still observed , albeit at a lower value ( Fig 4b and 4d , top row , right ) . In summary , we find that spatial sensing is favored when the repressor diffuses faster than the activator . This is because repressor produces at the front end is able to diffuse to the back to lower the signaling level of the output protein . This magnifies the difference between the signal output at the two ends , leading to higher signaling output for spatial sensing . When repressor diffuses slower than the activator , this amplification does not occur and temporal sensing is favored . To check that our findings are independent of the exact gradient profile , we repeat our analysis for an exponential gradient ( S2 Fig ) . We find that similar to results of the linear gradient , high ratio of cell speed to cell diameter favors temporal sensing and diffusivity of activator has to be smaller than diffusivity of repressor for spatial sensing to be preferred at low values of β . In our simulations , the cell is moving from a region of constant chemoattractant , into a region with a linear increase in chemoattractant and finally into another region of a higher constant chemoattractant level . In general , cells may be moving inside a steady state gradient . To show that the motion from a region of constant chemoattractant into a gradient does not affect the findings , we simulate the response of cells into a step change in chemoattractant ( S3 Fig ) . This will simulate the case where a cell suddenly encounters a gradient and moves into it , as opposed to moving inside a steady-state gradient . We find that the main findings are consistent with those for a linear gradient . We also repeat the simulations using a longer Ts = 20 and obtain similar findings ( S4 Fig ) . In Fig 5a , we summarize our findings . We find that sensing outcomes are determined by three dimensionless parameters: 1 ) the ratio of cell speed to the product of cell diameter and rate of signaling , 2 ) the diffusivities of the output protein of the two circuits and 3 ) the ratio of the diffusivities of the activator to inactivator protein . Temporal sensing is usually preferred whereas spatial sensing is preferred when all three parameters are low . To compare our theoretical results with experimental observations , we need to determine the diffusion rates , cell sizes and speeds of a wide range of chemotactic cells and organisms . While cell sizes and speeds are readily available , values of diffusion rates are much harder to find . Hence , we first compare our findings based on the ratio of cell speed to cell diameter with that of the sensing decisions of chemotactic cells and organisms . The most well-studied chemotactic organism is E . coli . E . coli is 2 μm in length [15] and swims at about 20 μm/s . The dephosphorylation rate of Che-Y has been found to be 2 . 2s−1[16 , 17] . This yields β = 4 . 5 , agreeing with our analysis that E . coli will adopt temporal sensing . Since reactions rates are difficult to characterize and the circuitry controlling chemotaxis is usually much more complicated than our canonical NFB and IFF circuits , we are unable to obtain lBC for many chemotactic cells . Nonetheless , we estimate reaction rates to be of the order of seconds based on the dephosphorylation rate of Che-Y [16 , 17] and the fast response time observed in chemotactic cells . Micropipette stimulation experiments showed that neutrophils took between 5–30s to extend their surface towards the chemotactic pipette [18] . We conduct an extensive literature search to obtain the diameters and velocities of many chemotactic cells and unicellular organisms such as bacteria [19–22] , Paramecium caudatum ( P . caudatum ) [23] , Tetrahymena thermophila[24] , alga [25]; sperm cells [26 , 27]; mammalian cells [28–33]; insect cells [34 , 35]; and amoeba [36–38] . We classified these chemotactic cells based on their mechanisms of motion , namely lamellipodia/filopodia , flagellar , pseudpodia and cilia . In general , the eukaryotic and insect cells are in the lamellipodia/filopodia group; bacteria and sperm cells are in the flagellar group; amoeba are in the pseudpodia group; and Tetrahymena thermophila and alga are in the cilia group . We find that cells using flagellar and cilia to move have higher ratio of velocity over cell diameter than cells using lamellipodia/filopodia and pseudpodia ( Fig 5b ) . In our simulations , we find that cells and organisms with high ratio of cell speed to cell diameter adopt temporal sensing . Assuming lBC = 1s−1 , cells and organisms above the the black horizontal line in Fig 5b will adopt temporal sensing . In general , lBC may be different in each cell , if lBC lies between 0 . 2s−1 − 5s−1 then the yellow region will be the separating boundary between cells with high and low values of β . Cells with high β values includes cells in the flagellar ( green ) group and agrees with the broad categorization that these cells adopt temporal sensing . Sperm cells have been shown to utilize temporal sensing despite being relatively big [26] . Our results suggests that temporal sensing is utilized as its high cell velocity makes temporal sensing more advantageous . One exception to the classification is the bipolar flagellated vibrioid bacteria that has been suggested to adopt spatial sensing [39] . This bacteria has a very fast response time as it was able to correct deviations from its swimming direction within a second . Further work elucidating the chemotaxis circuitry and reaction rates in this organism is necessary to determine the value of β . It is currently unclear whether P . caudatum adopts spatial or temporal sensing . The other ciliated organism ( blue ) , Tetrahymena thermophila , has been proposed to utilize temporal sensing [40] , agreeing with our prediction . Cells and organisms below the yellow region in Fig 5b have low values of β . We find that these cells would adopt spatial sensing if the activator diffuses slowly whereas the inactivator diffuses fast . Unfortunately it is difficult to obtain these diffusion rates as many of the activator and inactivator proteins involved are unknown . For example , in Dictyostelium discoideum , some literature suggests that the locally acting activator ( Protein A ) , PI3-kinase , and globally acting inactivator ( Protein B ) , PTEN , work together to control G-protein ( Protein C ) activation during chemotaxis [13] whereas other literature suggests that RasGEF and a RasGAP are the activator and inactivator proteins instead [12] . As diffusion rate is inversely proportional to the square root of the molecular weight , one could estimate the ratio of PTEN to PI3K diffusion rate and the ratio of RasGEF to RasGAP diffusion rate to be 83 , 598 47 , 166 2 = 1 . 33 and 57 , 010 54 , 556 2 = 1 . 02 respectively . The slight differences in these estimated diffusion rates are clearly inconsistent with the local and global activation roles suggested . This shows that even when there are candidate proteins for the activator and inactivator proteins , molecular weight is not a good approach for estimating diffusion rates in cells and suggests the presence of other active biological processes in controlling the movements of these proteins . From Fig 4 ( b ) and 4 ( d ) , we observe that the signaling output , OS , is highest at low activator diffusion rate and high inactivator diffusion rate ( D A ′ = 1 . 0 and D B ′ = 100 ) for low value of β . From an evolutionary point of view , this suggests that organisms would evolve towards having high D B ′ and low D A ′ to achieve better chemotactic response . Indeed , it has been shown experimentally that lamellipodia/filopodia ( black ) and pseudpodia cells ( red ) utilize spatial sensing . Hence we find that β is the most important determinant in the choice between spatial and temporal sensing . Next , we consider the effect of noise on the decision choice . Noise can exist in both the external chemoattractant and the internal signaling pathway and affects chemotaxis [41] . We focus our analysis on the regime where D A ′ is low and D B ′ is high as this was the region of parameter space that yields most interesting behavior in the deterministic analysis . We examine the decision choice for the following cases: ( 1 ) β = 0 . 25 , ( 2 ) β = 1 . 0 and ( 3 ) β = 4 . 0 at D A ′ = 1 and D B ′ = 100 as the amount of external or internal noise increases . Since each run is stochastic , ten runs are performed on each set of parameters and noise level to determine the average performance from spatial and temporal sensing . First , we focus on the presence of external noise in the chemoattractant gradient . The dynamics of protein C is plotted at different noise levels for β = 0 . 25 and β = 4 . 0 ( S5 Fig ) . η quantifies the amount of fractional noise . At low level of noise , η = 0 . 0625 , the dynamics of protein C is well behaved with the levels of protein C at the front always higher than that at the back ( S5a and S5d Fig ) . As η increases , the dynamics becomes noisier with levels of protein C showing more fluctuations ( S5b and S5e Fig ) . Furthermore , the level of protein C at the front of the cell is sometimes lower than that at the back . However , the average levels of protein C is still rather well-behaved , rising as the cell enters the chemoattractant gradient and adapting back to basal level as the cell exits the gradient . At high level of noise , η = 1 . 0 , the noise level dominates over the signal and the levels of protein C fluctuated randomly ( S5c and S5f Fig ) . Next , we want to determine which sensing strategy is more susceptible to noise . For each set of parameters , ten stochastic runs are performed . If all the runs yield positive signaling output for a particular sensing strategy that strategy is considered to be viable for that set of parameters . We plot the fraction of parameter set that fulfil the above criteria for spatial and temporal sensing ( Fig 6a and 6b ( red ) . We find that temporal sensing ( green ) was less susceptible to noise than spatial sensing ( red ) . Intuitively , this can be understand as taking average in temporal sensing is more robust than taking difference in spatial sensing . We also find that the fraction of parameters fulfilling the criteria increased as β decreased ( Fig 6a , ( red ) ) . This showed that spatial sensing is less susceptible to noise when the cell diameter is larger than cell velocity . Lastly , we determine the fraction of parameters that chooses temporal or spatial sensing . When noise level becomes too high , both sensing mechanisms fail as the signal had been completely dominated by noise . As shown in ( Fig 6c and 6d ) , spatial sensing performs better than temporal sensing for low values of β and low values of noise . As noise level increases , temporal sensing yields better results . Finally at very high noise levels , sensing using both strategies are infeasible . To introduce noise into the internal signaing pathway , we allow all the kinetic parameters ( kIA , kIB , lFA , lFB , kAC , lBC , kCB ) to be random variable with mean equal to their values in the noiseless case and variance , ν . We find that the sensing decision is independent of the amount of noise , ν ( Fig 7 ) . We examine the dynamics of protein C when subjected to external chemoattractant noise and internal signaling noise ( Fig 8 ) . We find that in the presence of internal noise , protein C fluctuates at high frequency about the expected value of C for the noiseless case . Integrating over time , the noise would cancel out , leading to an average performance similar to that of the noiseless case . On the other hand , protein C fluctuates at low frequency in the presence of external noise and its mean averaged over time can be quite different from the expected value of C for the noiseless case . Hence in this case , temporal sensing performs better .
Here , we determine the conditions favoring temporal and spatial sensing . We find that the behavior of the negative integral feedback and incoherent feedforward circuits were determined by five dimensionless constants , namely the three diffusion rates , D A ′ = D A d 2 l B C , D B ′ = D B d 2 l B C and D C ′ = D C d 2 l B C , the ratio of cell speed to the product of cell diameter and signaling rate , β = v d l B C , and the effective chemoattractant gradient , α = v k l B C . Both the negative integral feedback and incoherent feedforward circuits yielded similar behaviors when we varied the dimensionless constants . We summarize our findings in Fig 5a . In brief , temporal sensing is favored in most situations whereas spatial sensing is only favored when values for D C ′ and β are small and D A ′ , diffusion rate of the activator , is slower than that of D B ′ , diffusion rate of the inactivator . Comparing our findings with experimental observations , we found that the sensing choices observed experimentally agrees with that predicted based on their values of β . This does not mean that the other requirements on the diffusion rates are not important . We speculate that when β is low , the proteins would evolve over time to fulfil the requirements for spatial sensing as this strategy can yield much higher output than temporal sensing . We use the simplest form of NFB and IFF circuitries to study chemotaxis as they capture the requirements of perfect adaptation and senstivity in chemotaxis . Previous mathematical models put forward to explain spatial sensing [11 , 13 , 42–44] are more complicated versions of these circuits . Levchenko’s model is similar to the IFF circuit [13]; Hans’ model is an IFF circuit with two inhibitors [42]; and Ma’s model consists of two competing IFF circuits acting in parallel [11] . Narang’s and Rappel’s models are reminiscent of the NFB loop as simulation of an activator leads to production of a second messanger that inhibits the activator [43 , 44] . Through our simulations , we find that D A ′ < D B ′ is required for spatial sensing to be favored over temporal sensing . Indeed , this is also an assumption found in the previous models [11 , 13 , 42–44] . For example , membrane-bound PI4 , 5P2 with low diffusivity and cytosol IP3 with high diffusivity are proposed to be the activator and inhibitor respectively in Narang’s model [44]; whereas locally acting PI3-kinase and globally acting inactivator PTEN are suggested to be the activator and inhibitor respectively in Levchenko’s model [13] . There have also been recent works that modify the NFB and IFF circuitries to account for other behaviours observed in chemotaxis like fold-change detection of the chemoattractant [45] and rectified directional sensing [46] . Nonetheless , our simple circuits are able to reproduce the conditions necessary in other more complicated mathematical models and serve as good starting points for analyzing the impact of the different parameters . NFB circuit is often associated with temporal sensing [9 , 10] whereas IFF circuit is associated with spatial sensing [11 , 13] . This is in part historical due to the success of the NFB circuit in explaining temporal sensing for E . coli and IFF circuit in explaining spatial sensing for social amoeba . Recently , it was found that the IFF circuit , but not the NFB circuit , can explain the faster protein adaptation dynamics in social amoeba at higher chemoattractant stimulation [12] . This is a clear experimental result supporting the use of IFF circuit for spatial sensing while ruling out the NFB circuit . Here , we found that both the circuits yielded similar results and could be use for both temporal and spatial sensing . It remains to be seen whether the NFB circuit offers general advantages over IFF circuit for temporal sensing and vice versa . To answer this question , one needs to find inputs where the two circuits behave differently . Indeed , it was found that when subjected to a ramp input , IFF circuit adapt perfectly whereas NFB circuit come to a steady-state activity proportional to the gradient of the ramp [47] . It is argued that temporal sensing , which involves sampling concentrations in space , is measuring the rate of change of the input . This makes the NFB suited for temporal sensing as its steady state response is proportional to the gradient of the ramp [48] . Incorporating noise into the external chemoattractant , we found that spatial sensing is more suspectible to noise than temporal sensing . This suggests that temporal sensing which averages over the signal across the entire cell is more robust than spatial sensing which takes the difference between the front and back of the cell . However , we note that when for low β , spatial sensing still performs better than temporal sensing for low noise level . This may be parameter regime where cells adopting spatial sensing operates in . Strategies like receptor coupling [49] , memory [50] and cell-cell communication [51] could also be used in combination with our basic negative integral feedback and incoherent feedforward circuitries to buffer effects of noise and improve performance of spatial sensing .
|
Unicellular organisms and other single cells often have to migrate towards food sources or away from predators by sensing chemicals present in the environment . There are two ways for a cell to sense these external chemicals: temporal sensing , where the cell senses the external chemical at two different time points after it has moved through a certain distance , or spatial sensing , where the cell senses the external chemical at two different locations on its cellular surface ( e . g . , the front and rear of the cell ) simultaneously . It has been thought that small unicellular organisms employ temporal sensing as their small size prohibits sensing at two different locations on the cellular surface . Using computational modeling , we find that the choice between temporal and spatial sensing is determined by the ratio of cell velocity to the product of cell diameter and rate of signaling , as well as the diffusivities of the signaling proteins . Predictions from our model agree with experimental observations over a wide range of cells , where a fast-moving , small cell performs better comparing the chemoattractant at different times in its trajectory; whereas , a slow-moving , big cell performs better by comparing the chemoattractant concentration at its two ends .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion"
] |
[
"blood",
"cells",
"cell",
"motility",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"engineering",
"and",
"technology",
"signal",
"processing",
"immunology",
"germ",
"cells",
"protozoans",
"network",
"analysis",
"cellular",
"structures",
"and",
"organelles",
"neutrophils",
"sperm",
"computer",
"and",
"information",
"sciences",
"white",
"blood",
"cells",
"network",
"motifs",
"animal",
"cells",
"amoebas",
"chemistry",
"cell",
"membranes",
"chemotaxis",
"physics",
"mass",
"diffusivity",
"eukaryota",
"cell",
"biology",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"physical",
"sciences",
"chemical",
"physics",
"organisms"
] |
2018
|
A computational model for how cells choose temporal or spatial sensing during chemotaxis
|
Wolbachia pipientis is an endosymbiotic bacterium estimated to chronically infect between 40–75% of all arthropod species . Aedes aegypti , the principle mosquito vector of dengue virus ( DENV ) , is not a natural host of Wolbachia . The transinfection of Wolbachia strains such as wAlbB , wMel and wMelPop-CLA into Ae . aegypti has been shown to significantly reduce the vector competence of this mosquito for a range of human pathogens in the laboratory . This has led to wMel-transinfected Ae . aegypti currently being released in five countries to evaluate its effectiveness to control dengue disease in human populations . Here we describe the generation of a superinfected Ae . aegypti mosquito line simultaneously infected with two avirulent Wolbachia strains , wMel and wAlbB . The line carries a high overall Wolbachia density and tissue localisation of the individual strains is very similar to each respective single infected parental line . The superinfected line induces unidirectional cytoplasmic incompatibility ( CI ) when crossed to each single infected parental line , suggesting that the superinfection would have the capacity to replace either of the single constituent infections already present in a mosquito population . No significant differences in fitness parameters were observed between the superinfected line and the parental lines under the experimental conditions tested . Finally , the superinfected line blocks DENV replication more efficiently than the single wMel strain when challenged with blood meals from viremic dengue patients . These results suggest that the deployment of superinfections could be used to replace single infections and may represent an effective strategy to help manage potential resistance by DENV to field deployments of single infected strains .
The endosymbiotic bacterium Wolbachia pipientis was first discovered in 1924 by Marshall Hertig and Burt Wolbach in ovaries of the mosquito Culex pipiens [1] . Wolbachia is a Gram-negative , obligate endosymbiont that is maternally transmitted [2] . It is estimated that around 40–75% of all arthropod species are infected with Wolbachia [3 , 4] and the phenomenal success of this bacterium has been attributed to its ability to manipulate the reproductive biology of its host to provide it with a vertical transmission advantage in host populations [5] . These manipulations include feminization , parthenogenesis , cytoplasmic incompatibility ( CI ) and male-killing [6 , 7] . Of these reproductive phenotypes , CI is probably the best studied and describes the phenomenon of early embryonic death resulting from crosses between an infected male and uninfected female or in crosses involving two different Wolbachia strains [7 , 8] . More recently , Wolbachia has been shown to limit pathogen replication , in particular the enveloped , positive single-stranded RNA viruses such as dengue ( DENV ) , yellow fever ( YFV ) and chikungunya ( CHIKV ) [9–12] . Wolbachia also inhibits additional human pathogens transmitted by mosquitoes including filarial nematodes [13] and malaria parasites [14–16] . The mechanism of pathogen inhibition by Wolbachia is still being investigated , but blocking has been linked to priming of the host innate immune system and competition for limited resources between pathogens and Wolbachia [17 , 18] . The ability of Wolbachia to limit pathogen replication has led to the field deployment of Ae . aegypti transinfected with two Drosophila Wolbachia strains , wMel and wMelPop-CLA [19 , 20] . wMelPop-CLA is a pathogenic strain that grows to high densities in insect hosts and infected adult insects have significantly reduced lifespan [21] . In contrast , the closely related wMel strain is avirulent and grows to a lower density in most insect tissues . Correspondingly , total DENV inhibition in whole adult wMel-infected mosquitoes is lower than in wMelPop-CLA infected mosquitoes [12] . However , key to the success of such an approach is the use of Wolbachia strains that can successfully invade wild mosquito populations through the action of CI . The wMelPop-CLA Wolbachia strain imposes significant fitness costs to Aedes mosquitoes including reducing fecundity and egg longevity [9 , 12 , 22 , 23] . Although the wMelPop-CLA strain has a stronger inhibitory effect on total DENV replication in whole mosquito bodies , the significant fitness costs were predicted to prevent invasion of wild mosquito populations [24] . Semi-field cage experiments revealed that the wMel strain would likely invade wild mosquito populations at a faster rate than the virulent wMelPop-CLA strain [12] . Based on these findings , the wMel strain was released into two suburbs of Cairns , Australia in 2011 and reached fixation in mosquito populations within a few months [19] . The avirulent Wolbachia strain wAlbB , transinfected from closely related Aedes albopictus mosquitoes , also inhibits DENV replication in Ae . aegypti with smaller fitness costs than wMelPop-CLA [25] . If avirulent Wolbachia strains such as wMel or wAlbB induce the most favourable phenotypic effects for establishment in wild mosquito populations , the potential long-term development of resistance to the inhibitory effects on DENV must be considered . A strategy to overcome the potential development of DENV resistance to either the wMel or wAlbB strains in wild mosquito populations is to release a superinfected line that would ‘sweep over’ the existing single infection . In this study , we describe the generation of an Ae . aegypti mosquito line co-infected with Wolbachia strains wMel and wAlbB . The CI attributes of this superinfected line , named wMelwAlbB , indicate the superinfection should replace either single infection in a population and as such provide a potential mechanism to address resistance if it were to develop . In addition , the superinfected strain shows fitness costs compatible with a successful field deployment and inhibition of DENV that is predicted to have a large impact on dengue transmission in human populations .
Total Wolbachia density in the superinfected Ae . aegypti line was determined using qPCR and primers specific for the gene encoding the Wolbachia surface protein ( wsp ) in conjunction with the Ae . aegypti rps17 gene to ‘normalise’ for differences in mosquito size . After infection densities had stabilized by generation 18 ( G18 ) , the total Wolbachia density in the wMelwAlbB line was higher than in either parental line and comparable to the virulent wMelPop-CLA strain ( Fig 1A ) . The tissue localization within adult female mosquitoes of both the wMel and wAlbB Wolbachia strains in the superinfected line was determined by fluorescence in situ hybridisation ( FISH ) in formaldehyde-fixed , paraffin-embedded tissue sections using specific probes against wMel ( labelled in red ) and wAlbB ( labelled in green ) ( Fig 1B ) . The Wolbachia tissue tropism in the superinfected line was compared with the wMel and wAlbB strains in the parental , single infected lines . We confirmed the specificity and lack of cross-reactivity of the wMel and wAlbB FISH probes by using both probes against each of the parental lines . No Wolbachia signal was detected in wAlbB mosquitoes when using the wMel probe , and vice versa . Our FISH studies demonstrated the coexistence of both strains in various tissues within the adult female mosquito body . As expected for maternally transmitted symbionts , both wMel and wAlbB strains were particularly abundant in the ovaries ( Fig 1B ) . In addition , both strains were also found to co-localise in somatic tissues such as fat body , nervous tissue ( e . g . thoracic ganglia ) , Malpighian tubules and salivary glands ( S1 Fig ) . The density of wAlbB in all these tissues was similar in the wMelwAlbB line as in the single wAlbB-infected line . However , wMel was more abundant in the Malpighian tubules , fat body and muscle from the super infected line than in the parental wMel line . The density of wMel in salivary glands appeared to be similar in the super infected Ae . aegypti line as in the single wMel line . Interestingly , the wMel and wAlbB Wolbachia strains showed quite distinct localisation patterns in ovaries of superinfected wMelwAlbB line females . Whereas wMel was found evenly distributed throughout the whole egg chamber ( nurse cells and oocyte ) , wAlbB was concentrated in the posterior end of the egg chamber that contains the oocyte ( Fig 1B ) . This is similar to the tropism observed in each the parental lines ( Fig 1B ) . These differences in tropism could represent different patterns of binding of these two strains to the host microtubules and dynein as well as kinesin-1 that appear to drive the movement of Wolbachia into the oocyte during oogenesis [26 , 27] . Maternal transmission was determined from crosses between wMelwAlbB infected females and uninfected wild type males . We observed 100% maternal transmission for wAlbB across all generations and 97% , 98% and 100% transmission for wMel across generations G12 , G14 and G17 respectively ( Table 1 ) . Cytoplasmic incompatibility ( CI ) was determined by setting up a series of reciprocal crosses between wild type , wMel , wAlbB and wMelwAlbB infected mosquitoes . Viable offspring from each of the crosses was used to determine the level of CI induced by the wMelwAlbB line . Egg hatch rate percentages from different crosses are summarised in Table 2 . Crosses between wMelwAlbB infected females and wild type males as well as males infected with wMel , wAlbB and wMelwAlbB resulted in viable offspring while the reciprocal crosses resulted in no viable offspring . To determine the mosquito fitness costs of Wolbachia superinfection , the longevity ( Fig 2 ) and fecundity and egg survival ( Fig 3 ) of the superinfected line were compared to both uninfected mosquitoes as well as each parental infected line . To test the extent to which DENV replication is relatively inhibited in the wMelwAlbB line , we first challenged wild type , wMel , wAlbB , wMelPop-CLA and wMelwAlbB infected mosquitoes with DENV-2 using intrathoracic injections . A DENV-2 strain ET300 was injected at a titre of 104 genome copies/mL and mosquitoes were incubated for 7 days . Positive strand DENV-2 RNA genome copies were detected and quantified in whole mosquito bodies using qRT-PCR . Consistent with previous findings , we saw a significant ~ 1 log reduction of DENV-2 genome copies in wMel and wAlbB , whilst in wMelPop-CLA mosquitoes , DENV-2 genome copies were dramatically reduced by ~ 4 logs ( Fig 4A ) . No significant differences in DENV-2 copies between the wMelwAlbB superinfected line and each of the parental lines were observed ( Fig 4A ) . However , DENV-2 infection rates ( calculated as the percentage of DENV-2 infected mosquitoes of the total injected ) in wMelwAlbB were consistently lower ( 69% ) than both wMel ( 89% ) ( Fisher’s exact test , p = 0 . 034 ) and wAlbB ( 100% ) ( Fisher’s exact test , p>0 . 0001 ) . We next challenged wild type , wMel , wAlbB and wMelwAlbB mosquitoes with DENV-2 ( ET300 ) by oral feeding . Defribrinated sheep blood was inoculated with 107 DENV genome copies per ml and 5–6 day old females from each line were allowed to feed for 2 hours using artificial feeders . Fully fed females were selected and incubated for 14 days . Positive strand DENV-2 RNA genome copies were detected and quantified in whole mosquito bodies using qPCR . We found a significant ~1 . 5 log reduction in DENV-2 genome copies in wMel , wAlbB as well as wMelwAlbB mosquitoes compared to wild type . No significant difference in DENV-2 genome copies between the three Wolbachia-infected lines were found ( Fig 4B ) . We did observe non-significant , lower DENV infection rates in the wMelwAlbB infected line ( 15% ) as compared to the wMel ( 30% ) ( Fisher’s exact test , p = 0 . 41 ) and wAlbB ( 35% ) ( Fisher’s exact test , p = 0 . 24 ) infected lines ( Fig 4B ) . We then assessed the susceptibility of wild type , wMel and wMelwAlbB mosquitoes to DENV infection after feeding on human viremic blood from 43 dengue patients admitted to the Hospital for Tropical Diseases in Ho Chi Minh City , Vietnam . Two feeds were excluded from analysis; a flow chart describing the number of blood fed mosquitoes , their survival and the final cohorts for analyses are described in S2 Fig . The characteristics of the 41 blood donor patients are shown in S1 Table . DENV-1 and DENV-4 were the predominant infecting serotypes in the patient donors ( 88% of infectious feeds ) . The wMel and superinfected wMelwAlbB lines had lower frequencies of DENV infection than wild-type mosquitoes in abdomens and saliva ( Fig 5 and Table 3 ) . Across all time points , a total of 42 . 65% of wild-type mosquitoes had infectious saliva versus 6 . 57% for wMel and 2 . 89% for wMelwAlbB ( adjusted odds ratio ( OR ) 0 . 065; 95% CI = 0 . 038–0 . 112; p <0 . 001 for wMel , and OR 0 . 025; 95% CI = 0 . 014–0 . 043; p < 0 . 001 for wMelwAlbB versus wild-type ) ( Table 3 ) . wMelwAlbB further reduced the risk of females having infectious saliva compared to wMel-infected females ( OR = 0 . 377; 95% CI = 0 . 196–0 . 725; p = 0 . 003 ) . In addition , Wolbachia-infected mosquito strains also had significantly lower concentrations of DENV RNA in their abdomen and salivary gland tissues compared to wild-type mosquitoes ( Fig 6A and 6B and S2 Table ) . wMelwAlbB blocked DENV infection in the salivary glands more efficiently than wMel ( Fig 6B ) . Collectively , these data generated using clinically-relevant virus challenge methods , suggest that the wMelwAlbB strain delivers an incrementally improved DENV blocking phenotype compared to wMel .
Wolbachia has been shown to inhibit pathogen replication in both natural and transinfected insects [9–12 , 18 , 20] . Combined with Wolbachia’s remarkable evolutionary adaptations to ensure rapid spread and transmission , [5 , 28] this bacterium holds promise as an effective biocontrol agent against mosquito-borne diseases such as dengue [20] . Trials with the wMel strain of Wolbachia have shown its establishment and spread in both semi-field [12] and wild populations of Aedes aegypti mosquitoes [19] . However , not all Wolbachia strains are suitable for use in biocontrol strategies . The virulent wMelPop-CLA strain , for example , results in greater overall inhibition of DENV replication in adult female mosquitoes than the avirulent Wolbachia strains , but imparts significantly higher fitness costs [11 , 12 , 29] . Preliminary trials in Australia and Vietnam in which the wMelPop-CLA strain was released into wild mosquito populations indicate that these fitness costs prevented successful establishment [30] . Modelling projections suggest the establishment of Wolbachia strains in dengue endemic settings will result in a substantial reduction in disease burden [31] . The persistence of an inhibitory effect on DENV replication within wild Wolbachia-infected mosquitoes will be key to the success of any release program . Laboratory vector competence experiments with field ( F1 ) wMel-infected Ae . aegypti mosquitoes , collected one year following field release , indicated very low levels of DENV replication and dissemination [32] , demonstrating the persistence of the virus inhibition phenotype . The potential evolution of DENV resistance to Wolbachia’s inhibitory effects must be considered if this biocontrol strategy can be sustainable on a long-term basis . However , the ability to predict the likelihood of resistance development in virus populations will require a greater understanding of the mechanisms of Wolbachia-mediated viral inhibition . Host immune stimulation has been shown to result in antiviral effects in Ae . aegypti [10 , 25 , 33] but this is not universal for all Wolbachia-mediated antiviral inhibition [34–36] . The density and tissue tropism of Wolbachia strains in insect hosts appears to be the most important factors [12 , 37 , 38] and competition for shared host resources such as cholesterol has been shown to influence the strength of Wolbachia-induced antiviral effects [17] . High density Wolbachia strains in Drosophila flies provide strong inhibitory effects on insect viruses despite a long-term evolutionary association [11 , 39] . Thus , the non-specific nature of the anti-viral environment in Wolbachia-infected Ae . aegypti tissues , coupled with the dominant evolutionary process of purifying selection in DENV populations[40] , such that minor variant viruses that arise within individual hosts are lost because they are not infectious to both humans and mosquitoes , creates significant barriers to the emergence of DENV strains that are resistant to Wolbachia . Nonetheless , the association between density and viral inhibition in these natural Wolbachia-host endosymbiotic relationships suggest resistance is less likely to develop for Wolbachia strains that grow to high densities in transinfected insect hosts . Therefore , a superinfection that results in a cumulative higher density Wolbachia infection would be predicted to reduce the potential for DENV resistance development in Ae . aegypti . In the event DENV does evolve resistance to either the wMel or wAlbB strains in wild mosquito populations , one potential option would be to release a superinfected line that would ‘sweep over’ the existing single infection . For this resistance management strategy to be effective , favourable CI spread dynamics would be needed for a superinfected line to replace existing single Wolbachia infections in wild mosquito populations . The crossing patterns induced by wMelwAlbB ( Table 2 ) indicate that either the wMel or wAlbB strain could be replaced by a superinfection in wild mosquito populations . The density of Wolbachia strains in transinfected Ae . aegypti mosquitoes is also correlated with mosquito fitness costs [12] . The additive density of Wolbachia strains in the superinfected line , measured at G18 when the line was stable , was comparable to the virulent wMelPop-CLA strain ( Fig 1B ) . Despite the superinfected line resulting in a cumulative high density Wolbachia infection , the effects on the majority of mosquito fitness parameters were very similar to that observed for the single infected wMel line . Under laboratory conditions superinfected males and females had a marginally shorter adult lifespan than uninfected wild type mosquitoes ( ~10% reduction ) . The observed effects on adult mosquito longevity of the superinfected line are significantly less than those for the virulent wMelPop-CLA strain , which reduces the lifespan of adult Ae . aegypti mosquitoes by approximately ~50% [9] . In our study , no differences in the number of eggs laid by females ( fecundity ) from the superinfected line compared to wMel , wAlbB or wild type mosquitoes were observed . Under semi-field conditions , the virulent wMelPop-CLA strain reduced fecundity of Ae . aegypti females by ~60% [12] , which may have contributed to the inability of this strain to invade wild mosquito populations [41] . Minimal fecundity costs should increase the potential of the superinfected line to ‘sweep over’ existing single infections in wild mosquito populations . In contrast , survival of eggs from superinfected females during periods of embryonic quiescence was significantly lower than either parental line or wild type mosquitoes . Following two months of storage , ~50% of superinfected eggs were still viable . Although the hatch rates for the superinfected line were lower than that observed for the wAlbB- infected line , the hatch rates were very similar to that of the wMel infected line . Furthermore , hatch rates are still within the average 2-month survival rates ( 40–60% ) for Ae . aegypti eggs during dry seasons [42 , 43] . Further experiments under semi-field conditions will be needed to fully determine if the effect on embryonic quiescence is likely to impact the ability of the superinfected line to invade uninfected wild mosquito populations . The results of field releases to date ( using wMel ) suggest this is unlikely to be a major obstacle to establishing superinfections in the field . The wMel strain successfully invaded wild mosquito populations [19] and the infection remains stable in these release areas [44] despite the observed reduction on embryo hatch rates under laboratory conditions . The release of a superinfected line for virus resistance management would require the co-infection to provide strong inhibitory effects on DENV replication . Vector competence experiments carried out under laboratory conditions indicated all Wolbachia lines significantly reduced DENV replication as previously reported [12 , 25] , however the superinfected line provided the greatest resistance . After oral feeding on fresh human viremic blood , the most relevant model to assess mosquito susceptibility to DENV , very few superinfected mosquitoes had infectious virus in their saliva and viral RNA concentrations were substantially reduced in mosquito tissues . These data give reassurance that any population replacement strategy with the superinfected line would be expected to deliver stronger inhibition of DENV transmission than is conferred by wMel . In summary , the generation and characterisation of a superinfected line with the desired phenotypic effects to replace single wild infections provides a potential mechanism to overcome the emergence of DENV resistance . Both Wolbachia strains are stably maintained in the line with minimal mosquito fitness effects . Importantly , DENV replication is inhibited to a greater extent in the superinfected line compared to both parental lines . The observed CI phenotype induced by the superinfected line is of particular significance as it would enable the line to be released “on top of” existing wMel or wAlbB field releases in dengue endemic areas .
Wolbachia-uninfected Ae . aegypti eggs were collected from Cairns ( Queensland , Australia ) in 2013 ( JCU wild type ) . The Wolbachia-infected wMel and wAlbB mosquito lines have been described previously [12 , 25] . All Ae . aegypti mosquitoes were reared and maintained as described in [9] with the following modification . For hatching , eggs were placed in hatching water ( distilled H2O , boiled and supplemented with 50 mg/L fish food [Tetramin] ) and allowed to hatch for 24 h . Larvae were subsequently reared at a set density of ~150 in 3 L of distilled water as described in [9] . To prevent genetic drift between wild type and the Wolbachia infected mosquito lines used for analyses , females from each generation of the infected lines were backcrossed with a small proportion ( 10% ) of uninfected field collected male mosquitoes . Embryonic microinjection , isofemale line establishment and selection for stably-infected lines were done as previously described [9] . In short , the wMel strain was purified from wMel-infected mosquitoes and microinjected into the posterior-pole of wAlbB-infected preblastoderm embryos using methodology previously described [12] . Surviving G0 adult females ( ~600 ) from microinjection were mated to wild type males and blood fed for oviposition of the G1 generation . G0 females that laid fertile egg batches were screened using quantitative PCR as described by [17] and primers specific for wMel ( forward primer: 5’-CAAATTGCTCTTGTCCTGTGG-3’ , reverse primer: 5’-GGGTGTTAAGCAGAGTTACGG-3’ ) and wAlbB ( forward primer: 5’-CCTTACCTCCTGCACAACAA-3’ , reverse primer: 5’-GGATTGTCCAGTGGCCTTA-3’ ) . For each sample , quantitative PCR amplification of DNA was performed in duplicate with a LightCycler 480 II Instrument ( Roche ) using LightCycler 480 SYBR Green I Master ( Roche ) according to the manufacturer’s protocol . From the ~600 females screened , 21 wMel positives were identified and pooled into two lines . The female progeny from both lines of superinfected females were mated to uninfected field-males for 5 generations ( G0-G4 ) before the lines were considered stably infected with both strains of Wolbachia . One line was selected for further characterisation . Wolbachia density and distribution in the superinfected line was compared to each of the parental strains using qPCR and fluorescence in situ hybridisation ( FISH ) . Quantitative PCR to determine the total relative Wolbachia densities of infected lines was performed as described by [17] using primers specific to the gene coding for the Wolbachia surface protein ( wsp ) ( forward primer 5’ GCATTTGGTTAYAAAATGGACGA-3’ , reverse primer 5’- GGAGTGATAGGCATATCTTCAAT-3’ ) , as well as the Ae . aegypti rps17 gene ( forward primer 5’-TCCGTGGTATCTCCATCAAGCT-3’ , reverse primer: 5’-CACTTCCGGCACGTAGTTGTC-3’ ) . Wolbachia was localized in sections of paraffin-embedded 5–7 day old female mosquitoes by FISH , as described in [10] , except that only one probe against 16S rRNA was used against each strain and their concentration was increased by 10-fold to improve the signal . wMel was detected using the probe MelPopW6: 5’-GCTTAGCCTCGCGACTTTGCAG-3’ , labelled with Alexa 594 dye ( red ) , whereas wAlbB was localized using AlbBW5: 5’-CTTAGGCTTGCGCACCTTGCAA-3’ , labelled with Alexa 488 dye ( green ) . 16S rRNA is highly conserved between wMel and wAlbB , therefore the probe was designed against a part of the gene that includes several SNPs . We confirmed the specificity and lack of cross-reactivity of each probe by testing them against the single infected lines ( wMel and wAlbB ) . Both probes were added simultaneously to the wMel , wAlbB and wMelwAlbB mosquito sections in order to obtain the images . DAPI was also used to stain total DNA . The propagation and maintenance of dengue virus serotype 2 ( DENV-2 ) ET300 was carried out as previously described [18] . For adult microinjections , 40 Ae . aegypti female mosquitoes were anesthetized by briefly exposing them to -20°C . The mosquitoes were subsequently injected intrathoracically with 50 nL of virus solution ( 104 genomic copies/ml in RPMI [Sigma-Aldrich] media ) using a pulled glass capillary and a handheld microinjector ( Nanoject II , Drummond Sci . ) . Injected mosquitoes were incubated for 7 days ( 40 mosquitoes per cup ) at 26°C with 65% relative humidity and a 12h light/dark cycle . For feeding experiments with DENV-2 ( ET300 ) infected blood , 80 Ae . aegypti female mosquitoes were placed in 500 mL plastic containers , starved for 25 hours and allowed to feed on a 50:50 mixture of defibrinated sheep blood and tissue culture supernatant containing 107 genome copies/mL of DENV-2 . Feeding was done through a piece of desalted porcine intestine stretched over a water-jacketed membrane feeding apparatus preheated to 37°C for approximately three hours . Fully engorged mosquitoes were placed in 500 mL containers and incubated for 14 days at 26°C with 65% relative humidity and a 12h light/dark cycle . To quantify DENV-2 genomic copies , total RNA was isolated from DENV-2 injected mosquitoes using the Nucleospin 96 RNA kit ( Macherey-Nagel ) . DENV-2 qPCR analysis was done using cDNA prepared from individual mosquitoes according to [10] . Statistical significance for differences in DENV titres between treatments was determined using a one-way ANOVA with Tukey’s multiple comparison tests ( Graph Pad Prism 6c ) . Cohorts of 3–5 day old mosquitoes were allowed to feed on fresh , viremic blood from 43 NS1 rapid test-positive patients admitted to the Hospital for Tropical Diseases , in Ho Chi Minh City , Vietnam . Mosquitoes were fed via membrane feeders for a maximum of 1 hour . Fully engorged mosquitoes were placed in 250 mL containers and incubated at 28°C/80% humidity with a 12h light/dark cycle . Mosquitoes were harvested from each blood fed cohort 10 , 14 and 18 days later . Detection of infectious virus in the saliva of each mosquito was as described previously [45] . Statistical analyses were performed with the statistical software R , version 3 . 1 . 3 ( R Foundation for Statistical Computing , Vienna , Austria ) . Marginal regression models for binary ( infected/uninfected mosquitoes ) and continuous ( tissue viral load ) outcomes were fitted using generalized estimating equations with working exchangeable correlation structure to account for potential within-patient correlation . Blood feeding by volunteers ( Monash University human ethics permit no CF11/0766-2011000387 ) for this study was approved by the Monash University Human Research Ethics Committee ( MUHREC ) . All adult volunteers provided informed written consent; no child participants were involved in the study . The protocol for feeding mosquitoes with viremic human blood was reviewed and approved by the Ethics Committee of Hospital for Tropical Diseases ( HTD ) , Ho Chi Minh City , Vietnam ( approval number CS/ND/12/16 ) , and the Oxford University Tropical Research Ethics Committee ( OxTREC ) ( approval number OxTREC 30–12 ) . All enrolled subjects provided informed written consent .
|
Dengue fever is a viral disease transmitted by Aedes aegypti mosquitoes and more than 30% of the world’s population is at risk . The control of dengue virus ( DENV ) transmission has been problematic as no vaccines or drugs are effective against the four serotypes . Vector control of mosquitoes during epidemics is considered the only option to prevent transmission . Recently , a novel biocontrol method using the endosymbiotic bacterium Wolbachia has been developed in which DENV replication is significantly inhibited in Wolbachia-infected Ae . aegypti . This bacterium also induces a reproductive phenotype called cytoplasmic incompatibility that allows rapid invasion of uninfected mosquito populations . Like any control method , evolutionary responses are expected of the system that might limit its future effectiveness . Here we report the generation and characterization of a superinfected Ae . aegypti line containing two Wolbachia strains ( wMel and wAlbB ) . We show that stable Wolbachia superinfections are more effective at blocking dengue than single infections . Superinfections also demonstrate a cytoplasmic incompatibility phenotype that should enable them to replace single infections in the field . This represents a potential mechanism for resistance management in regions where single infections have already been deployed .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"abdomen",
"microbiology",
"saliva",
"animals",
"wolbachia",
"insect",
"vectors",
"bacteria",
"digestive",
"system",
"infectious",
"diseases",
"epidemiology",
"exocrine",
"glands",
"superinfection",
"viral",
"replication",
"hematology",
"disease",
"vectors",
"insects",
"arthropoda",
"mosquitoes",
"blood",
"anatomy",
"virology",
"physiology",
"salivary",
"glands",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2016
|
Establishment of a Wolbachia Superinfection in Aedes aegypti Mosquitoes as a Potential Approach for Future Resistance Management
|
The YbeB ( DUF143 ) family of uncharacterized proteins is encoded by almost all bacterial and eukaryotic genomes but not archaea . While they have been shown to be associated with ribosomes , their molecular function remains unclear . Here we show that YbeB is a ribosomal silencing factor ( RsfA ) in the stationary growth phase and during the transition from rich to poor media . A knock-out of the rsfA gene shows two strong phenotypes: ( i ) the viability of the mutant cells are sharply impaired during stationary phase ( as shown by viability competition assays ) , and ( ii ) during transition from rich to poor media the mutant cells adapt slowly and show a growth block of more than 10 hours ( as shown by growth competition assays ) . RsfA silences translation by binding to the L14 protein of the large ribosomal subunit and , as a consequence , impairs subunit joining ( as shown by molecular modeling , reporter gene analysis , in vitro translation assays , and sucrose gradient analysis ) . This particular interaction is conserved in all species tested , including Escherichia coli , Treponema pallidum , Streptococcus pneumoniae , Synechocystis PCC 6803 , as well as human mitochondria and maize chloroplasts ( as demonstrated by yeast two-hybrid tests , pull-downs , and mutagenesis ) . RsfA is unrelated to the eukaryotic ribosomal anti-association/60S-assembly factor eIF6 , which also binds to L14 , and is the first such factor in bacteria and organelles . RsfA helps cells to adapt to slow-growth/stationary phase conditions by down-regulating protein synthesis , one of the most energy-consuming processes in both bacterial and eukaryotic cells .
Escherichia coli harbors a core set of about 190 genes that are conserved in more than 90% of all completely sequenced genomes [1] . Most of them encode well-understood proteins involved in metabolism , transcription , translation , or replication . However , a few of these highly conserved proteins remain functionally uncharacterized and thus enigmatic . One of these mysterious proteins is YbeB . In 2004 it was proposed by Galperin and Koonin as one of 10 top targets of conserved hypothetical proteins for experimental characterization [2] . In recent interactome studies , we and others found this protein to interact with various proteins , including several ribosomal components [3] , [4] , [5] , [6] . Moreover , YbeB was shown to co-sediment with the large ribosomal subunit ( LRS ) [7] , suggesting that it functions in protein translation . Recently it has been suggested that its mitochondrial homologue , C7orf30 , is involved in ribosome biogenesis and/or translation [5] , [8] although these studies were not able to explain their observations mechanistically . In this work we characterize YbeB's molecular function by identifying its binding site on the LRS and reveal a molecular mechanism of YbeB action: it is down-regulating protein synthesis under nutrient shortage by binding to protein L14 of the LRS , acting as a ribosomal silencing factor ( “RsfA” ) by blocking ribosome subunit joining . Thus , we will use the term “RsfA” below .
In the Pfam database ( V26 . 0 ) RsfA sequence homologues are known for at least 2 , 928 species , including nearly all bacteria as well as almost all eukaryotic species ( Pfam entry PF02410 , Interpro IPR004394 ) . However , the RsfA protein family is conspicuously absent in archaea ( Figure 1A ) . In the STRING 9 . 0 database [9] RsfA is clustered with the orthologous protein group “COG0799” , consisting of 932 RsfA homologues in 920 different species , indicating that there is usually one rsfA gene per genome . A multiple sequence alignment of ten representative RsfA orthologues , however , exhibits only limited conservation when compared to ribosomal protein L14 ( Figure S1 ) . Interestingly , more than 80% of all eukaryotic RsfA orthologues are predicted to localize to mitochondria or chloroplasts according to the WoLF PSort program [10] . For the yeast orthologue ATP25 , the mitochondrial localization has been experimentally confirmed [11] and the Zea mays homologue , Iojap , was found in chloroplast fractions [12] . This strongly suggests that RsfA functions in a strictly conserved process of bacterial origin . Previously , Butland and colleagues reported L14 , L19 , L4 , L7/L12 and others as interaction partners of RsfA based on protein complex data [3] . Similarly , we found that several interactors of RsfA's Treponema pallidum orthologue TP0738 were involved in protein synthesis [6] . Although these observations provided the first experimental hint that RsfA might function in translation , this has never been functionally demonstrated . Since previous studies have revealed RsfA's association with the large ribosomal subunit ( LRS ) which offers multiple binding sites , we re-tested all previously detected interactions of T . pallidum RsfA that are involved in protein translation . As expected , several proteins indeed tested positive ( Figure 1B ) . However , the interaction of RsfA with L14 was by far the strongest as determined by using increasing concentrations of 3-amino-triazole ( 3-AT ) , a competitive inhibitor of the yeast two-hybrid reporter gene HIS3 . In fact , only the interaction with L14 was detectable at more than 1 mM 3-AT . Furthermore , the L14-RsfA interaction was the only one that was detectable in a reciprocal screen , i . e . with RsfA used as both bait and prey . Given the conservation of RsfA , we wanted to establish to which extent the interactions of RsfA of T . pallidum are conserved in other species . To this end , we first retested whether the interactions of T . pallidum RsfA are conserved in E . coli . We also included eight putative interaction partners that have been identified in a protein complex together with E . coli RsfA and L14 [3] and four interologous pairs detected by Y2H in Campylobacter jejuni [13] . Surprisingly , only the interaction with L14 was conserved in E . coli as a strong ( up to 50 mM 3-AT ) and reciprocal interaction ( Figure 1C , all tested interactions and reference sets are listed in Table S2 and the complete Y2H assays are shown in Figure S2 ) . Moreover , we confirm the interaction of RsfA with L14 from E . coli independently in a pull-down experiment ( Figure S3A ) . Thus , we conclude that L14 is the primary and specific binding target of RsfA on the LRS and that all other interactions are species specific or even artifacts . Next we tested whether this particular interaction is conserved in other bacteria . Notably , we could verify the interaction in all tested species , including gram-positive Streptococcus pneumoniae and the cyanobacterium Synechocystis PCC 6803 ( Figure S3B and S3C ) . In addition , we confirmed the interaction between the corresponding orthologues of RsfA/L14 of both human ( C7orf30/mitochondrial L14 ) and Zea mays ( Iojap/chloroplastic RPL14 ) as shown in Figure 1D and 1E , respectively . In HeLa cells human C7orf30 co-localized with L14mt exclusively to mitochondria ( Figure 1F ) . This supports the hypothesis that eukaryotic RsfA orthologues are functionally active only in organelles . Finally , we verified the human protein interaction in vivo by a bimolecular fluorescence complementation assay using C-terminally tagged Split-Venus constructs ( Figure 1G ) . In summary , these results strongly suggest that the interaction of RsfA and L14 is universally conserved in all species that encode RsfA homologues and that in fact their specific binding site at the LRS is in the ribosomal protein L14 . In order to map the exact binding site of RsfA we used the LRS 3D structure ( PDB id: 2AWB [14]: first , we identified amino acids of L14 that ( i ) are highly conserved ( Figure 2A ( a ) and 2A ( b ) ) and that ( ii ) are located on the surface exposed towards the 30S small subunit interface . These criteria identified T97 , R98 , K114 , and S117 . ( Figure 2A ( b , c ) ) . In fact , docking a homology model of RsfA and a crystal structure of L14 predicted these residues to be at their interaction interface ( Figure 2B ) . In order to test whether the identified residues of L14 are indeed essential for the L14-RsfA interaction , we substituted T97 , R98 , K114 , and S117 with a single alanine each and tested these L14 constructs if they still bound RsfA by another Y2H experiment ( Figure 2C ) : the K114A and T97A mutants lost the interaction with RsfA already in the presence of 0 to 1 mM 3-AT , while in R98A the interaction was lost at 10 mM and higher concentrations . S117A did not appear to affect the interaction . Several control mutations including moderately conserved amino acids ( D80A , F100A , E121A ) and none-conserved ones ( R49A , K51A ) did not show any difference in the Y2H assay compared to the assayed wild type L14 ( Figure 2A , 2C ) . In summary , the interaction epitope assay confirms that the docking model ( Figure 2B ) is largely correct . The RsfA-interaction epitope of L14 involves the highly conserved residues K114 , T97 , and R98 ( but not S117 ) while K114 and T97 are the most critical ones . Notably , T97 and R98 are involved in bridge B8 ( Figure 2A ( d ) ) that contacts the small ribosomal subunit [15] . The docking model predicts that binding of RsfA to these residues , as a consequence , would sterically interfere with ribosome subunit joining ( Figure 2B ( b ) ) and thus might block translation . Although RsfA is phylogenetically highly conserved , its gene deletion has been reported not to result in any obvious growth disadvantage in E . coli [7] , [16] . We designed a sensitive growth experiment , which compares the WT and the rsfA deletion strain under competitive growth conditions: we mixed equal amounts of both cell types and monitored the populations at constant time intervals under log-phase conditions . Figure 3A demonstrates that the amounts of mutant cells decreased continuously . In other words , WT cells in rich medium steadily overgrew the mutant cells leaving only about 10 to 25% of mutant cells after 35 generations . This modest effect reveals that RsfA mutant cells suffer from a disadvantage when competing with WT cells . Strikingly , a much stronger difference was observed , when cells grown in rich medium were diluted in minimal medium: the WT strain overgrew the mutant ΔrsfA strain within only five generations . The opposite growth transition ( poor→rich media ) is better tolerated by the mutant strain . The addition of amino acids to the minimal medium completely rescues this striking growth defect of the rsfA mutant in the rich→poor media transition ( see Discussion ) . These strong defects seen with the ΔrsfA strain in minimal medium rather than in rich medium should be evident also in a direct determination of the doubling times of wild type versus mutant in separate cultures . In rich medium the generation times of WT and mutant strains were not significantly different ( 30 and 32 min , respectively; Figure 3B ) . However , a change from rich to poor medium revealed a dramatic difference: initially the ΔrsfA mutant strain showed a growth like the WT strain for about 7 h , but then growth was abrogated for about 14 h before it resumes almost with the same doubling time as the WT strain ( 130 versus 120 min ) . The growth block for many hours demonstrates that the lack of the rsfA gene poses a serious adaptation problem on the cells after a transition from rich to poor medium . It has been reported that the rsfA ( formerly ybeB ) knock-out can cause a defect in cell separation in a distinct genetic background , and this defect can be complemented with genes of the rsfA operon downstream of the rsfA gene indicating a polarity effect of the rsfA deletion [17] . Therefore , we tested whether we can complement the strong mutant phenotype observed in Figure 3A and 3B by introducing a plasmid carrying the rsfA gene . If so , it would prove that the mutant phenotype is caused by the absence of the RsfA factor . To this end , we removed the kanamycin cassette in place of the chromosomal rsfA gene and introduced a plasmid with the rsfA gene under the native promoter; the expressed RsfA carried a His-tag at the C-terminus to monitor the expression by anti-His antibodies . Figure 3C demonstrates that the mutant phenotype could not be cured probably due to the fact that after the shift to the poor medium RsfA was not sufficiently expressed , whereas taking up growth after 30 h was accompanied by a strong RsfA expression ( see red bars in Figure 3C ) . Therefore , we performed the same experiment but now with the rsfA gene under a tac promoter . The forced RsfA expression could heal the mutant phenotype ( Figure 3D; red closed circles ) . We conclude that ( i ) the RsfA expression is regulated in a way we do not yet understand , and ( ii ) that the lack of RsfA is responsible for the mutant phenotype . Figure 3A and 3B demonstrate that mutant and WT strains showed almost the same growth behavior under log-phase conditions in rich medium ( LB ) . But what happens in a batch culture , when a mixture of both strains reaches the stationary phase in rich medium and protein synthesis has to be down regulated ? This was tested in the next experiment . The stationary phase is reached after about 7 h ( red line in Figure 3E ) . At various time points aliquots were taken and the fraction of ΔrsfA mutant strains were determined ( blue bars ) . Until reaching the stationary phase the fraction of mutant cells remains constant at about 35% , but thereafter the fraction of mutant cells sharply declined to less than 10% . This viability competition assay indicates that the mutant cells have serious problems to form stable stationary-phase cells . The experiments shown in Figure 3A–3E disclose two strong phenotypes caused by the lack of RsfA: ( i ) The cells adapt poorly after the transition from rich to poor media , and ( ii ) the viability of cells is dramatically impaired during the stationary phase , eventually causing cell death . Given RsfA's physical association with the large ribosomal subunit/L14 , we wondered whether RsfA has an effect on protein synthesis . To this end we expressed β-galactosidase ( as an L-arabinose inducible reporter ) in an E . coli gene deletion strain ( ΔrsfA ) and wild type ( WT ) cells . At stationary phase the β-galactosidase expression was strongly repressed in wild type cells as expected ( Figure 3F ) . In striking contrast , the ΔrsfA mutant exhibited a significant accumulation of β-galactosidase in the stationary phase . These results demonstrate that RsfA acts as a negative modulator of protein translation in vivo in the stationary phase . Together with the viability assay ( Figure 3E ) these results suggest that silencing protein synthesis plays an important role for reorganization of the metabolic conversion on the way to the stationary phase . Next we tested whether RsfA interferes with ribosomal elongation in vitro using a highly resolved E . coli system just containing purified elongation factors EF-Tu , EF-Ts , EF-G , purified precharged [14C]Phe-tRNA , poly ( U ) programmed ribosomes and GTP as energy source . We added 30S subunits to an excess of 50S subunits in order to facilitate association to 70S ribosomes . Purified RsfA suppressed the translational activity dramatically down to about 20% , when RsfA was added to the 50S subunits before the oligo ( Phe ) synthesis ( Figure 4A , left panel ) . To test whether RsfA blocks ribosomal activities via interfering with association of the subunits as suggested by our protein docking model ( Figure 2B ) , we subjected an aliquot to a sucrose-gradient analysis before incubating for oligo ( Phe ) synthesis ( Figure 4B ) . The gradients demonstrate that in the absence of RsfA clearly more 70S ribosomes are formed on the cost of ribosomal subunits . However , when RsfA was added to programed 70S ribosomes carrying an AcPhe-tRNA at the ribosomal P site , no inhibition was observed indicating that RsfA does not interfere with ribosomal functions during the elongation phase ( Figure 4A , right panel ) . We conclude that RsfA blocks association of the ribosomal subunits to functional 70S ribosomes . Corresponding experiments with the translational elements of mitochondrial ribosomes from mammalian cells ( pig liver ) confirmed these results . In the presence of purified mitochondrial factors mtEF-Tu , mtEF-Ts , mtEF-G1 , poly ( U ) and [14C]Phe-tRNA oligo ( Phe ) synthesis was severely reduced upon addition of the mitochondrial RsfA orthologue C7orf30 ( mtRsfA; Figure 4C ) . The results suggest that the function of RsfA is conserved from bacteria to eukaryotic mitochondria .
The cellular synthesis machinery runs at high speed in the exponential ( logarithmic ) phase of bacterial growth . The growth rate slows in semi-log phase and finally comes to a halt at higher cell density in the stationary phase , usually caused by nutrient depletion . Several bacterial factors bind to ribosomes and thus support the dormant state of the ribosomes in the stationary phase , such as the ribosome modulation factor ( RMF ) , hibernation promoting factor ( HPF ) or stationary-phase-induced ribosome-associated protein ( SRA ) [18] , [19] , [20] , [21] . RMF ( homologues exist only in the γ-proteobacteria ) alone or together with the more broadly distributed HPF are essential for the formation of 70S dimers in the stationary phase , so called 100S particles; an inactivation of the RMF gene causes a viability defect at prolonged periods in stationary phase [22] , [23] . Phenotypical effects of knock-out strains concerning the other factors have not been reported . A first analysis of RsfA-binding partners identified a group of proteins including a number of ribosomal proteins [6] . Similarly , other groups suggested various ribosomal proteins as binding partners [3] , [4] , [5] , the common denominator being that all proteins were derived from the large subunit . Thorough analyses presented here identified the ribosomal protein L14 as the docking station ( Figure 1B–1G , Figure 2 ) , and mutation of conserved amino acid residues of L14 at the surface of this protein abolished RsfA binding , clearly demonstrating L14 as the binding protein ( Figure 2 ) . Interestingly , the three most conserved residues of RsfA as shown by the multiple sequence alignment ( Figure S1A ) are located at the interface with L14 predicted by docking . The three residues are W120 , D124 and R140 ( alignment numbers ) , corresponding to residue numbers W77 , D81 and R95 in E . coli RsfA . D81 is predicted to be in direct contact with R98 of L14 that was shown to disrupt the interaction when mutated . Another such critical residue , K114 of L14 , is predicted to be in contact with a fairly conserved residue with RsfA L103 ( position 148 in the alignment ) . The only other known protein that like RsfA also docks to the ribosomal protein L14 of eukaryotic ribosomes is the so-called initiation factor eIF6 , which is not a homologue to RsfA and is thought to block ribosome association in archaea and in eukaryotes from yeast to man [24] , [25] , [26] , [27] , [28] , [29] . However , in eukaryotes eIF6 is rather a 60S assembly factor and plays an essential role in the late pre-25S rRNA processing and the export of the 60S subunit from the nucleolus to the cytoplasm [30] . Depletion of eIF6 is eventually lethal , in contrast to RsfA . Interestingly , eIF6 is restricted to the eukaryotic nucleus/cytoplasm and to archaea [27] , while RsfA is present in almost all bacteria and their descendent eukaryotic organelles ( Figure 1A ) . Studies with the human mitochondrial homologue of RsfA , C7orf30 , have recently suggested that this protein is involved in ribosomal assembly and/or translation [5] , [8] . Our results do not indicate any assembly defects as deletion strains of rsfA appear to have perfectly assembled ribosomes ( sucrose gradients not shown ) and actually translate as well as wild type strains at logarithmic phase ( Figure 3F ) . In addition , we could show that C7orf30 inhibits translation by mitochondrial ribosomes ( Figure 4C ) . It remains possible that C7orf30 has multiple roles in mitochondria or that its role in ribosome assembly is indirect . In rich medium bacterial cells produce proteins at maximum rates to sustain cell division . Furthermore , bacterial cells take up many metabolic precursors such as amino acids and thus block corresponding synthesis pathways . In contrast , in poor/minimal medium protein synthesis must be down-regulated in a concerted fashion in order to save energy and resources , and at the same time many synthesis pathways such as those for the synthesis of amino acids have to be switched on [31] , [32] . The results presented here suggest that RsfA plays a prominent role in this down-regulation by silencing ribosome activities . We observe two strong phenotypes with the ΔrsfA strain: ( i ) the viability is strongly impaired in the stationary phase ( Figure 3E ) and ( ii ) after a transition from rich to poor media the adaptation phase lasts more than 10 hours before resuming growth again in striking contrast to WT cells ( Figure 3B ) , which overgrow the mutant strain in a few generations . Just adding casamino acids to the minimal medium relieves the strong growth defects of the ΔrsfA strain ( Figure 3A ) . Adding amino acids will switch off most of the amino-acid synthesis pathways similar to the situation during the logarithmic phase in the presence of rich medium , when the silencing effect of RsfA is not strictly required . In contrast , during starvation and in the absence of ribosomal silencing ( ΔrsfA ) , energy would be wasted affecting the conversion of the metabolic network , eventually causing deleterious growth defects . Accordingly , protein synthesis is seriously attenuated in the stationary phase , when RsfA is present ( i . e . wild type cells ) in contrast to protein synthesis in the ΔrsfA strain ( Figure 3F ) . Attenuation of protein synthesis by RsfA seems to be of utmost importance for reorganization the metabolic state on the way to the stationary phase , since the absence of this factor threatens seriously the viability in the stationary phase ( Figure 3E ) , and it explains the well-known effect that ribosomes are much less active , when derived from the stationary rather than from log-phase cells [33] . When RsfA is added to ribosomal subunits it blocks 70S formation and thus protein synthesis ( Figure 4A and 4B ) , whereas the factor does not interfere with the elongation phase of protein synthesis when added to ribosomes that have passed the initiation phase ( Figure 4A , right panel ) . We conclude that RsfA , as a ribosomal silencing factor , is damping the translational activity under restricted energy ( stationary phase ) or nutrient conditions ( growth in poor medium ) thus harmonizing translation with the general metabolic state , i . e . RsfA works in line with the stringent response [34] and thus plays a key role in the physiology of the stationary phase and the translational adaptation during the transition from rich to poor medium . Our experiments suggest a direct silencing effect of RsfA sketched in Figure 5: when the ribosomal activity should be silenced , RsfA binds to the ribosomal protein L14 at the interface of the large subunit and by impairing association of the ribosomal subunits translation is hampered . We demonstrated that RsfA damps the ribosomal elongation in bacterial and mammalian mitochondrial systems ( Figure 4A and 4C ) . The importance of RsfA in eukaryotic organelles is indicated by the fact that a mutation in the gene of the RsfA orthologue Iojap in Zea mays leads to irregular albino patterns on maize leafs and germless seeds due to failure of proplastids to differentiate into chloroplasts [35] , [36] , [37] , [38] . Photosynthesis and respiration can vary enormously in plastids and mitochondria , respectively , and as suggested by the experiment shown in Figure 4C , the RsfA orthologue might accordingly regulate protein synthesis in these organelles using the mechanism suggested here .
ORFs were cloned into pDONR207 by using the Gateway Technology ( Invitrogen ) . Zea mays cDNA was kindly provided by F . Hochholdinger ( Tübingen , Germany ) , HeLa cDNA by O . Kassel ( Karlsruhe , Germany ) , S . pneumoniae TIGR4 DNA by D . Nelson ( UMBI , MD , USA ) , T . pallidum DNA by T . Palzkill ( Houston , USA ) , and Synechocystis PCC 6803 DNA by T . Lamparter ( Karlsruhe , Germany ) . All ORFs were cloned with a stop codon at the 3′-ends . Entry plasmids were sequenced , shuttled into expression vectors ( see below ) , and finally verified by PCR reactions . For the interologous tests E . coli ORFs were kindly provided as pENTR/Zeo clones by S . V . Rajagopala [39] except for RsfA and L14 which have been cloned in this study . E . coli L14 ( b3310 ) alanine substitutions were directionally introduced by performing standard fusion PCR reactions using mutagenic primers . For cloning PrimeStar HS DNA Polymerase was used ( Takara Bio Inc . ) . Entry plasmids were recombined with the bait and prey vector pGBKT7g and pGADT7g ( Clontech ) [40] . These were individually transformed into the haploid yeast strains AH109 and Y187 [41] , [42] . After mating the haploids and enrichment of diploids , yeast growth was observed on solid starvation medium lacking Leucine , Tryptophan , and Histidine . The medium contained various concentrations of 3-AT ( 0 to 100 mM ) . Detailed procedures were done as described elsewhere [43] . In case of the L14-interaction epitope mapping experiment bait and prey plasmids were sequentially cotransformed into haploid yeast strain CG-1945 ( Clontech ) and then assayed as described above . ORFs were shuttled from entry plasmids into pNusA ( Santhera , Liestal , Switzerland ) , pETG-40A , or pETG-30A ( EMBL , Heidelberg , Germany ) and transformed or co-transformed into E . coli BL21 ( DE3 ) ( combinations , see main text , Figure 1D and 1E and Figure S3A ) . Proteins were expressed following standard protocols . Cell pellets were lysed in 500 µl buffer ( 50 mM Tris-HCL pH 8 . 0 , 100 mM NaCl , 50 µg/ml chicken egg white lysozyme , 50 µM PMSF , Sarcosyl/Triton-X 100 0 . 1% , each ) and then sonicated and centrifuged . The supernatants were used for pull-down experiments: for E . coli RsfA and L14 corresponding volumes of 50 µg soluble protein fractions of co-expressed proteins were applied to beads and aliquots saved as input controls . For human and Zea mays proteins 25 µg soluble fractions were mixed and then applied to the beads . MBP fusions were co-purified with their GST baits on 20 µl glutathione beads and NusA-tagged preys with their MBP fusions on 20 µl amylose beads under buffer conditions indicated above but w/o lysozyme . Binding occurred at room temperature for 30 min . Then , the beads were washed and finally boiled in 50 µl Laemmli buffer . 10 µl of output ( ∧ = 10 µg protein input ) and 10 µg input samples were separated by SDS PAGE using 12% gels . Proteins were transferred onto a polyvinylidene fluoride membrane by semi-dry Western blotting . The recombinant bait and prey proteins were labeled by standard immunodetection procedure and then analyzed by enhanced chemiluminescence . Human C7orf30 ( mtRsfA ) and L14mt full-length ORFs were cloned into pcDNA3 . 1-HA-mCherry [44] , pcDNA3 . 1 ( + ) -HA-VN , and pcDNA3 . 1 ( + ) -HA-VC [45] ( Note: an N-terminal HA tag from the vector backbones was removed under consideration that the native mitochondrial localization peptides of mtRsfA ( = C7orf30 ) and L14mt are N-terminally exposed ) . For localization studies , Hela cells were transfected ( 100 ng , each plasmid ) with mCherry-tagged C7orf30 or L14mt using Promofectin ( Promokine , Germany ) . 100 ng pECFP-Mem ( Clontech ) was co-transfected to stain cell membranes . 24 h later , MitoTracker Green FM ( 100 nM f . c . , Invitrogen ) was added . After washing , DRAQ5 ( 1∶2 , 000 , Biostatus ) was added fur nuclear staining . For BiFC assays [46] , Hela cells were prepared correspondingly . Exceptions: Mitotracker staining was not done and instead of localization constructs , cells were co-transfected with BiFC plasmid constructs ( 50 ng , each ) in combinations as given in Figure 1G . 30 min post DRAQ5 administration cells were analyzed by fluorescence microscopy using a Zeiss LSM 510 Meta confocal laser scanning microscope . Multiple alignments were generated using ClustalW [47] with the L14 amino acid sequences from E . coli , T . pallidum , S . pneumoniae , Synechocystis PCC 6803 , C . jejuni , H . sapiens , Zea mays , Chromobacterium violaceum , Bacillus halodurans , and S . cerevisiae using default parameters . Based on that alignment the conservation scores were calculated with the ConSurf Server [48] . 3D images ( Figure 2A ) were presented using PyMol 1 . 5 ( http://pymol . org ) . Structures of unbound proteins: the E . coli L14 structure was taken from 2AWB PDB entry , chain K [14] . Because the crystal structure of E . coli RsfA is not available , we used I-TASSER server [49] to build a model of that protein . The server built a single model using as templates 2ID1_A and 2O5A_A . The server has estimated the accuracy of the model as 0 . 90±0 . 06 ( TM-score ) and 1 . 6±1 . 4 Å ( RMSD ) . An unconstrained rigid body docking was performed of individual L14 and RsfA structures with GRAMM-X [50] . We then used the coordinates of L14 to superimpose 100 top scored docking models onto the entire 70S unit ( 2AWB and 2AW7 PDB IDs ) . Then , each model was evaluated for the backbone clashes between the predicted RsfA position and the rest of the 50S subunit . We defined a clash as having less than 2 Å distance between backbone atoms in order to tolerate some degree of unknown conformational re-arrangement of the 50S components that were not used in docking . Model #17 was the first one in order of the docking score where RsfA had no clashes with other parts of 50S ( parts not seen by the docking procedure ) . Model #17 contained certain surface exposed amino acid residues of L14 that are highly conserved ( Figure 2B ) . To test whether these are involved in mediating the interaction with RsfA they were subjected to alanine substitution constructs ( see above and Figure 2A ) and analyzed in Y2H experiments ( Figure 2C ) . The interface contacts were defined as having less than 4 . 6 Å distance between any heavy atoms of the docking subunits . We used PyMol 1 . 5 ( http://pymol . org ) for the post-docking analysis and graphics . ΔrsfA ( b0637 ) [16] and wild type ( BW25113 ) were transformed with a β-galactosidase reporter plasmid , pBAD24-lacZ-HA ( based on pBAD24HA ) [51] , [52] and selected on LB agar containing 50 µg/ml ampicillin . Both were grown overnight in LB in the presence of 50 µg/ml ampicillin and 0 . 4% glucose as inhibitor of leaky expression . For stationary phase expression cultures were centrifuged at 5 , 000 rpm ( 15 min ) and pellets were resuspended in the cell-free supernatant of an LB overnight culture ( BW25113/ΔrsfA , no plasmid ) lacking glucose . β-galactosidase expression was induced with 2% arabinose; the resuspension was adjusted to the same cell density as the previous stationary-phase culture . For logarithmic phase expression overnight cultures were centrifuged at 5 , 000 rpm for 15 min and pellets were resuspended in fresh LB medium ( no glucose ) with 50 µg/ml ampicillin for both strains . Cultures were then diluted to OD600 = 0 . 05 and grown for 2 h . β-galactosidase expression was induced by adding 2% arabinose to the medium . The cultures were shaken at 37°C . Every hour 300 µl suspension was withdrawn , 100 µl from it was loaded into a well of a 96-well plate ( flat bottom ) and the growth was followed by monitoring the extinction at 600 nm ( ELISA spectrophotometer ) . The rest of aliquots were centrifuged at 12 , 000 rpm for 5 min and pellets were resuspended in 20 µl loading buffer ( 2× ) Tris-glycine SDS and incubated at 95°C for 5 min to denature proteins . Samples were loaded on SDS-polyacrylamide gel ( 10% ) and the β-galactosidase amount was quantified as relative protein-band intensity using ImageJ 1 . 45 . For growth competition assays ( Figure 3A ) the same amount of cells from overnight cultures of wild type and ΔrsfA strains were mixed , yielding a final OD600 of 0 . 01 in a volume of 5 ml , and incubated with mild shaking either in LB ( rich ) or M9 medium with 0 . 4% glucose ( poor ) . Aliquots were withdrawn every 3 h or 6 h or 24 h ( depending on the growth rate ) and OD600 was measured . Simultaneously , dilutions to approximately 5 , 000 cells/ml ( according to the assumption that 1 OD600 corresponds roughly to 109 cells ) were made and 100 µl of each was plated in duplicates on either LB plates or LB plates containing 25 µg/ml kanamycin . The number of colonies ( ΔrsfA contained a kanR-cassette , WT not ) was counted after incubation at 37°C for overnight . For viability competition experiment in stationary phase ( LB medium; Figure 3F ) ΔrsfA mutant and wild type strain were separately grown overnight . Subsequently two cultures were diluted to OD600 = 0 . 005 and incubated with shaking till 0 . 5 OD600 . Then two cultures were mixed and the fitness of ΔrsfA was monitored as numbers of colonies on LB plates ( mutant and wild type colonies ) and LB plates containing kanamycin ( only mutant colonies ) after 2 , 6 , 9 , 21 , 32 , 52 , 78 hours of incubation at 37°C . The kanamycin resistance gene that substituted the rsfA was removed by introducing a flippase-encoding plasmid pCP20 as described elsewhere [53] . The successful flip-out was verified by a genotyping PCR . For the media shift ( Figure 3B ) wild type and ΔrsfA strains were grown overnight in LB medium ( rich ) and then diluted in either LB ( rich ) or M9 medium ( poor ) yielding a start OD600 = 0 . 005 . Cultures were incubated at 37°C with shaking ( 200 rpm ) and growth was monitored measuring the OD600 over a time of up to 40 hours . For curing the phenotype of the ΔrsfA strain during the transition from rich to poor ( Figure 3C and 3D ) ΔrsfA cells lacking the kanamycin resistance gene and wild type cells were transformed with a plasmid harbouring the gene coding for RsfA fused with a C-terminal His-tag under control of either the native promoter or the IPTG inducible tac-promoter and with the corresponding empty plasmid . The transformed strains were grown overnight in rich ( LB ) medium at 37°C and then diluted in poor M9 medium yielding a start OD600 = 0 . 005 and incubated like described above . At several time points samples were withdrawn and the expression of RsfA was analysed after SDS-PAGE and Western-blot using an antibody directed against the His-tag . The intensity of the RsfA-His bands was quantified using ImageQuant 5 . 2 and normalized for correction of the input to a non-altered protein band of the Coomassie stained gel . The gene coding for E . coli RsfA ( b0637 ) was expressed as an N-terminal His6 tag fusion in E . coli BL21 ( DE3 ) . Expression was induced at OD600 = 0 . 4 with 0 . 1 mM IPTG and carried out for 2 h at 30°C to decrease the formation of inclusion bodies . The soluble protein was purified via nickel-nitrilotriacetic-acid-agarose ( Qiagen , according to the manufacturer's manual ) and anion exchange chromatography ( Source 15Q , GE Healthcare ) . The purified protein was dialyzed against 20 mM Hepes , 6 mM Mg-acetate , 150 mM K-acetate , 4 mM β-mercaptoethanol , pH 7 . 6 at 0°C . The gene coding for the mature human mitochondrial RsfA ( C7orf30; amino acids 23–234 ) was expressed and the protein purified like the E . coli RsfA orthologue . Both proteins were expressed using the Gateway System-compatible plasmid pHGWA [54] . Ribosomes and ribosomal subunits were prepared from E . coli strains CAN20-12E [55] as described [56] . Preparation of mammalian mitochondrial ribosomes and ribosomal subunits ( pig liver ) followed [57] with minor modifications . Hepes-buffer and TCEP were utilized instead of Tris-buffer and 2-mercaptoethanol , respectively . Isolation of mitochondrial factors are described in [58] . 18 pmol 50S ribosomes were incubated with 180 µg poly ( U ) with or without 360 pmol RsfA in 90 µl for 10 min at 37°C in binding buffer ( 20 mM Hepes , pH 7 . 6 at 0° C , 4 . 5 mM Mg-acetate , 150 mM K-acetate , 4 mM β-mercaptoethanol , 2 mM spermidine , 0 . 05 mM spermine , H20M4 . 5K150SH4Spd2Spm0 . 05 ) . Reaction was further incubated with 10 pmol 30S ribosomes for 10 min at 37°C and then analyzed in poly ( U ) dependent oligo ( Phe ) synthesis and sucrose gradient centrifugation . 15 µl of the reaction was used for oligo ( Phe ) synthesis . 2 . 4 pmol EF-G together with the ternary complex mix were added yielding 30 µl in binding buffer H20M4 . 5K150SH4Spd2Spm0 . 05 . The ternary complex mix contained in 15 µl 30 pmol [14C]Phe-tRNAPhe , 45 pmol EF-Tu , 45 pmol EF-Ts , 3 mM GTP and was preincubated 5 min at 37°C . Incubation was at 30°C for 2 min and 12 . 5 µl aliquots were precipitated with TCA , incubated at 90°C in the presence of 2 drops of 1% ( w/v ) BSA and filtered through glass filters and counted . 60 µl of the reaction was mixed with 40 µl H20M4 . 5K150SH4Spd2Spm0 . 05 and loaded onto a 10–30% sucrose gradient prepared in the same buffer . Centrifugation was carried out at 42 , 000 rpm for 4 h in an SW60 rotor . The gradient was pumped out from bottom to top and the A260 was measured to obtain the ribosome profile . The corresponding assay with mitochondrial components from pig liver was performed in H20M4 . 5K150SH4Spd2Sp0 . 05 pH7 . 5 ( at 0°C ) . mtRsfA was pre-incubated with 2 . 5 pmol large subunit 39S in 80 molar excess over ribosomes , before the same amount of 28S subunits were added; likewise 2 . 5 pmol 55S ribosomes were incubated with the same amount of RsfA . EF-G1 was added in a 0 . 8-fold excess over ribosomes . 37 . 5 pmol of [14C]Phe-tRNA were present and the mitochondrial factors mtEF-Tu and mtEF-Ts , were added both in an excess of 1 . 5 over Phe-tRNA . The total volume was 100 µl , the main incubation 20 min at 30°C . The following processing was as described above . The oligo ( Phe ) synthesis with reassociated 70S ribosomes ( Figure 4A , right panel ) was performed in the following way: 3 pmol 70 S ribosomes were incubated with 30 µg poly ( U ) and 6 pmol Ac-Phe-tRNA for 10 min at 37°C . When indicated 60 pmol RsfA was added and the oligo ( Phe ) synthesis performed as described above . The total volume was 20 µl , the mixture was incubated for 5 min at 37°C .
|
The YbeB/DUF143 family of proteins is one of the most widely conserved proteins with homologues present in almost all bacteria and eukaryotic organelles such as mitochondria and chloroplasts ( but not archaea ) . While it has been shown that these proteins associate with ribosomes , their molecular function remained mysterious . Here we show that a knock-out of the ybeB gene causes a dramatic adaptation block during a shift from rich to poor media and seriously deteriorates the viability during stationary phase . YbeB of six different species binds to ribosomal protein L14 . This interaction blocks the association of the two ribosomal subunits and , as a consequence , translation . YbeB is thus renamed “RsfA” ( ribosomal silencing factor A ) . RsfA inhibits translation when nutrients are depleted ( or when cells are in stationary phase ) , which helps the cell to save energy and nutrients , a critical function for all cells that are regularly struggling with limited resources .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"bacteriology",
"cellular",
"structures",
"subcellular",
"organelles",
"functional",
"genomics",
"protein",
"interactions",
"macromolecular",
"assemblies",
"microbiology",
"bacterial",
"biochemistry",
"gene",
"function",
"protein",
"synthesis",
"protein",
"structure",
"molecular",
"genetics",
"microbial",
"physiology",
"proteins",
"biology",
"proteomics",
"biochemistry",
"cell",
"biology",
"gene",
"identification",
"and",
"analysis",
"genetics",
"genomics",
"molecular",
"cell",
"biology",
"computational",
"biology",
"genetics",
"and",
"genomics",
"macromolecular",
"structure",
"analysis"
] |
2012
|
RsfA (YbeB) Proteins Are Conserved Ribosomal Silencing Factors
|
Identifying the genetic determinants of phenotypes that impact disease severity is of fundamental importance for the design of new interventions against malaria . Here we present a rapid genome-wide approach capable of identifying multiple genetic drivers of medically relevant phenotypes within malaria parasites via a single experiment at single gene or allele resolution . In a proof of principle study , we found that a previously undescribed single nucleotide polymorphism in the binding domain of the erythrocyte binding like protein ( EBL ) conferred a dramatic change in red blood cell invasion in mutant rodent malaria parasites Plasmodium yoelii . In the same experiment , we implicated merozoite surface protein 1 ( MSP1 ) and other polymorphic proteins , as the major targets of strain-specific immunity . Using allelic replacement , we provide functional validation of the substitution in the EBL gene controlling the growth rate in the blood stages of the parasites .
Malaria parasite strains are genotypically polymorphic , leading to a diversity of phenotypic characteristics that impact on disease severity . Discovering the genetic basis for such phenotypic traits can inform the design of new drugs and vaccines . Both association mapping and linkage analyses approaches have been adopted to understand the genetic mechanisms behind various phenotypes of malaria parasites [1–5] and with the application of whole genome sequencing ( WGS ) , the resolution of these methodologies has been dramatically improved , allowing the discovery of selective sweeps as they arise in the field [6] . However , both approaches suffer from drawbacks when working with malaria parasites: linkage mapping requires the cloning of individual recombinant offspring , a process that is both laborious and time-consuming , and association studies require the collection of a large number of individual parasites ( usually in the thousands ) from diverse geographical origins and over periods of several months or years to produce enough resolution for the detection of selective sweeps . Linkage Group Selection ( LGS ) , like linkage mapping , relies on the generation of genetic crosses , but bypasses the need for extracting and phenotyping individual recombinant clones . Instead , it relies on quantitative molecular markers to measure allele frequencies in the recombinant progeny and identify loci under selection [7 , 8] . This approach bears similarity to Bulked Segregant Analysis ( BSA ) [9] , a technique developed to study disease resistance in plants . In BSA , individuals from a population are segregated based upon their phenotype ( e . g . disease resistance ) , following which the frequencies of genetic markers in each population are analysed , identifying loci at which different alleles are found for the differently phenotypes populations . Segregating individuals by phenotype , while relatively straight forward for large organisms such as plants , is not feasible for unicellular pathogens such as malaria parasites . Instead , in LGS , the segregating population is grown both in the presence or absence of a selection pressure ( e . g . drug treatment , immune pressure , etc . ) . Selection removes susceptible individuals in the selected “pool” , while leaving both susceptible and resistant individuals in the unselected “pool” . In its original implementation , LGS was successfully applied in studying strain-specific immunity ( SSI ) [10 , 11] , drug resistance [7 , 12] and growth rate [8] in malaria and SSI in Eimeria tenella [13] . LGS is essentially identical to the extreme QTL approach ( xQTL ) that was independently developed by yeast researchers based on BSA [14] . In both the original implementations of BSA and LGS a limiting factor is the availability of molecular markers differentiating the two populations . One step in increasing the number of molecular markers was through the use of array hybridisation that allowed the identification of thousands of SNPs as molecular markers in Arabidopsis thaliana [15] . BSA ( still using pre-selected pools ) was also combined with tiling microarray hybridisation and used probe intensities to detect a gene underlying xylose utilisation in yeast [16] . The xQTL method increased the power and rapidity of the approach by making use of available yeast microarray data as well as Next Generation Sequencing ( NGS ) of DNA hybridised to microarray probes to identify a large number of markers across the genome , this time comparing selected and unselected populations , rather then generating pools based on phenotype [14 , 17] . In the absence of microarray databases , an alternative approach was to use NGS short reads to identify genome-wide SNPs between two parents and then use these SNPs as molecular markers to identify target genes in the selected progeny population compared against the unselected population , as done to study chloroquine resistance in malaria [18] . In this study , we apply an improved LGS approach for the identification of genes controlling two independent and naturally occurring phenotypic differences between two strains of the rodent malaria parasite Plasmodium yoelii; growth rate , and strain-specific immunity . A mathematical model , built upon methodological improvements in the analysis of genetically crossed populations [19 , 20] , was developed to analyze the data . This modified LGS approach relies on the generation and selection of at least two independent crosses between the strains . The progeny from both crosses pre- and post-selection are then subjected to high-throughput WGS , and SNP marker movement analyzed using best fitting modeling ( Fig 1 ) . Our novel statistical framework both accounts for the influence of clonal growth in the cross population , and allows for a locally variable recombination rate in the parasite population , unlike previous analyses applied to comparable data [21] . Applying this framework to crosses between two strains of P . yoelii that induce SSI , and which differ in their growth rates , we were able to identify three genomic regions and alleles controlling both phenotypes , demonstrating that the approach can be used to analyze multiple complex phenotypes concomitantly with high genomic resolution within a short space of time .
The difference in blood-stage parasite growth rate between the two clones was followed in vivo for nine days in CBA mice . A likelihood ratio test using general linear mixed models indicated a more pronounced growth rate for 17X1 . 1pp compared to CU clone by time interaction term , L = 88 . 60 , df = 21 , p<0 . 0001 , Fig 2A ) . To verify that the two malaria clones could also be used to generate protective SSI , groups of mice were immunized with 17X1 . 1pp , CU or mock immunized , prior to challenge with a mixture of the two clones ( S1 Fig ) . The relative proportions of the two clones were measured on day four of the infection by real time quantitative PCR ( Q-RT-PCR ) targeting the polymorphic msp1 locus [22] . A strong , statistically significant SSI was induced by both parasite strains in CBA mice ( Fig 2B ) . Two kinds of selection pressure were applied in this study: growth rate driven selection and SSI . Two independent genetic crosses between 17X1 . 1pp and CU were produced , and both these crosses were subjected to immune selection ( in which the progeny were grown in mice made immune to either of the two parental clones ) , and grown in non-immune mice . Progeny were harvested from mice four days after challenge , at which point strain-specific immune selection in the immunized mice , and selection of faster growing parasites in the non-immune mice had occurred . Using deep sequencing by Illumina technology , a total of 29 , 053 high confidence genome-wide SNPs that distinguish the parental strains were produced by read mapping with custom-made Python scripts . SNP frequencies from these loci from each population were filtered using a likelihood ratio test to remove sites where alleles had been erroneously mapped to the wrong genome location . A hidden Markov model was applied to the data to identify allele frequency changes ( Table 1 ) that were likely to have arisen from the clonal growth of individuals within the cross population or possible incorrect assembly of the reference genome , as described in the Materials and Methods section and in more detail in the supplementary mathematical methods ( S1 Appendix ) . In a genetic cross population , an especially high fitness clone generated by random recombination events can grow to substantial frequency , this being manifested as sudden jumps in allele frequency occurring at the recombination points in this individual [23] . Jumps of this type were primarily identified in the 17X-immunized population , where the increased virulence of the 17X strain had less of an effect in driving alleles to high frequency , and in the first replica experiment; the data in the first experiment seemed to have been more affected by clonal growth in the population . The consistency of identified jumps between treatment conditions reflects the common origin of the differently treated populations; the jump at the end of chromosome XIV inferred in both replicas may be artefactual . Based upon an analytical evolutionary model describing patterns of allele frequencies following selection , a maximum likelihood approach was used to define confidence intervals for the positions of alleles under selection in each of the genetic cross populations . In the absence of selection acting for a variant in a region of the genome , the allele frequencies in that region are expected to be locally constant . In common with a previous approach to identifying selected alleles [21] , a search was therefore made for regions of the genome in which allele frequencies varied substantially according to their position in the genome . Next , wherever deviations of this form were consistently identified in both replica experiments a model of selection was applied to the data , inferring for each set of replica data the position in that region of the genome that was most likely to be under selection; this model was based upon expected changes in allele frequency under a constant local rate of recombination and is described further in the Methods section . Regions of the genome in which this inference of selection produced consistent results across replica datasets were then identified ( Table 2 ) . Of a total of 11 genomic regions suggesting evidence of non-neutrality , six showed sufficient evidence of consistent selection . For each of these regions of the genome , a more sophisticated evolutionary model , accounting for variation in the local recombination rate , was then applied to the data , refining the position of the putatively selected allele . At this point , a putative selected allele in chromosome IV was removed from consideration , leaving five cases of potential alleles under selection in three regions of the genomes; confidence intervals for the positions of the selected loci are given in Table 3 . Optimal positions of variant loci derived from each replicate are detailed in S1 Table; results of the variable recombination rate model are shown in S2 Table , with inferred recombination rates in S3 Table . Of the final three putative loci , two were detected under multiple experimental conditions ( Fig 3 ) . When considering the combined largest intervals , a selective sweep was inferred at position 1 , 436–1 , 529 kb on Chromosome ( Chr ) XIII in replicate crosses grown in both non-immunized mice and 17X1 . 1pp-immunized mice , resulting from selection against CU-specific alleles at the target locus . A second sweep was inferred at position 1 , 229–1 , 364 kb on Chr VIII , detected in the parasite crosses grown in both CU and 17X1 . 1pp immunized mice , though not in the non-immunized mice . Here , selection pressure acted against different alleles according to the strain against which mice were immunized . The third sweep was detected at a locus between positions 725–814 kb on Chr VII . This event was only detected in mice replicates immunized with the 17X1 . 1pp strain , albeit that a consistent change in allele frequencies was also observed between replicas grown under these conditions ( Fig 3B ) . The remaining loci ( on Chrs VIII and XIII ) were not consistently detected between replicates ( S1 Table ) and were thus considered to be non-significant . All the genes in the combined conservative intervals of the three main loci under selection are listed in S4–S6 Tables , along with annotation pertaining to function , structure , orthology with P . falciparum genes and Non-synonymous/Synonomous SNP ( NS/S ) ratio in the P . falciparum orthologue , which is calculated by the PlasmoDB website ( 6 . 2 ) based on SNP data from 202 individual strains . These include both laboratory strains and field isolates obtained from six collections ( see Methods for more details ) . The locus associated with SSI on Chr VIII contains 41 genes . We considered the presence of either transmembrane ( TM ) domains or a signal peptides as necessary features of potential antigen-encoding genes . Only 16 genes met these criteria . Functional annotation indicated 10 likely candidates among these; eight genes described as “conserved Plasmodium proteins” , and two encoding RhopH2 and merozoite surface protein 1 ( MSP1 ) . Of these genes , the P . falciparum orthologue of msp1 had the highest NS/S SNP ratio ( 8 . 43 ) . MSP1 is a well characterized major antigen of malaria parasites that has formed the basis of several vaccine studies [24] and has been previously linked to SSI in Plasmodium chabaudi [10–12] . The locus under selection on Chr VII consists of 21 genes . Only seven contained TM domains and/or a signal peptide motif . Based on functional annotation , four of these could be potential targets for SSI . One of these genes , PY17X_0721800 , encodes an apical membrane protein orthologous to Pf34 in P . falciparum . This protein has recently been described as a surface antigen that can elicit an immune response [25] . Three conserved proteins of unknown function ( PY17X_0720100 , PY17X_0721500 and PY17X_0721600 ) also displayed potential signatures as target antigens . The growth rate associated selected locus on Chr XIII contains 29 genes . In this case , the presence of TM domains or signal peptide motifs were not considered informative criteria . Only eight genes contained NS SNPs between the parental strains 17X1 . 1pp and CU according to the WGS data . Among these was a duffy binding protein , Pyebl . Pyebl , is a gene that has been previously implicated in growth rate differences between strains of P . yoelii [8 , 26] . A single NS SNP was predicted from the WGS data in this gene . Due to the very high likelihood of its involvement based on previous work , this gene was considered for further analysis . Examining the Pyebl gene , Sanger capillary sequencing re-confirmed the existence in 17X1 . 1pp of an amino acid substitution ( Cys >Tyr ) at position 351 within region 2 of the encoded protein . When aligned against other P . yoelii strains and other Plasmodium species , this cysteine residue is highly conserved , and the substitution observed in 17X1 . 1pp was novel ( Fig 4A ) . Crucially , no other polymorphisms were detected in the coding sequence of the gene , including in region 6 , the location of the SNP previously implicated in parasite virulence in other strains of P . yoelii [8] . Structural modeling of the EBL protein in both wild-type and 17x1 . 1pp ( C351Y ) mutants predicted the abolition of a a disulphide bond between C351 and C420 in the mutant parasites that alters the tertiary structure of the receptor binding region of the ligand in these parasites ( Fig 4B and 4C ) . The functional role of this polymorphism was verified by experimental means . In order to study the functional consequences of the polymorphism , the Pyebl alleles of slow growing CU and faster growing 17X1 . 1pp clones were replaced with the alternative allele ( i . e . CU-EBL-351C>Y and 7x1 . 1pp-EBL-351Y>C ) , as well as with the homologous allele ( i . e . CU-EBL-351C>C and 17x1 . 1pp-EBL-351Y>Y ) . The latter served as a control for the actual allelic swap , as the insertion of the plasmid for allelic substitution could potentially affect parasite fitness independently of the allele being inserted . To establish whether the C351Y substitution affected EBL localization , as was shown for the previously described region 6 mutation , Immunoflurescence Analysis ( IFA ) was performed . This revealed that , unlike the known mutation in region 6 [8] , the EBL proteins of 17X1 . 1pp and CU were both found to be located in the micronemes ( Fig 5 and S2 Fig ) . Transgenic clones were grown in mice for 10 days alongside wild-type clones . Pair-wise comparisons between transgenic clones with the parental allele against transgenic clones with the alternative allele ( that is CU-EBL-351C>C vs CU-EBL-351C>Y and 17x1 . 1pp-EBL-351Y>Y vs 17x1 . 1pp-EBL-351Y>C ) showed that allele substitution could switch growth phenotypes in both strains ( Fig 6A and 6B ) . This confirmed the role of the C351Y mutation as underlying the observed growth rate difference . RNA-seq analysis revealed that transfected EBL gene alleles were expressed normally , ( S3 Fig ) , thus indicating a structural effect of the polymorphism on parasite fitness , rather than an alteration in protein expression .
The development of LGS has facilitated functional genomic analysis of malaria parasites over the past decade . In particular , it has simplified and accelerated the detection of loci underlying selectable phenotypes such as drug resistance , SSI and growth rate [7 , 8 , 10] . Here we present a radically modified LGS approach that utilizes deep , quantitative WGS of parasite progenies and the respective parental populations , multiple crossing and mathematical modeling to identify loci under selection at ultra-high resolution . This enables the accurate definition of loci under selection and the identification of multiple genes driving selectable phenotypes within a very short space of time . This modified approach allows the simultaneous detection of genes or alleles underlying multiple phenotypes , including those with a multigenic basis . Applying this modified LGS approach to study SSI and growth rate in P . yoelii , we identified three loci under selection that contained three strong candidate genes controlling both phenotypes . Two loci were implicated in SSI; the first time LGS has identified multigenic drivers of phenotypic differences in malaria parasites in a single experimental set-up . The strong locus under selection in Chr VIII , associated with the gene encoding MSP1 , is consistent with existing knowledge of malaria immunity . The Chr VII locus , which includes the orthologue of Pf34 as well as other potential unannotated antigens , underscores the power for hypothesis generation and gene detection of the LGS approach using multiple crosses . Our approach also provided a genetic rationale for the difference in growth rate of the parental clones CU and 17X1 . 1pp . Phenotypically , this occurs due to the ability of 17X1 . 1pp to invade both reticulocytes and normocytes , while CU is restricted to reticulocytes [22] . Previously , differences in growth rates between strains of P . yoelii have been linked to a polymorphism in Region 6 of the Pyebl gene that alters its trafficking so that the protein locates in the dense granules rather than the micronemes [8 , 26] . In the case of 17x1 . 1pp however , direct sequencing of the Pyebl gene revealed a previously unknown SNP in region 2 , the predicted receptor-binding region of the protein , with no polymorphism in region 6 . Consistent with this , the EBL protein of 17X1 . 1pp was shown to be located in the micronemes , indicating that protein trafficking was unaffected by the region 2 substitution . Allelic replacement of the parasite strains with the alternative allele resulted in a switching of the growth rate to that of the other clone , thus confirming the role of the substitution . Region 2 of the Pyebl orthologues of P . falciparum and Plasmodium vivax [27–29] are known to interact with receptors on the red blood cell ( RBC ) surface . Furthermore , the substitution falls within the central portion of the region , which has been previously described as being the principal site of receptor recognition in P . vivax [29] . Wild-type strains of P . yoelii ( such as CU ) preferentially invade reticulocytes but not mature RBCs , whereas highly virulent strains are known to invade a broader repertoire of RBCs [30] . Further structural and functional studies are required to elucidate how the polymorphism described here enables mutant parasites to invade a larger repertoire of erythrocytes than wild type parasites . We show that the cysteine residue at position 351 in EBL forms a disulphide bond with a cysteine at position 420 , and that this is abolished following the C351Y substitution , altering the tertiary structure of the binding region . This leads to the possibility that such an alteration of the shape of the binding domain may enable the ligand to bind to a larger repertoire of receptors . LGS with multiple crosses offers a powerful and rapid methodology for identifying genes or non-coding regions controlling important phenotypes in malaria parasites and , potentially , in other apicomplexan parasites . Through bypassing the need to clone and type hundreds of individual progeny , and by harnessing the power of genetics , genomics and mathematical modeling , genes can be linked to phenotypes with high precision in a matter of a few months , rather than years . Here we have demonstrated the ability of LGS to identify multiple genetic polymorphisms underlying two independent phenotypic differences between a pair of malaria parasite strains; growth rate and SSI . This methodology has the potential power to identify the genetic components controlling a broad range of selectable phenotypes , and can be applied to studies of drug resistance , transmissibility , virulence , host preference , etc . , in a range of apicomplexan parasites that are amenable to genetic crossing . The applicability of the approach to human malaria species has been recently demonstrated: the original LGS approach was successfully applied to study P . falciparum immune evasion in mosquitoes in vivo [31] , while we recently tested its applicability in vitro to detect loci under selection following antifolate drug treatment and in vitro growth rate competition . With the advent of humanized mice that are able to support the complete malaria life cycle , the generation of new genetic crosses between strains of human malaria has become more feasible , as recently demonstrated [32] . With the ability to maintain these crosses without the need of simian hosts , application of a broader range of selection pressures ( excluding , for now , selection mediated by the presence of a complete immune response ) is now more feasible in vivo , thus extending the application of the LGS approach to medically relevant malaria species .
Laboratory animal experimentation was performed in strict accordance with the Japanese Humane Treatment and Management of Animals Law ( Law No . 105 dated 19 October 1973 modified on 2 June 2006 ) , and the Regulation on Animal Experimentation at Nagasaki University , Japan . The protocol was approved by the Institutional Animal Research Committee of Nagasaki University ( permit: 1207261005–2 ) . Plasmodium yoelii CU ( with slow growth rate phenotype ) and 17X1 . 1pp ( with intermediate growth rate phenotype ) strains [33] were maintained in CBA mice ( SLC Inc . , Shizuoka , Japan ) housed at 23°C and fed on maintenance diet with 0 . 05% para-aminobenzoic acid ( PABA ) -supplemented water to assist with parasite growth . Anopheles stephensi mosquitoes were housed in a temperature and humidity controlled insectary at 24°C and 70% humidity , adult flies were maintained on 10% glucose solution supplemented with 0 . 05% PABA . Plasmodium yoelii parasite strains were typed for growth rate in groups of mice following the intravenous inoculation of 1 × 106 iRBCs of either CU , 17X1 . 1pp or transfected clones per mouse and measuring parasitaemia over 8–9 days . In order to verify the existence of SSI between the CU and 17X1 . 1pp strains , groups of five mice were inoculated intravenously with 1 × 106 iRBCs of either CU or 17X1 . 1pp parasite strains . After four days , mice were treated with mefloquine ( 20mg/kg/per day , orally ) for four days to remove infections . Three weeks post immunization , mice were then challenged intravenously with 1 × 106 iRBCs of a mixed infection of 17X1 . 1pp and CU parasites . A group of five naïve control mice was simultaneously infected with the same material . After four days of growth 10 μl of blood were sampled from each mouse and DNA extracted . Strain proportions were then measured by Quantitative Real Time PCR using primers designed to amplify the msp1 gene [34] . All measurements were plotted and standard errors calculated using the Graphpad Prism software ( v6 . 01 ) ( http://www . graphpad . com/scientific-software/prism/ ) . Wilcoxon rank sum tests with continuity corrections were used to measure the SSI effect , and were performed in R [35] . Linear mixed model analyses and likelihood ratio tests to test parasite strain differences in growth rate were performed on log-transformed parasitaemia by choosing parasitaemia and strain as fixed factors and mouse nested in strain as a random factor , as described previously [22] . Pair-wise comparisons of samples for the transfection experiments were performed using multiple 2-way ANOVA tests and corrected with a Tukey’s post-test in Graphpad Prism software ( v6 . 01 ) . SNP frequencies were processed to filter potential misalignment events . We note that , during the cross , a set of individual recombinant genomes are generated . Considering the individual genome g , we define the function ag ( i ) as being equal to 1 if the genome has the CU allele at locus i , and equal to 0 if the genome has the 17X1 . 1pp allele at this locus . In any subsequent population of N individuals , the allele frequency q ( i ) at locus i can then be expressed as q ( i ) = 1 N ∑ g n g a g ( i ) ( 1 ) for some set of values ng , where ng is the number of copies of genome g in the population . To filter the allele frequencies , we note that each function ag ( i ) changes only at recombination points in the genome g . As such , q ( i ) should change relatively smoothly with respect to i . Using an adapted version of code developed for the inference of subclones in populations [39] , we therefore modeled the reported frequencies q ( i ) as being ( beta-binomially distributed ) emissions from an underlying diffusion process ( denoted by x ( i ) ) along each chromosome , plus uniformly distributed errors , using a hidden Markov model to infer the variance of the diffusion process , the emission parameters , and an error rate . A likelihood ratio test was then applied to identify reported frequencies that were inconsistent with having been emitted from the inferred frequency x ( i ) at locus i relative to having been emitted from an inferred global frequency distribution fitted using the Mathematica package via Gaussian kernel estimation to the complete set of values {x ( i ) }; this test filters out reported frequencies potentially arising from elsewhere in the genome . Next , the above logic was extended to filter out clonal growth . In the event that a specific genome g is highly beneficial , this genome may grow rapidly in the population , such that ng becomes large . Under such circumstances the allele frequency q ( i ) gains a step-like quality , mirroring the pattern of ag ( i ) . Such steps may potentially mimic selection valleys , confounding any analysis . As such , a jump-diffusion variant of the above hidden Markov model was applied , in which the allele frequency can change either through a diffusion process or via sudden jumps in allele frequency , modeled as random emissions from a uniform distribution on the interval [0 , 1] . For each interval ( i , i + 1 ) the probability that a jump in allele frequency had occurred was estimated . Where potential jumps were identified , the allele frequency data were split , such that analyses of the allele frequencies did not span sets of alleles containing such jumps . The resulting segments of genome were then analyzed under the assumption that they were free of allele frequency change due to clonal behavior . Inference of the presence of selected alleles was performed using a series of methods . In the absence of selection in a chromosome , the allele frequency is likely to remain relatively constant across each chromosome . A ‘non-neutrality’ likelihood ratio test was applied to each contiguous section of genome , calculating the likelihood difference between a model of constant frequency x ( i ) and the variable frequency function x ( i ) inferred using the jump-diffusion model . Next , an inference was made of the position of the allele potentially under selection in each region . Under the assumptions that selection acts for an allele at locus i , and that the rate of recombination is constant within a region of the genome , previous work on the evolution of cross populations [19 , 20] can be extended to show that the allele frequencies within that region of the genome at the time of sequencing are given by x ( i ) = x + Δ x ( 2 ) x ( j ) = [ X + 1 2 ( 1 - X ) ( 1 + e - ρ Δ i j ] x + [ 1 2 X ( 1 - e - ρ Δ i j ) ] ( 1 - x ) + Δ x ( 3 ) for each locus j not equal to i , where X is the CU allele frequency at the time of the cross , ρ is the local recombination rate , Δij is the distance between the loci i and j , x is an allele frequency , and Δx describes the effect of selection acting upon alleles in other regions of the genome . A likelihood-based inference was used to identify the locus at which selection was most likely to act . In regions for which the ‘non-neutrality’ test produced a positive result for data from both replica crosses , and for which both the inferred locus under selection , and the direction of selection acting at that locus were consistent between replicas , an inference of selection was made . For regions in which an inference of selection was made , an extended version of the above model was applied , in which the assumption of locally constant recombination rate was relaxed . Successive models , including an increasing number of step-wise changes in the recombination rate , were applied , using the Bayesian Information Criterion [40] for model selection . A model of selection at two loci within a region of the genome was also examined . Given an inference of selection , a likelihood-based model was used to derive confidence intervals for the position of the locus under selection . For each combined conservative interval of relevant loci under selection , genes were listed based on the annotation available in version 6 . 2 of PlasmoDB and verified against the current annotation ( release 26 ) . For each gene , information on predicted transmembrane domains , signal peptides and P . falciparum orthologues . For the P . falciparum orthologues , the NS/S SNP ratios were obtained from PlasmoDB , based on the count of synonymous and non-synonymous SNPs found in 202 individual strains collected from 6 data sets stored on the website . More details on the data sets can be found at the following link: https://goo . gl/lUwKn1 . All primer sequences are given in Supplementary S7 Table . Plasmids were constructed using MultiSite Gateway cloning system ( Invitrogen ) . To assess the course of infection of wild type and transgenic parasite lines , 1 × 106 pRBCs were injected intravenously into five 8-week old female CBA mice for each parasite line . Since the 17X1 . 1p and CU-recipient strains were transfected on separate occasions , the transgenic lines were tested separately . Thin blood smears were made daily , stained with Giemsa’s solution , and parasitaemias were examined microscopically . Since the atomic structures of EBL protein of P . yoelii Wild Type: ( Py17X-WT ) and its mutant P . yoelii ( C351Y ) : ( Py17X1 . 1pp ) are not known , homology models were generated . The homology models were generated using P . vivax Duffy Binding Protein ( PvDBP ) atomic structure ( PDB ID: 3RRC , [46] with the Swiss-Model server ( https://swissmodel . expasy . org ) [47–50] . The homology models showed maximum amino acid sequence homology of 32% with Py17X-WT EBL , compared to another homologous protein P . falciparum Erythrocyte Binding Antigen 140 ( PfEBA-140/BAEBL ) ( PDB ID: 4GF2 , [51] , that had 26% sequence homology . These models were then subsequently stabilized by minimizing their energies for at least 10 times each , to attain reasonably well equilibrated structures using the YASARA server ( www . yasara . org ) . The prediction of disulfide bonds in our homology models were performed using DISULFIND ( http://disulfind . dsi . unifi . it ) [52–55] . Our analysis showed high probability of disulfide bond formation by this Cys351 residue . Confirming that C351 is a potential residue for forming a disulfide bond , the energy minimized stable homology models were subjected to Disulfide bond visualization to check whether the Cys351 is involved in any disulfide bond formation with any other Cys and what is the effect of the C351Y substitution . The homology models along with their disulfide bonds were visualized ( Fig 4B and 4C ) and the images were obtained using the “Disulfide by Design 2 . 0” server ( http://cptweb . cpt . wayne . edu ) [56] . Code used in this project is available online from https://github . com/cjri/LGSmalaria
|
Developing a greater understanding of malaria genetics is a key step in combating the threat posed by the disease . Here we use a novel approach to study two important properties of the parasite; the rate at which parasites grow within a single host , and the means by which parasites are affected by the host immune system . Two malaria strains with different biological properties were crossed in mosquitoes to produce a hybrid population , which was then grown in naïve and vaccinated mice . Parasites with genes conveying increased growth or immune evasion are favoured under natural selection , leaving a signature on the genetic composition of the cross population . We describe a novel mathematical approach to interpret this signature , identifying selected genes within the parasite population . We discover new genetic variants conveying increased within-host growth and resistance to host immunity in a mouse malaria strain . Experimental validation highlights the ability of this rapid experimental process for generating insights into malaria biology .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"plasmodium",
"yoelii",
"tropical",
"diseases",
"cloning",
"parasitic",
"diseases",
"alleles",
"parasitic",
"protozoans",
"protozoans",
"molecular",
"biology",
"techniques",
"mammalian",
"genomics",
"research",
"and",
"analysis",
"methods",
"malarial",
"parasites",
"molecular",
"biology",
"genetic",
"loci",
"animal",
"genomics",
"genetics",
"biology",
"and",
"life",
"sciences",
"malaria",
"genomics",
"organisms"
] |
2017
|
Rapid identification of genes controlling virulence and immunity in malaria parasites
|
Since 2001 , outbreaks of Nipah virus have occurred almost every year in Bangladesh with high case-fatality rates . Epidemiological data suggest that in Bangladesh , Nipah virus is transmitted from the natural reservoir , fruit bats , to humans via consumption of date palm sap contaminated by bats , with subsequent human-to-human transmission . To experimentally investigate this epidemiological association between drinking of date palm sap and human cases of Nipah virus infection , we determined the viability of Nipah virus ( strain Bangladesh/200401066 ) in artificial palm sap . At 22°C virus titers remained stable for at least 7 days , thus potentially allowing food-borne transmission . Next , we modeled food-borne Nipah virus infection by supplying Syrian hamsters with artificial palm sap containing Nipah virus . Drinking of 5×108 TCID50 of Nipah virus resulted in neurological disease in 5 out of 8 hamsters , indicating that food-borne transmission of Nipah virus can indeed occur . In comparison , intranasal ( i . n . ) inoculation with the same dose of Nipah virus resulted in lethal respiratory disease in all animals . In animals infected with Nipah virus via drinking , virus was detected in respiratory tissues rather than in the intestinal tract . Using fluorescently labeled Nipah virus particles , we showed that during drinking , a substantial amount of virus is deposited in the lungs , explaining the replication of Nipah virus in the respiratory tract of these hamsters . Besides the ability of Nipah virus to infect hamsters via the drinking route , Syrian hamsters infected via that route transmitted the virus through direct contact with naïve hamsters in 2 out of 24 transmission pairs . Although these findings do not directly prove that date palm sap contaminated with Nipah virus by bats is the origin of Nipah virus outbreaks in Bangladesh , they provide the first experimental support for this hypothesis . Understanding the Nipah virus transmission cycle is essential for preventing and mitigating future outbreaks .
Nipah virus first emerged in 1998 during a large outbreak of encephalitis and respiratory disease in Malaysia and Singapore , causing 276 cases of encephalitis with 106 fatalities [1] . Since 2001 , outbreaks of Nipah virus have occurred almost every year in Bangladesh with a strikingly high case-fatality rate of up to 90% [2] , with 24 cases of Nipah virus occurring to date in 2013 [3] . The recurrent outbreaks of Nipah virus in Bangladesh have been epidemiologically associated with the consumption of date palm sap , which has led to the hypothesis that Nipah virus zoonosis is a result of drinking date palm sap contaminated by infected fruit bats [4] , [5] . In Bangladesh , date palm sap is harvested at nighttime from October to March [6] , which overlaps with the occurrence of Nipah virus outbreaks . Although to our knowledge Nipah virus has so far not been detected in date palm sap , human observation and analysis by infrared camera has shown that bats frequently drink from date palm sap during collection [7] , [8] . Since bats can shed Nipah virus in their urine and saliva [9]–[12] , it is thought that bats contaminate the date palm sap while drinking from the sap stream or date palm sap collection vessel . In addition to the initial zoonotic transmission , subsequent human-to-human transmission also plays an important role in the epidemiology of Nipah virus outbreaks in Bangladesh . It was estimated for the outbreaks in Bangladesh between 2001 and 2007 that approximately 50% of cases was the result of human-to-human transmission [13] . Thus far , the epidemiological association between drinking date palm sap and Nipah virus infection has not been confirmed experimentally . Therefore , we set out to assess the ability of date palm sap to function as a vehicle for zoonotic transmission of Nipah virus using a well-established small animal model for Nipah virus pathogenesis and transmission , the Syrian hamster [14]–[16] . We showed that , upon drinking of artificial palm sap containing high doses of Nipah virus , hamsters became infected and developed neurological signs of disease . Moreover , hamsters infected through the drinking route transmitted Nipah virus to naïve hamsters via direct contact .
The composition of palm sap was derived from a published report [17] and artificial palm sap was produced in the laboratory consisting of 13% sucrose and 0 . 21% BSA in water , pH 7 . 0 . The stability of three different doses ( 103 , 105 and 107 TCID50/ml ) of Nipah virus ( strain Bangladesh/200401066 ) in artificial palm sap was determined at 22°C . Of note , between October and March ( the date palm sap harvesting season ) the average temperature in Bangladesh fluctuates between 20°C and 28°C . No significant difference was detected between the slopes of the three lines ( slope ± standard error: 0 . 001140±0 . 001026 for the 103 TCID50 , 0 . 002115±0 . 0008557 for the 105 TCID50 and 0 . 0007124±0 . 001272 for the 107 TCID50 ) . In addition , the slopes of the three groups were also not significantly different from zero ( p = 0 . 2812 for the 103 TCID50 , p = 0 . 0950 for the 105 TCID50 , p = 0 . 5823 for the 107 TCID50 ) . Thus , no significant reduction in Nipah virus infectious titer was observed at any of the dilutions for at least 8 days at 22°C ( Fig . 1 ) . At 28°C , Nipah virus titers decreased only 5–10 fold during the eight day incubation period ( Fig . S1A ) , thus indicating preservation of Nipah virus viability in date palm sap at both temperatures . Since heating of date palm sap before consumption has been suggested as a means of inactivating Nipah virus and thereby preventing Nipah virus infection , we tested the stability of Nipah virus in artificial palm sap at 70°C and 100°C . At both temperatures , virus titers decreased about 4 log in the first 5 minutes ( Fig . S1B ) . However , incubation of Nipah virus in artificial palm sap at 70°C for 1 hour did not inactivate all infectious virus . In contrast , incubation at 100°C for more than 15 minutes completely inactivated Nipah virus . To assess the ability of Nipah virus to establish an infection upon ingestion of virus , hamsters were inoculated esophageally with 107 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) . Eight hamsters were monitored daily for signs of disease for up to 28 days post inoculation ( dpi ) . One hamster succumbed to respiratory disease on 6 dpi; a second hamster was euthanized on 12 dpi ( Fig . 2A ) . Two out of the six surviving hamsters seroconverted by 28 dpi , indicating that 4 out of 8 hamsters were likely infected with Nipah virus after esophageal inoculation . For comparison , a group of hamsters was inoculated intranasally with 107 TCID50 of Nipah virus . All eight intranasally inoculated animals were euthanized due to severe disease signs between 5 and 14 dpi ( Fig . 2A ) . On 2 , 4 , and 8 dpi , groups of 4 hamsters inoculated esophageally or intranasally were euthanized and 17 tissues were collected from each hamster for virus titration . In addition , whole blood was collected for analysis of the presence of viral RNA by qRT-PCR . For both inoculation routes , virus was mainly detected in nasal turbinates , trachea and lungs; virus in non-respiratory tissues was observed mostly in animals with evidence of viremia . In the esophageally inoculated hamsters , virus was detected on 2 dpi in 2 out of 4 hamsters ( Fig . 3 & Table S1 ) , with viremia detected by qRT-PCR in one of the two remaining hamsters . On 4 dpi , virus was detected in only one hamster , including a low amount of virus in the esophagus of this animal . By 8 dpi , virus could no longer be detected in tissues or blood . On 2 and 4 dpi , virus was detected in respiratory tissues of all intranasally inoculated animals; on 8 dpi , virus could not be detected in the tissues of all three remaining animals , although viremia was detected by RT-PCR in one animal at this time-point ( Fig . 3 and Table S1 ) . Histopathological examination did not reveal evidence of virus replication in intestinal tissues in hamsters that were inoculated esophageally ( Table S2 ) . Next , the ability of Nipah virus to infect Syrian hamsters via drinking of virus-containing artificial palm sap was assessed . Eight animals were singly housed and their drinking water was replaced with 30 ml artificial palm sap containing 107 TCID50 of Nipah virus; animals drank the artificial palm sap containing Nipah virus in approximately 2 days . Animals were assessed for signs of disease and survival for up to 28 days . On 10 and 11 days after supplying the hamsters with artificial palm sap containing Nipah virus , one hamster was euthanized due to neurological signs of disease ( Fig . 2B ) . The remaining six hamsters did not show signs of disease until the end of the experiment at day 28; however , two out of these six hamsters seroconverted , indicating that these animals were likely infected with Nipah virus . Tissues from 4 animals collected on 2 and 4 days after supplying the hamsters with artificial palm sap containing Nipah virus were negative in virus titration and viremia could not be detected by qRT-PCR ( Table S1 ) . On day 8 , a low amount of virus was detected in the kidney of 1 animal; all other tissues of this hamster and all tissues of three other hamsters were negative ( Table S1 ) . Since human-to-human transmission plays an important role in Nipah virus outbreaks in Bangladesh and virus shedding is a prerequisite for transmission , we collected nasal , oropharyngeal , urogenital and rectal swabs daily for nine days after supplying the hamsters with artificial palm sap containing Nipah virus and analyzed these for the presence of Nipah virus RNA . Nipah virus RNA could be detected in only six of the collected swabs; all PCR-positive swabs were oropharyngeal swabs . One of the seroconverted hamsters had a positive swab on 4 and 6 days after supplying it with artificial palm sap; one of the hamsters that was euthanized with neurological signs had positive oropharyngeal swabs on days 5 , 6 and 7 after inoculation ( data not shown ) , likely indicating an active infection . In comparison , all intranasally inoculated hamsters shed virus from the nose for up to 7 days and from the throat up to 11 dpi ( Fig . 4 ) . To determine whether virus shedding increased when animals were supplied with a higher dose of Nipah virus in artificial palm sap , we repeated the drinking experiment with 5×108 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) in artificial palm sap . Nasal , oropharyngeal , urogenital and rectal swabs were collected up to 11 days after supplying the hamsters with artificial palm sap . All eight tested hamsters shed virus for several days , mainly via the oropharynx and , at later time points , the intestinal tract ( Fig . 4 ) . Between 10 and 18 days after supplying the hamsters with artificial palm sap containing Nipah virus , 5 hamsters had to be euthanized due to the severity of disease; with neurological signs apparent in 4 out of 5 hamsters ( Fig . 2B ) . The three remaining hamsters survived until the end of the experiment at day 28; 2 out of three survivors had seroconverted at that time , indicating that 7 of 8 hamsters likely became infected after drinking artificial palm sap containing a high dose of Nipah virus ( strain Bangladesh/200401066 ) . Histopathological examination of tissues collected from hamsters euthanized due to severity of disease revealed signs of bronchointerstitial pneumonia with syncytial cells , fibrin and edema in all 5 hamsters; 2 out of 5 hamsters demonstrated signs of subacute meningitis ( Fig . 5 and Table S2 ) . On day 2 after supplying the hamsters with artificial palm sap containing Nipah virus , virus could only be detected in the nasal turbinates of one out of 4 tested hamsters; all other tested tissues were negative in virus titration and whole blood was negative by RT-PCR ( Fig . 3 and Table S1 ) . On day 4 , infectious virus was detected in respiratory tissues of 3 out of 4 tested hamsters , but no viral RNA was detected in blood . On day 8 , infectious virus was detected in tissues of two out of four hamsters ( Fig . 3 and Table S1 ) . Again , histopathological examination of tissues did not implicate involvement of the intestinal tract in virus replication or initiation of infection ( Table S2 ) . To understand how the different inoculation routes could result in virus replication in the lower respiratory tract , we fluorescently labeled Nipah virus and inoculated hamsters with 107 TCID50 of this labeled virus intranasally , esophageally or via drinking . After 10 minutes , hamsters were euthanized and the lungs and head prepared for ex vivo imaging , to visualize where the inoculum was deposited . In agreement with our tissue distribution data , a large proportion of virus was deposited in the lungs , regardless of whether animals were inoculated intranasally , esophageally or via drinking ( Fig . 6 ) . Of note , virus deposition in the stomach could not be assessed due to background fluorescence in this organ . Deposition of virus in the lungs upon esophageal inoculation was likely a result of trace inoculum entering the trachea when the gavage needle was inserted or removed; drinking may result in the generation of aerosols and/or small droplets that were subsequently deposited in the lungs . We have recently shown that Nipah virus ( strain Malaysia ) is transmitted between Syrian hamsters primarily through direct contact [14] . We have also determined the transmission route of Nipah virus ( strain Bangladesh/200401066 ) . Groups of eight hamsters were inoculated intranasally with 107 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) and singly housed to examine transmission via fomites , direct contact or aerosols as described previously [14] . At 1 dpi , a naïve hamster was added to each cage . Inoculated and naïve hamsters were swabbed daily . At 28 dpi all naïve hamsters were euthanized and sera were tested for antibodies to Nipah virus . None of the hamsters exposed through fomites or aerosols seroconverted ( Fig . 7 ) . Two out of 8 hamsters exposed via direct contact seroconverted ( Fig . 7 ) , indicating that transmission of Nipah virus ( strain Malaysia ) and Nipah virus ( strain Bangladesh/200401066 ) occurs via a similar route and at a similar rate [14] . Next , we set out to determine if hamsters infected with Nipah virus through drinking of artificial palm sap containing Nipah virus could transmit the virus to naïve hamsters via direct contact . Of note , a larger number of animals was used in this experiment since the hamsters infected with Nipah virus through drinking of artificial palm sap shed a lower amount of virus than intranasally inoculated hamsters ( Fig . 4 ) and transmission was therefore expected to be less efficient . Twenty-four hamsters were supplied with 30 ml of artificial palm sap containing 5×108 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) . After two days , when the hamsters had drunk the artificial palm sap , drinking bottles were replaced with bottles containing water and a naïve hamster was added to each cage . At 28 dpi the naïve hamsters were euthanized and sera were collected . Out of 24 naïve hamsters , 2 hamsters showed presence of antibodies directed against Nipah virus in ELISA , likely indicating that Nipah virus was transmitted to these hamsters .
Epidemiological investigations in Bangladesh suggest that Nipah virus is introduced into the human population via the consumption of raw date palm sap . This study provides the first experimental evidence for the transmission of Nipah virus via the consumption of palm sap containing Nipah virus , resulting in neurological signs of disease in Syrian hamsters . Although these findings do not directly demonstrate that date palm sap contaminated with Nipah virus by bats is the origin of Nipah virus outbreaks in Bangladesh , they provide experimental support for the current hypothesis , based on epidemiological observations , of the zoonotic introduction of Nipah virus via contaminated date palm sap . Nipah virus was very stable in artificial palm sap , likely due to its neutral pH and high sugar content . Nipah virus was preserved much better in artificial palm sap than on the surface of fruit or in fruit juice [18] . In fruit bats , Nipah virus is predominantly shed via urine [11] , [19] , [20] . Although bat urine itself may not preserve Nipah virus very well [18] , the urine would be quickly diluted in the palm sap . Thus , palm sap is likely a very suitable carrier for foodborne transmission of Nipah virus . The rapid decrease in virus titer upon heat treatment of Nipah virus-containing palm sap suggests that this might reduce the risk of Nipah virus transmission to humans . Based on virus distribution in tissues of infected hamsters and the presence of vRNA mainly in throat swabs rather than urogenital or rectal swabs , the porte d'entrée for the initial Nipah virus infection upon drinking of artificial palm sap containing Nipah virus was the respiratory tract rather than the intestinal tract . This finding was further strengthened using fluorescently labeled Nipah virus to visualize the deposition of virus upon inoculation via the nose , gavage or upon drinking . These experiments clearly showed that during drinking the virus does not only end up in the intestinal tract but some of the volume is also deposited in the lungs , thereby explaining the replication of virus in respiratory tissues . However , the main disease manifestation in hamsters infected through drinking was the development of neurological signs , suggesting that the animals became infected with a relatively low dose , despite the high amount of virus present in the artificial palm sap . Previous studies have shown that inoculation of hamsters with a low dose of Nipah virus results in neurological signs of disease , whereas a high dose results in respiratory disease [15] , [21] . In one of the hamsters infected with Nipah virus through drinking of palm sap , virus was detected by immunohistochemistry in the olfactory bulb , indicating that virus traveled from olfactory neurons in the nasal turbinates through the cribriform plate into the olfactory bulb and from there further into the central nervous system . Although this route of Nipah virus into the brain has been described before [22] , [23] , this is the first time it is described in animals without deliberate inoculation of the nasal cavity . Besides the ability of Nipah virus to infect hamsters via the drinking route , we showed here that Syrian hamsters infected with Nipah virus through drinking of palm sap containing Nipah virus can transmit the virus through direct contact with naïve hamsters . Transmission upon drinking of Nipah virus-containing artificial palm sap was less efficient than upon intranasal inoculation with Nipah ( strain Bangladesh/200401066 ) , likely as a result of decreased virus shedding upon infection through drinking . Within the transmission model for Nipah virus in Syrian hamsters no differences were observed in transmission route or efficiency between the Malaysian [14] and a virus isolate from Bangladesh . Although experimental infection of ferrets suggested that there is increased oral shedding with a Nipah virus isolate from Bangladesh as compared to a virus isolate from Malaysia , this study did not include transmission experiments [24] . Thus it is currently not clear whether the differences in virus shedding observed in the ferret model result in differences in transmission efficiency . Different Nipah virus isolates from several Nipah virus outbreaks in Bangladesh would have to be tested in the different animal models to assess the transmission efficiency of this virus properly . Prophylactic or therapeutic intervention measures are currently not available to prevent , treat or contain zoonotic transmission of Nipah virus . Moreover , medical interventions might be difficult to implement in rural outbreak areas . Therefore , our best hope to prevent or intervene in future outbreaks of Nipah virus lies in the potential to efficiently block zoonotic and human-to-human transmission and thereby spread of the outbreak . Currently , efforts are underway in Bangladesh to prevent zoonotic transmission of Nipah virus from fruit bats to people by restricting access of bats to date palm collection pots and thereby preventing contamination of the date palm sap with Nipah virus [8] , [25] . The data presented here stress the importance of these efforts in Bangladesh in the prevention of Nipah virus outbreaks .
All animal experiments were approved by the Institutional Animal Care and Use Committee of the Rocky Mountain Laboratories , and performed following the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care , International ( AAALAC ) by certified staff in an AAALAC-approved facility . Nipah virus ( strain Bangladesh/200401066 ) was kindly provided by the Special Pathogens Branch of the Centers for Disease Control and Prevention , Atlanta , Georgia , United States . This strain ( SPBLOG# 200401066 ) was isolated from a throat swab collected from patient #3001 on January 22 2004 in Bangladesh . This patient was a 10-year old male who developed altered mental status on January 21 and cough and breathing difficulties later that day . The patient was admitted to Goalando Hospital , Bangladesh , on January 22 . None of the patient's contacts developed Nipah virus infection; the patient is presumed to have been infected via direct spillover from the bat reservoir ( dr . Steve Luby , personal communication ) . The virus isolate was propagated in Vero C1008 cells in DMEM ( Sigma ) supplemented with 10% fetal calf serum , 1 mM L-glutamine ( Lonza ) , 50 U/ml penicillin and 50 µg/ml streptomycin ( Gibco ) . For fluorescent labelling of virus , Nipah virus-containing cell supernatant was cleared by low speed centrifugation and virus was pelleted by spinning 2 hours at 21000 rpm in the ultracentrifuge . The pellet was resuspended in 1 ml PBS and loaded onto a 20%–60% ( w/w ) sucrose gradient and centrifuged overnight at 39000 rpm . The virus fraction was collected and pelleted once again by centrifuging 2 hours at 21000 rpm; the pellet was resuspended in 1 ml PBS . Purified Nipah virus particles were labeled using an Alexa Fluor 680 Protein Labeling Kit ( Molecular Probes ) . Excess dye was removed by dialyzing against PBS . Artificial palm sap was prepared based on a literature report [17] and consisted of 13% sucrose ( w/v ) and 0 . 21% BSA in water . The pH of the artificial palm sap was 7 without any adjustments . Nipah virus ( strain Bangladesh/200401066 ) was added to artificial palm sap at the desired concentration , aliquotted into 1 ml aliquots and incubated at 22°C or 28°C for up to eight days . The stability data were analyzed using the linear regression model in the GraphPad prism 6 software package . For inactivation of Nipah virus in artificial palm sap , Nipah virus ( strain Bangladesh/200401066 ) was added to artificial palm sap at 107 TCID50/ml , aliquotted into 1 ml aliquots and incubated at 70°C or 100°C for up to one hour . Four groups of 40 6–8 week old female Syrian hamsters ( HsdHantm:AURA , Harlan Laboratories ) were inoculated with Nipah virus via different routes . One group received 107 TCID50 Nipah virus ( strain Bangladesh/200401066 ) via intranasal inoculation in a total volume of 80 µl . One group received 107 TCID50 Nipah virus ( strain Bangladesh ) via gavage in a total volume of 500 µl . The remaining two groups received 107 and 5×108 TCID50 Nipah virus ( strain Bangladesh/200401066 ) , respectively through drinking of artificial palm sap . Animals were housed singly and supplied with 30 ml artificial palm sap containing a total dose of 107 or 5×108 TCID50 Nipah virus instead of drinking water . When animals had drunk all artificial palm sap , in about 2 days , they were again supplied with drinking water . Nasal , oral , urogenital and rectal swabs were collected daily from eight hamsters inoculated via all four different routes . Swabs were collected in vials containing 1 ml DMEM supplemented with 50 U/ml penicillin and 50 µg/ml streptomycin . On days 2 , 4 , 8 , 12 and 28 post inoculation 8 animals from each inoculation group were euthanized and blood , trachea , lungs , heart , liver , spleen , kidney , esophagus , stomach , duodenum , jejunum , ileum , cecum , colon ( proximal and distal ) , bladder , brain and nasal turbinates were collected for virological ( 4 animals/time point ) and histopathological ( 4 animals/time point ) analysis . Hamsters used for histopathological analysis were anaesthetized using ketamine ( 80–100 mg/kg ) and xylazine ( 7–10 mg/kg ) and perfused with PBS containing 5 mM EDTA , followed by 4% paraformaldehyde . Tissues of interest were then further fixed according to BSL4 standard operating procedures for a minimum of 7 days in 10% neutral buffered formalin . To visualize the deposition of virus in the hamster respiratory tract after inoculation , 3 hamsters per inoculation route were inoculated intranasally , esophageally and via drinking as described above with 107 TCID50 of fluorescently labeled Nipah virus ( strain Bangladesh/200401066 ) . To prevent fusion of virus particles with target cells , hamsters were euthanized ten minutes after inoculation and the respiratory tract was excised; ex vivo imaging was subsequently performed on the head and respiratory tract in an IVIS Spectrum imager ( PerkinElmer ) . Images were acquired using an excitation wavelength of 675 nm with an emission scan at 720 , 740 , 760 and 780 nm at field of view B ( 6 . 6 cm ) in auto-exposure mode with medium binning , f-stop 3 . Following acquisition , images were unmixed in Living Image 4 . 2 with tissue autofluorescence subtracted . The resulting AF680 image was used for subsequent analysis . Rectangular Regions of Interest were drawn around the entire lung , trachea or nasal tract . The resulting average radiant efficiency was used to determine the quantity of labelled virus that was detected in the respiratory tract ( combined trachea and lungs ) or in the nasal tract ( head ) . To determine the transmission route for Nipah virus ( strain Bangladesh/200401066 ) we used the recently described Syrian hamster transmission model [14] . For fomite transmission experiments , eight 6–8 week old female singly housed Syrian hamsters , housed in a plastic cage with wood shavings , a feeder and a water bottle , were inoculated intranasally with 107 TCID50 of Nipah virus ( strain Bangladesh ) in a total volume of 80 µl . On day 4 post inoculation , hamsters were euthanized and a single naïve hamster was placed in each cage . For direct contact transmission experiments , eight 6–8 week old female singly housed Syrian hamsters were inoculated intranasally with 107 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) in a total volume of 80 µl . On day 1 post inoculation , a naïve hamster was added to each cage . For aerosol transmission experiments , eight 6–8 week old female Syrian hamsters were inoculated intranasally with 107 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) in a total volume of 80 µl and singly housed in specially designed aerosol transmission cages . On 1 dpi , a naïve hamster was placed on the opposite side of the inoculated hamster . The hamsters were separated by two stainless steel grids , allowing airflow from the inoculated to the naive hamster but preventing direct contact and fomite transmission . In all transmission experiments , nasal and oropharyngeal swabs were obtained from inoculated and naïve hamsters daily and the bodyweight of naïve hamsters was determined . Upon signs of severe disease , inoculated and naïve hamsters were euthanized; remaining hamsters were euthanized four weeks post exposure . To determine whether Nipah virus ( strain Bangladesh/200401066 ) is transmitted via direct contact after infection through palm sap containing Nipah virus , 24 6–8 week old female singly housed Syrian hamsters were supplied with 30 ml artificial palm sap containing 5×108 TCID50 of Nipah virus ( strain Bangladesh/200401066 ) instead of drinking water . When animals had drunk all artificial palm sap they were again supplied with drinking water and a naïve hamster was added to each cage . Nasal and oropharyngeal swabs were obtained from inoculated and naïve hamsters daily and bodyweight of naïve hamsters was determined . On signs of severe disease , inoculated and naïve hamsters were euthanized; remaining hamsters were euthanized four weeks post exposure . Virus titrations were performed by end-point titration in VeroC1008 cells . VeroC1008 cells were inoculated with tenfold serial dilutions of swab medium or tissue homogenates . One hour after inoculation , the inoculum was removed and replaced with 200 µl DMEM supplemented with 10% fetal calf serum , 1 mM L-glutamine ( Lonza ) , 50 U/ml penicillin and 50 µg/ml streptomycin ( Gibco ) . Five days after inoculation , cytopathic effect ( CPE ) was scored and the TCID50 was calculated from 5 replicates by the method of Spearman-Karber . Tissue homogenates were prepared by adding 1 ml DMEM to the weighed tissue and homogenizing using a TissueLyzer II ( Qiagen ) . Homogenates were centrifuged to clear the homogenate before inoculating cells . Histopathology and immunohistochemistry was performed on hamster tissues . Necropsies and tissue sampling were performed according to a standard protocol approved by the Institutional Biosafety Committee . After fixation for 7 days in 10% neutral-buffered formalin and embedding in paraffin , tissue sections were stained with hematoxylin and eosin ( H&E ) and an immunohistochemical method using a rabbit polyclonal antiserum against the Nipah virus nucleoprotein [26] ( 1∶5000; kindly provided by L . Wang , CSIRO Livestock Industries , Australian Animal Health Laboratory , Australia ) as a primary antibody for detection of Nipah virus antigen . For the histopathological analysis of the nasal turbinates ( NT ) whole hamster skulls were used . The skulls were decalcified using a 20% EDTA solution in sucrose ( Newcomer Supply ) and allowed to sit at room temperature for 3 weeks . The 20% EDTA/sucrose solution was changed 2 times prior to gross sectioning the skull . The following tissues were examined: NT , trachea and lungs . Lesions were assigned a subjective score from 0 to 4 based on the percentage of the tissue that was immunopositive . The slides were evaluated by a board-certified veterinary pathologist . RNA was extracted from swab samples using the NucleoSpin 96 Virus Core kit ( Macherey-Nagel ) and a Corbett Robotics model CAS 1820 automatic RNA extractor . RNA was eluted in 100 µl . 5 µl RNA was used in a one-step real-time RT-PCR targeted at the NP gene using the Rotor-Gene probe kit ( Qiagen ) according to instructions of the manufacturer ( primer and probe sequences are available on request ) . In each run , standard dilutions of a titered virus stock were run in parallel , to calculate TCID50 equivalents in the samples . Antibody responses were measured in an enzyme-linked immunosorbent assay ( ELISA ) using inactivated Nipah virus ( strain Malaysia ) as the antigen . Nipah virus-containing cell culture supernatant was concentrated and purified by centrifuging for two hours at 21000 rpm over a 20% sucrose cushion . The pellet was resuspended in PBS and Triton X-100 was added to a final concentration of 1%; the preparation was then inactivated with γ-radiation according to standard operating procedures . This suspension was used to coat immuno 96 microwell maxisorp plates ( NUNC ) at 4°C overnight . Subsequently , plates were blocked with 5% skim milk in PBS containing 0 . 05% Tween 20 ( PBST ) for 1 . 5 hours at 4°C . After 3 washes with PBST , 50 µL of diluted serum samples were added , and the plates were incubated for 1 hour at 37°C . Bound antibodies were detected after 3 washes using an anti-hamster secondary antibody conjugated with horseradish peroxidase ( HRP; KPL ) . Following incubation for 1 hour at 37°C , bound HRP was detected using the ABTS Peroxidase Substrate System ( KPL ) . The absorbance at 405 nm was measured using a microplate spectrophotometer . Sera were considered positive when absorbance was higher than three standard deviations above the mean of negative control sera .
|
In Bangladesh , outbreaks of Nipah virus occur almost every year , resulting in respiratory and neurological disease with high case-fatality rates . Based on epidemiological data Nipah virus is thought to be transmitted from fruit bats to humans via drinking of date palm sap contaminated by bats that drink from the sap stream or collection vessel during collection . Additionally , human-to-human transmission has been shown to occur . Here , we experimentally modeled the proposed transmission cycle of Nipah virus in Bangladesh in Syrian hamsters . Hamsters that drank artificial palm sap containing high doses of Nipah virus became infected with the virus and developed neurological signs of disease . Virus replication occurred mainly in the respiratory rather than the intestinal tract . Most importantly , hamsters infected with Nipah virus through drinking of contaminated palm sap could transmit the virus to uninfected cage mates . As treatments for Nipah virus are currently unavailable and medical interventions are difficult to implement in rural outbreak areas , our best hope to prevent or intervene in future outbreaks of Nipah virus lies in the potential to block transmission from bats to humans and from human to human . Understanding how Nipah virus is transmitted is essential to achieve this .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"viral",
"transmission",
"and",
"infection",
"emerging",
"viral",
"diseases",
"nipah",
"fever",
"microbiology",
"animal",
"models",
"model",
"organisms",
"viral",
"pneumonia",
"emerging",
"infectious",
"diseases",
"neglected",
"tropical",
"diseases",
"infectious",
"disease",
"control",
"viral",
"load",
"animal",
"models",
"of",
"infection",
"infectious",
"diseases",
"zoonoses",
"biology",
"infectious",
"disease",
"modeling",
"pathogenesis",
"virology",
"viral",
"diseases"
] |
2014
|
Foodborne Transmission of Nipah Virus in Syrian Hamsters
|
Human genome-wide association studies ( GWASs ) are revealing the genetic architecture of anthropomorphic and biomedical traits , i . e . , the frequencies and effect sizes of variants that contribute to heritable variation in a trait . To interpret these findings , we need to understand how genetic architecture is shaped by basic population genetics processes—notably , by mutation , natural selection , and genetic drift . Because many quantitative traits are subject to stabilizing selection and because genetic variation that affects one trait often affects many others , we model the genetic architecture of a focal trait that arises under stabilizing selection in a multidimensional trait space . We solve the model for the phenotypic distribution and allelic dynamics at steady state and derive robust , closed-form solutions for summary statistics of the genetic architecture . Our results provide a simple interpretation for missing heritability and why it varies among traits . They predict that the distribution of variances contributed by loci identified in GWASs is well approximated by a simple functional form that depends on a single parameter: the expected contribution to genetic variance of a strongly selected site affecting the trait . We test this prediction against the results of GWASs for height and body mass index ( BMI ) and find that it fits the data well , allowing us to make inferences about the degree of pleiotropy and mutational target size for these traits . Our findings help to explain why the GWAS for height explains more of the heritable variance than the similarly sized GWAS for BMI and to predict the increase in explained heritability with study sample size . Considering the demographic history of European populations , in which these GWASs were performed , we further find that most of the associations they identified likely involve mutations that arose shortly before or during the Out-of-Africa bottleneck at sites with selection coefficients around s = 10−3 .
Much of the phenotypic variation in human populations , including variation in morphological , life history , and biomedical traits , is “complex” or “quantitative” , in the sense that heritable variation in the trait is largely due to small contributions from many genetic variants segregating in the population [1 , 2] . Quantitative traits have been studied since the birth of biometrics over a century ago [1–3] , but only in the past decade have technological advances made it possible to systematically dissect their genetic basis [4–6] . Notably , since 2007 , genome-wide association studies ( GWASs ) in humans have led to the identification of many thousands of variants reproducibly associated with hundreds of quantitative traits , including susceptibility to a wide variety of diseases [4] . While still ongoing , these studies already provide important insights into the genetic architecture of quantitative traits , i . e . , the number of variants that contribute to heritable variation , as well as their frequencies and effect sizes . Perhaps the most striking observation to emerge from these studies is that , despite the large sample size of many GWASs , all variants significantly associated with a given trait typically account for less ( often much less ) than 25% of the narrow sense heritability ( [4 , 7 , 8] , but see [9] ) . ( Henceforth , we use “heritability” to refer to narrow sense heritability . ) While many factors have been hypothesized to contribute to the “missing heritability” [7 , 8 , 10–14] , the most straightforward explanation and the emerging consensus is that much of the heritable variation derives from variants with frequencies that are too low or effect sizes that are too small for current studies to detect . Comparisons among traits also suggest that there are substantial differences in architectures . For example , recent meta-analysis GWASs uncovered 7 times as many variants for height ( 697 ) as for body mass index ( 97 ) , and together , the variants for height account for more than 4 times the heritable variance than the variants for body mass index do ( approximately 20% versus approximately 3%–5% , respectively ) , despite comparable sample sizes [15 , 16] . These first glimpses underscore the need for theory that relates the findings emerging from GWASs with the evolutionary processes that shape genetic architectures . Such theory would help to interpret the “missing heritability” [17–20] and to explain why architecture differs among traits . It may also allow us to use GWAS findings to make inferences about underlying evolutionary parameters , helping to answer enduring questions about the processes that maintain phenotypic variation in quantitative traits [5 , 21] . Development of such theory can be guided by empirical observations and first-principles considerations . New mutations affecting a trait arise at a rate that depends on its “mutational target size” ( i . e . , the number of sites at which a mutation would affect the trait ) . Once they arise , the trajectories of variants through the population are determined by the interplay between genetic drift , demographic processes , and natural selection acting on them . These processes determine the number and frequencies of segregating variants underlying variation in the trait . The genetic architecture further depends on the relationship between the selection on variants and their effects on the trait . Notably , selection on variants depends not only on their effect on the focal trait but also on their pleiotropic effects on other traits . We therefore expect both direct and pleiotropic selection to shape the joint distribution of allele frequencies and effect sizes . Multiple lines of evidence suggest that many quantitative traits are subject to stabilizing selection , i . e . , selection favoring an intermediate trait value [5 , 22–27] . For instance , a decline in fitness components ( e . g . , viability and fecundity ) is observed with displacement from mean values for a variety of traits in human populations [28–30] , in other species in the wild [31 , 32] , and in experimental manipulations [31 , 33] . While less is known about complex diseases , they may often reflect large deviations of an underlying continuous trait from an optimal value [1] , with these continuous traits subject to directional ( purifying ) selection in some cases and to stabilizing selection in others . What remains unclear is the extent to which stabilizing selection is acting directly on variation in a given trait or is “apparent” , i . e . , results from pleiotropic effects of this variation on other traits . Other lines of evidence suggest that pleiotropy is pervasive . For one , theoretical considerations about the variance in fitness in natural populations and its accompanying genetic load suggest that only a moderate number of independent traits can be effectively selected on at once [34] . Thus , the aforementioned relationships between the value of a focal trait and fitness are likely heavily affected by the pleiotropic effects of genetic variation on other traits [25 , 34–36] . Second , many of the variants detected in human GWASs have been found to be associated with more than one trait [37–41] . For example , a recent analysis of GWASs revealed that variants that delay the age of menarche in women tend to delay the age of voice drop in men , decrease body mass index , increase adult height , and decrease risk of male pattern baldness [37] . More generally , the extent of pleiotropy revealed by GWASs appears to be increasing rapidly with improvements in power and methodology [37 , 42–45] . These considerations and others [45 , 46] point to the general importance of pleiotropic selection on quantitative genetic variation . The discoveries emerging from human GWASs further suggest that genetic variance is dominated by additive contributions from numerous variants with small effect sizes . Dominance and epistasis may be common among newly arising mutations of large effect ( e . g . , [47–51] ) , but both theory and data suggest that they play a minor role in shaping quantitative genetic variation within populations ( e . g . , [9 , 52–56] ) . Indeed , for many traits , most or all of the heritability explained in GWASs arises from the additive contribution of variants with squared effect sizes that are substantially smaller than the total genetic variance ( e . g . , [15 , 16 , 57 , 58] ) . Moreover , statistical quantifications of the total genetic variance tagged by genotyping ( i . e . , not only due to the genome-wide significant associations ) suggest that such contributions may account for most of the heritable variance in many traits ( e . g . , [9 , 59–61] ) . Finally , considerable efforts to detect epistatic interactions in human GWASs have , by and large , come up empty-handed [9 , 56 , 62] , with few counterexamples , mostly involving variants in the major histocompatibility complex region ( [53 , 56 , 63 , 64] , but see [65] ) . Thus , while the discovery of epistatic interactions may be somewhat limited by statistical power [56] , theory and current evidence suggest that nonadditive interactions play a minor role in shaping human quantitative genetic variation . Motivated by these considerations , we model how direct and pleiotropic stabilizing selection shape the genetic architecture of continuous , quantitative traits by considering additive variants with small effects and assuming that together they account for most of the heritable variance . To date , there has been relatively little theoretical work relating population genetics processes with the results emerging from GWASs . Moreover , the few existing models have reached divergent predictions about genetic architecture , largely because they make different assumptions about the effects of pleiotropy . Focusing on disease susceptibility , Pritchard [19] considered the “purely pleiotropic” extreme , in which selection on variants is independent of their effect on the trait being considered . In this case , we expect the largest contribution to genetic variance in a trait to come from mutations that have large effect sizes but are also weakly selected or neutral , allowing them to ascend to relatively high frequencies . Other studies considered the opposite extreme , in which selection on variants stems entirely from their effect on the trait under consideration [26 , 66–70] , and have shown that the greatest contribution to genetic variance would arise from strongly selected mutations [67 , 68] ( we return to this case below ) . In practice , we expect most traits to fall somewhere in between these extremes . While there are compelling reasons to believe that quantitative genetic variation is highly pleiotropic , the effects of variants on different traits are likely to be correlated . Thus , even if a given trait is not subject to selection , variants that have a large effect on it will also tend to have larger effects on traits that are under selection ( e . g . , by causing large perturbation to pathways that affect multiple traits [36 , 45] ) . Motivated by such considerations , Eyre-Walker ( 2010 ) [20] , Keightley and Hill ( 1990 ) [18] , and Caballero et al . ( 2015 ) [71] considered models in which the correlation between the strength of selection on an allele and its effect size can vary between the purely pleiotropic and direct selection extremes . These models diverge in their predictions about architecture , however . Assuming , as seems plausible , an intermediate correlation between the strength of selection and effect size , Eyre-Walker finds that genetic variance should be dominated by strongly selected mutations [20] , whereas Keightley and Hill and Caballero et al . conclude that the greatest contribution should arise from weakly selected ones [18 , 71] . Their conclusions differ because of how they chose to model the relationship between selection and effect size , a choice based largely on mathematical convenience . We approach this problem by explicitly modeling stabilizing selection on multiple traits , thereby learning , rather than assuming , the relationship between selection and effect sizes .
We model stabilizing selection in a multidimensional phenotype space , akin to Fisher’s geometric model [72] . An individual’s phenotype is a vector in an n-dimensional Euclidian space , in which each dimension corresponds to a continuous quantitative trait . We focus on the architecture of one of these traits ( say , the first dimension ) , where the total number of traits parameterizes pleiotropy . Fitness is assumed to decline with distance from the optimal phenotype positioned at the origin , thereby introducing stabilizing selection . Specifically , we assume that absolute fitness takes the form W ( r→ ) =exp ( −r22w2 ) , ( 1 ) where r→ is the ( n-dimensional ) phenotype , r=‖r→‖ is the distance from the origin , and w parameterizes the strength of stabilizing selection . However , we later show that the specific form of the fitness function does not matter . Moreover , the additive environmental contribution to the phenotype can be absorbed into w ( [73]; Section 1 . 1 in S1 Text ) ; we therefore consider only the genetic contribution . The genetic contribution to the phenotype follows from the multidimensional additive model [74] . Specifically , we assume that the number of genomic sites affecting the phenotype ( the target size ) is very large , L ≫ 1 , and that allelic effects on the phenotype at these sites are vectors in the n-dimensional trait space . An individual’s phenotype then follows from adding up the effects of her or his alleles , i . e . , r→=∑l=1L ( a→l+a→l′ ) , ( 2 ) where a→l and a→l′ are the phenotypic effects of the parents’ alleles at site l . The population dynamics follows from the standard model of a diploid , panmictic population of constant size N , with nonoverlapping generations . In each generation , parents are randomly chosen to reproduce with probabilities proportional to their fitness ( i . e . , Wright-Fisher sampling with viability selection ) , followed by mutation , free recombination ( i . e . , no linkage ) , and Mendelian segregation . We further assume that the mutation rate per site , u , and the population size are sufficiently small such that no more than 2 alleles segregate at any time at each site ( i . e . , that θ = 4Nu ≪ 1 ) and therefore an infinite sites approximation applies . The number of mutations per gamete per generation therefore follows a Poisson distribution with mean U = Lu; based on biological considerations ( see Sections 4 . 1 and 4 . 2 in S1 Text ) , we also assume that 1 ≫ U ≫ 1/2N . The size of mutations in the n-dimensional trait space , a ( =‖a→‖ ) , is drawn from some distribution , assuming only that a2 ≪ w2 . We later show that this requirement is equivalent to the standard assumption about selection coefficients satisfying s ≪ 1 ( also see Section 4 . 3 in S1 Text ) . The directions of mutations are assumed to be isotropic , i . e . , uniformly distributed on the hypersphere in n-dimensions defined by their size , although we later show that our results are robust to relaxing this assumption as well .
In the first 3 sections , we develop the tools that we later use to study genetic architecture . We start by considering the equilibrium distribution of phenotypes in the population and generalizing previous results for the case with a single trait [26 , 66 , 67 , 70] . Under biologically sensible conditions , this distribution is well approximated by a tight multivariate normal centered at the optimum . Namely , the distribution of n-dimensional phenotypes , r→ , in the population , is well approximated by the probability density function: f ( r→ ) =1 ( 2πσ2 ) n/2exp ( −r22σ2 ) , ( 3 ) where σ2 is the genetic variance of the phenotypic distances from the optimum ( see Eq A25 in S1 Text for closed form ) ; and under plausible assumptions about the rate and size of mutations ( i . e . , when 1 ≫ U ≫ 1/2N and a2 ≪ w2 ) , it satisfies σ2 ≪ w2 , implying small variance in fitness in the population ( Section 4 . 2 in S1 Text ) . Intuitively , the phenotypic distribution is normal because it derives from additive and ( approximately ) independently and identically distributed contributions from many segregating sites . Moreover , the population mean remains extremely close to the optimum because stabilizing selection becomes increasingly stronger with the displacement from it and because any displacement is rapidly offset by minor changes to allele frequencies at many segregating sites . With phenotypes close to the optimum , only the curvature of the fitness function at the optimum ( i . e . , the multidimensional second derivative ) affects the selection acting on individuals . In addition , it is always possible to choose an orthonormal coordinate system centered at the optimum , in which the trait under consideration varies along the first coordinate and a unit change in other traits ( along other coordinates ) near the optimum has the same effect on fitness . These considerations suggest that the equilibrium behavior is insensitive to our choice of fitness function around the optimum . Moreover , in S1 Text ( Section 5 ) , we show that the rapid offset of perturbations of the population mean from the optimum ( by minor changes to allele frequencies at numerous sites ) lends robustness to the equilibrium dynamics with respect to the presence of major loci , moderate changes in the optimal phenotype over time , and moderate asymmetries in the mutational distribution . Next , we consider the dynamic at a segregating site and generalize previous results for the case with a single trait [68–70] . This dynamic can be described in terms of the first 2 moments of change in allele frequency in a single generation ( see , e . g . , [75] ) . To calculate these moments for an allele with phenotypic effect a→ and frequency q ( =1-p ) , we note that the phenotypic distribution can be well approximated as a sum of the expected contribution of the allele to the phenotype , 2qa→ , and the distribution of contributions to the phenotype from all other sites , R→ . From Eq 3 , it then follows that the distribution of background contributions is well approximated by probability density: f ( R→|a→ , q ) =1 ( 2πσ2 ) n/2exp ( − ( R→+2qa→ ) 22σ2 ) . ( 4 ) By averaging the fitness of the 3 genotypes at the focal site over the distribution of genetic backgrounds , we find that the first moment is well approximated by E ( Δq ) ≈a2w2 pq ( q−12 ) , ( 5 ) assuming that a2 and σ2 ≪ w2 ( Section 4 in S1 Text ) . By the same token , we find that V ( Δq ) ≈pq2N , ( 6 ) which is the standard second moment with genetic drift . The functional form of the first moment is equivalent to that of the standard viability selection model with underdominance . This result is a hallmark of stabilizing selection on ( additive ) quantitative traits: with the population mean at the optimum , the dynamics at different sites are decoupled , and selection at a given site acts to reduce its contribution to the phenotypic variance ( 2a2pq ) , thereby pushing rare alleles to loss . Comparison with the standard viability selection model shows that the selection coefficient in our model is s = a2/w2 , or S = 2Ns = 2Na2/w2 in scaled units . In other words , the selection acting on an allele is proportional to its size squared in the n-dimensional trait space ( where w translates effect size into units of fitness ) . The statistical relationship between the strength of selection acting on mutations and their effect on a given trait follows from the aforementioned geometric interpretation of selection . Specifically , all mutations with a given selection coefficient , s , lie on a hypersphere in n-dimensions with radius a=ws , and any given mutation satisfies s=1w2a2=1w2∑i=1nai2 , ( 7 ) where ai is the allele’s effect on the i-th trait ( Fig 1A ) . Our assumption that mutation is isotropic then implies that the probability density of mutations on the hypersphere is uniform . The distribution of effect sizes on a focal trait , a1 , corresponding to a given selection coefficient , s , follows . Given that mutation is symmetric in any given trait , E ( a1|s ) = 0 , and given that it is symmetric among traits , E ( a12|s ) =a2 /n= ( w2/n ) s . ( 8 ) More generally , the probability density corresponding to an effect size a1 is proportional to the volume of the ( n − 2 ) –dimensional cross section of the hypersphere with projection a1 ( Fig 1A ) . For a single trait , this implies that a1 = ±a with probability ½ , and for n > 1 , it implies the probability density φn ( a1|a ) =Γ ( n/2 ) /Γ ( ( n−1 ) /2 ) n/212π ( a2/n ) ( 1−1na12 ( a2/n ) ) n−32 ( 9 ) ( Section 1 . 2 in S1 Text ) . Intriguingly , when the number of traits n increases , this density approaches a normal distribution , i . e . , a1a2/n~N ( 0 , 1 ) , ( 10 ) implying that the distribution of effect sizes given the selection coefficient becomes a1~N ( 0 , ( w2/n ) s ) . ( 11 ) This limit is already well approximated for a moderate number of traits ( e . g . , n = 10; Fig 1B ) . The limit behavior also holds when we relax the assumption of isotropic mutation . This generalization is important because , having chosen a parameterization of traits in which the fitness function near the optimum is isotropic , we can no longer assume that the distribution of mutations is also isotropic [76] . Specifically , mutations might tend to have larger effects on some traits than on others , and their effects on different traits might be correlated . In Section 5 . 4 in S1 Text , we show that the limit distribution ( Eq 11 ) also holds for anisotropic mutation ( excluding pathological cases ) . To this end , we introduce the concept of an effective number of traits , ne , which can take any real value ≥1 and is defined as the number of equivalent traits required to generate the same relationship between the strength of selection on mutations and their expected effects on the trait under consideration ( i . e . , replacing n in Eq 11 ) . The robustness of our model , along with mounting evidence that genetic variation is highly pleiotropic ( see “Introduction” ) , suggests that the limit form may apply quite generally . In that regard , we note that even in this limit , the strength of selection on mutations and their effects on the focal trait are correlated , implying that the kind of “purely pleiotropic” extreme postulated in previous works cannot arise [18–20] . We can now derive closed forms for summary statistics of the genetic architecture ( see Section 2 . 3 in S1 Text ) . For mutations with a given selection coefficient , the frequency distribution follows from the diffusion approximation based on the first 2 moments of change in allele frequency ( Eqs 5 and 6; [75] ) , and the distribution of effect sizes follows from the geometric considerations of the previous section . Conditional on the selection coefficient , these distributions are independent , and therefore , the joint distribution of frequency and effect size equals their product . Summaries of architecture can be expressed as expectations over the joint distribution of frequencies and effect sizes for a given selection coefficient and then weighted according to the distribution of selection coefficients . While we know little about the distribution of selection coefficients of mutations affecting quantitative traits , we can draw general conclusions from examining how summaries of architecture depend on the strength of selection . We focus on the distribution of additive genetic variances among sites , a central feature of architecture that is key to connecting our model with GWAS results . We start by considering how selection affects the expected contribution of a site to additive genetic variance in a focal trait . We include monomorphic sites in the expectation , such that the expected total variance is given by the product of the expectation per-site and the population mutation rate , 2NU . Under the infinite sites assumption , sites are monomorphic or biallelic , and their expected contribution to variance is E ( 2a12pq|S ) =E ( a12|S ) E ( 2pq|S ) =w22NnSE ( 2pq|S ) ( 12 ) ( expressed in terms of the scaled selection coefficient S ) . Thus , the degree of pleiotropy only affects the expectation through a multiplicative constant . This multiplicative factor would have a discernable effect in generalizations of our model in which the degree of pleiotropy varies among sites . For example , if the degree of pleiotropy of one set of sites was k and of another set was l > k , and both sets were subject to the same strength of selection , then the expected contribution to genetic variance of sites in the first set would be l/k times greater than in the second ( from Eq 12 ) . While such generalizations may prove interesting in the future , here we focus on the model in which the degree of pleiotropy is constant . In this case , the multiplicative factor introduced by pleiotropy is not identifiable from data , because even if we could measure genetic variance in units of fitness ( e . g . , rather than in units of the total phenotypic variance ) , we still would not be able to distinguish between the effects of w and n on the genetic variance per site . We therefore focus on the effect of selection on the relative contribution to variance , which is insensitive to the degree of pleiotropy in our model . The effect of selection on the relative contribution to genetic variance was described by Keightley and Hill ( in the one-dimensional case [68] ) and is depicted in Fig 2A . When selection is strong ( roughly corresponding to S > 30 ) , its effect on allele frequency ( which scales with 1/S ) is canceled out by its relationship with the effect size ( Eq 8 ) , yielding a constant contribution to genetic variance per site , vS = 2w2/nN , regardless of the selection coefficient ( Section 3 . 1 in S1 Text; Fig 2A and Fig A1b in S1 Text ) . Henceforth , we measure genetic variance in units of vS . When selection is effectively neutral ( roughly corresponding to S < 1 ) and thus too weak to affect allele frequency , the expected contribution of a site to genetic variance scales with the effect size and equals ½S ( ·vs ) and therefore is lower than under strong selection ( Section 3 . 1 in S1 Text; Fig 2A and Fig A1a in S1 Text ) . In between these selection regimes , selection effects on allele frequency are more complex and are influenced by underdominance ( Section 3 . 1 in S1 Text ) . As the selection coefficient increases , the expected contribution to variance reaches vS at S ≈ 3 and continues to increase until it reaches a maximal contribution that is approximately 30% greater at S ≈ 10 ( Fig 2A ) , after which it slowly declines to the asymptotic value of vS ( Fig 2A and Fig A1b in S1 Text ) . Henceforth , we refer to this selection regime as intermediate ( not to be confused with the nearly neutral range , which is much narrower and does not include selection coefficients with S > 10 ) . These results suggest that effectively neutral sites should contribute much less to genetic variance than intermediate and strongly selected ones [67 , 68] . While intermediate and strongly selected sites contribute similarly to variance , their minor allele frequencies ( MAFs ) can differ markedly ( Fig 2B ) . As an illustration , segregating sites with MAF > 0 . 1 account for approximately 72% and approximately 49% of the additive genetic variance for intermediate selection coefficients of S = 3 and 10 , respectively , when almost no segregating sites would be found at such high MAF for a strong selection coefficient of S = 100 ( Fig 2B ) . Thus , within the wide range of selection coefficients characterized as intermediate and strong , genetic variance arises from sites segregating at a wide range of MAFs ranging from common to exceedingly rare . Next , we consider how genetic variance is distributed among sites with a given selection coefficient . We focus on the distribution among segregating sites ( including monomorphic effects would just add a point mass at 0 ) . This distribution is especially relevant to interpreting the results of GWASs , because , to a first approximation , a study will detect only sites with contributions to variance exceeding a certain threshold , v ( =2a12pq ) , which decreases as the study size increases ( see “Discussion” ) . We therefore depict the distribution in terms of the proportion of genetic variance , G ( v ) , arising from sites whose contribution to genetic variance exceeds a threshold v . We begin with the case without pleiotropy ( n = 1 ) , in which selection on an allele determines its effect size ( Fig 3A ) . When selection is strong ( S > 30 ) , the proportion of genetic variance exceeding a threshold v is also insensitive to the selection coefficient and takes a simple form , with G ( v ) =exp ( −2v ) ( 13 ) ( Fig 3A; Section 3 . 2 in S1 Text ) . In contrast , in the effectively neutral range ( S < 1 ) , G ( v ) =1−v/vmax , ( 14 ) where the dependency on the selection coefficient enters through vmax=18S , which is the maximal contribution to variance and corresponds to an allele frequency of ½ ( Fig A4a; Section 3 . 2 in S1 Text ) . In the intermediate selection regime , G ( v ) is also intermediate and takes a more elaborate functional form ( Section 3 . 2 in S1 Text ) . These results suggest how genetic variance would be distributed among sites given any distribution of selection coefficients ( Fig 3A ) : starting from sites that contribute the most , the distribution would at first be dominated by strongly selected sites , and then the intermediate selected sites would begin to contribute , whereas effectively neutral sites would enter only for v<18S≪1 . Pleiotropy causes sites with a given selection coefficient to have a distribution of effect sizes on the focal trait , thereby increasing the contribution to genetic variance of some sites and decreasing it for others . In Section 3 . 2 of S1 Text , we show that increasing the degree of pleiotropy , n , increases the proportion of genetic variance , G ( v ) , for any threshold , v , regardless of the distribution of selection coefficients ( Fig A5 in S1 Text ) . When variation in a trait is sufficiently pleiotropic for the distribution of effect sizes to attain the limit form ( Eq 11 ) G ( v ) = ( 1+2v ) exp ( −2v ) ( 15 ) for strongly selected sites and G ( v ) =exp ( −4v/S ) ( 16 ) for effectively neutral ones ( Fig 3B and Fig A4b in S1 Text; Section 3 . 2 in S1 Text ) . The intermediate selection range is split between these behaviors: on the weaker end , roughly corresponding to S < 5 , G ( v ) is similar to the effectively neutral case ( Fig A4b and Section 3 . 2 in S1 Text ) ; and on the stronger end , roughly corresponding to S > 5 , G ( v ) is similar to the case of strong selection , with measurable differences only when v ≫ vs ( inset in Fig 3B and Section 3 . 2 in S1 Text ) . We would therefore expect that as the sample size of a GWAS increases and the threshold contribution to variance decreases , intermediate and strongly selected sites ( more precisely , sites with S > 5 ) will be discovered first , and effectively neutral sites will be discovered much later . In S1 Text ( Section 3 . 2 and Fig A3 in S1 Text ) , we also derive corollaries for the distribution of numbers of segregating sites that make a given contribution to genetic variance .
In humans , GWASs for many traits display a similar behavior: when sample sizes are small , the studies discover almost nothing , but once they exceed a threshold sample size , both the number of associations discovered and the heritability explained begin to increase rapidly [4 , 77] . Intriguingly though , both the threshold study size and rate of increase vary among traits . These observations raise several questions: How is the threshold study size determined ? How should the number of associations and explained heritability increase with study size once this threshold is exceeded ? With an order of magnitude increase in study sizes into the millions imminent , how much more of the genetic variance in traits should we expect to explain ? The theory that we developed provides tentative answers to these questions . To relate the theory to GWASs , we must first account for the power to detect loci that contribute to quantitative genetic variation . In studies of continuous traits , the power can be approximated by a step function , where loci that contribute more than a threshold value v* to additive genetic variance will be detected and those that contribute less will not ( Section 6 . 1 in S1 Text ) . The threshold depends on the study size , m , and on the total phenotypic variance in the trait , VP , where v* ∝ VP/m ( Section 6 . 1 in S1 Text; [77] ) ; conversely , the study size m needed to detect loci with contributions above v* is proportional to VP/v* . Given a trait and study size , the number of associations discovered and heritability explained then follow from our predictions for the distribution of variances among sites . When genetic variation in a trait is sufficiently pleiotropic , our results suggest that the first loci to be discovered in GWASs will be intermediate or strongly selected , with correspondingly large effect sizes ( i . e . , S≈2Nnw2a12>5 ) . The functional form of the distribution of variances among these loci ( Eq 15 and Fig 3B ) implies that for GWASs to capture a substantial proportion of the genetic variance , their threshold variance for detection v* has to be on the order of the expected variance contributed by strongly selected sites , vs , or smaller . We therefore expect the threshold study size for the discovery of intermediate and strongly selected loci to be proportional to VP/vs . When the study size exceeds this threshold , the number of associations detected and proportion of variance explained depend on the study size measured in units of VP/vs ( Fig 4 ) and follow from the functional forms that we derived ( Eq 15 and Table A1 in S1 Text ) . The dependence on VP/vs makes intuitive sense , as the total phenotypic variance VP is background noise for the discovery of individual loci whose contributions to variance are on the order of vs . Some results are modified when variation in a trait is only weakly pleiotropic , which is probably less common: notably , the threshold study size for strongly selected loci would be higher , and loci under intermediate selection would begin to be discovered only after the strongly selected ones ( Fig A22 in S1 Text , Eq 13 , and Eq A35 in S1 Text ) . Regardless of the degree of pleiotropy , effectively neutral loci would only begin to be discovered at much larger study sizes , after the bulk of intermediate and strongly selected variance has been mapped ( Fig 4 and Fig A22 in S1 Text ) . Thus , the dependence of the explained heritability on study size is largely determined by VP/vs and by the proportion of heritable variance arising from intermediate and strongly selected loci , whereas the number of associations also depends on the mutational target size , providing a tentative explanation for why the performance of GWASs varies among traits . Importantly , these theoretical predictions can be tested . As an illustration , we consider height and body mass index ( BMI ) in Europeans , 2 traits for which GWASs have discovered a sufficiently large number of genome-wide significant ( GWS ) associations ( 697 for height [16] and 97 for BMI [15] ) for our test to be well powered . We fit our theoretical predictions to the distributions of variances among GWS associations reported for each of these traits , assuming that these distributions faithfully reflect what they would look like for the true causal loci ( see Section 6 . 3 in S1 Text ) . We further assume that these loci are under intermediate or strong selection ( as our predictions suggest ) and that they are highly pleiotropic ( see "Introduction"; [37 , 42 , 45] ) . Under these assumptions , we expect the distribution of variances to be well approximated by a simple form ( Eq A89 in S1 Text ) , which depends on a single parameter , vs . We find that the theoretical distribution with the estimated vs fits the data for both traits well ( Fig 5A ) : we cannot reject our model based on the data for either trait ( by a Kolmogorov-Smirnov test , p = 0 . 14 for height and p = 0 . 54 for BMI; Section 7 . 5 in S1 Text ) . By comparison , without pleiotropy ( n = 1 ) , our predictions provide a poor fit to these data ( by a Kolmogorov-Smirnov test , p < 10−5 for height and p = 0 . 05 for BMI; Fig A14 in S1 Text ) . Fitting the model to GWAS results further allows us to make inferences about evolutionary parameters ( Sections 7 . 1 and 7 . 3 in S1 Text ) . By including the degree of pleiotropy ( n ) as an additional parameter , we find that for both height and BMI , n is sufficiently large for it to be indistinguishable from the high pleiotropy limit . Based on the shape of the distributions in this limit and on scaling the threshold values of v* in units of our estimates for vs , we estimate that the proportion of variance arising from mutations within the range of detectable selection effects is approximately 50% for height and approximately 15% for BMI . Further relying on the number of associations that fall above the thresholds , we infer that , within this range , height has a mutational target size of approximately 5 Mb , whereas BMI has a target size of approximately 1 Mb ( Table A2 in S1 Text ) . These parameter estimates can help to interpret GWAS results . They suggest that , despite their comparable sample sizes , the GWAS for height succeeded in mapping a substantially greater proportion of the heritable variance than the GWAS for BMI ( approximately 20% compared to approximately 3%–5% ) primarily because the proportion of variance arising from mutations within the range of detectable selection effects for height is much greater than for BMI . Moreover , the estimates of target sizes and the relationship between sample size and threshold contribution to variance can be used to predict how the explained heritability and number of associations should increase with sample size ( Fig 5B and 5C ) . These predictions are likely underestimates as the range of detectable selection effects itself should also increase with sample size . We can also examine to what extent our inferences are consistent with data and estimates from earlier studies . For example , the distribution of variances that we inferred for height fits those obtained in a recent GWAS of height based on exome genotyping ( Kolmogorov-Smirnov test , p = 0 . 99; Fig A15b and Section 8 . 1 in S1 Text ) . In addition , the proportion of variance that we estimate to arise from the range of selection effects detectable in existing GWASs for height and BMI is consistent with estimates of the heritable variance tagged by all single-nucleotide polymorphisms ( SNPs ) with MAF > 1% [60 , 61]; Section 8 . 2 in S1 Text . While we have assumed that quantitative traits have been subject to long-term stabilizing selection , recent studies indicate that some traits , and height in particular , have also been subject to recent directional selection [78–82] . Under plausible evolutionary scenarios , recent directional selection can induce large changes to the mean phenotype through the collective response at many segregating loci while having a negligible effect on allele frequencies at individual loci [21 , 83] . This very subtle effect on allele frequencies is likely one reason why polygenic adaptation is so difficult to detect and why studies have to pool faint signals across many loci to do so [78–82] . In Section 5 . 1 of S1 Text , we show that the distribution of allele frequencies on which our results rely is insensitive to sizable recent changes to the optimal phenotype . Importantly then , even when recent directional selection has occurred and its effects are discernable , the genetic architecture of a trait is nonetheless likely to be dominated by the effects of longer-term stabilizing selection . In contrast , recent changes in the effective population size are likely to have had a dramatic effect on allele frequencies and thus on the genetic architecture of quantitative traits [84 , 85] . In particular , European populations in which the GWASs for height and BMI were performed are known to have experienced dramatic changes in population size , including an Out-of-Africa ( OoA ) bottleneck about 100 KYA and explosive growth over the past 5 KY [86–89] . To study how these changes would have affected genetic architecture , we simulated allelic trajectories under our model and historical changes in population sizes in Europeans ( relying on the model of [89]; Section 9 in S1 Text ) . Our results suggest that the individual segregating sites with the greatest contribution to the extant genetic variance have selection coefficients around s = 10−3 and are due to mutations that originated shortly before or during the OoA bottleneck ( Fig 6A and Section 9 in S1 Text ) . These mutations ascended to relatively high frequencies during the bottleneck and minimally decreased in frequency during subsequent , recent increases in population size , thereby resulting in large contributions to current genetic variance . Segregating sites under weaker selection contribute much less to variance because of their smaller effect sizes ( i . e . , for the same reason that applied in the case with a constant population size ) . Finally , and in contrast to the case with a constant population size , individual segregating sites under stronger selection ( e . g . , s ≥ 10−2 . 5 ) contribute much less to current variance than those with s ≈ 10−3 . Mutations at these sites are younger and arose after the bottleneck , when the population size was considerably larger , resulting in much lower initial and current frequencies and therefore a lower per ( segregating ) site contribution to variance ( as distinct from the proportion of strongly selected sites that are currently segregating , which will have greatly increased , resulting in the same total contribution to variance; [84 , 85] ) . In Section 10 in S1 Text , we discuss one implication of these demographic effects: that the reliance on genotyping rather than resequencing in GWASs had a minimal effect on the discovery of associations . Segregating loci with s ≈ 10−3 not only make the largest contributions to the current variance but also are likely to account for most of the GWS associations in the GWASs of height and BMI ( Section 9 in S1 Text ) . When we account for the discovery thresholds of these studies , the expected distribution of variances for loci with s ≈ 10−3 closely matches the distribution observed among GWS associations ( Fig 6B and Fig A20b in S1 Text ) . Moreover , these distributions closely match our theoretical predictions for s ≈ 10−3 and an Ne ≈ 5 , 000 ( Fig 6B ) —roughly the effective population size experienced by mutations that originated shortly before or during the bottleneck . This match likely explains why the results predicted on a constant population size fit the data well nonetheless . Our interpretation of GWAS findings is supported by other aspects of the data ( Section 9 in S1 Text ) . Our conclusions about the high degree of pleiotropy of genetic variation for height and BMI and the differences between these traits are likely robust to demographic effects , given how well our model fits the distributions of variances among loci , once we account for European demographic history . However , we might be underestimating the mutational target sizes and total heritable variances associated with the selection effects currently visible in GWASs , as simulations with European demographic history indicate that the proportion of variance arising from loci with s ≈ 10−3 explained by current GWASs is lower than our equilibrium estimates ( approximately 42% compared to approximately 53% for height and approximately 29% compared to approximately 38% for BMI ) . By the same token , we likely underestimated the future increases in explained heritability with increases in study sizes ( Fig 5B and 5C ) . In summary , a ground-up model of stabilizing selection and pleiotropy can go a long way toward explaining the findings emerging from GWASs . Important next steps involve explicitly using more information from GWASs in the inferences . In particular , we can learn more about the selection acting on quantitative genetic variation by explicitly incorporating information about frequency and effect size ( rather than their combination in terms of variance ) and by including information from associations that do not attain genome-wide significance . Doing so will further require directly incorporating the effects of recent demographic history on genetic architecture [84 , 85] . An extended version of the inference , applied to the myriad traits now subject to GWASs , should allow us to learn about differences in the genetic architectures of traits and answer long-standing questions about the evolutionary forces that shape quantitative genetic variation .
|
One of the central goals of evolutionary genetics is to understand the processes that give rise to phenotypic variation in humans and other taxa . Genome-wide association studies ( GWASs ) in humans provide an unprecedented opportunity in that regard , revealing the genetic basis of variation in numerous traits . However , exploiting this opportunity requires models that relate genetic and population genetic processes with the discoveries emerging from GWASs . We present such a model and show that it can help explain the results of GWASs for height and body mass index . More generally , our results offer a simple interpretation of the findings emerging from GWASs and suggest how they relate to the evolutionary and genetic forces that give rise to phenotypic variation .
|
[
"Abstract",
"Introduction",
"The",
"model",
"Results",
"Discussion"
] |
[
"genome-wide",
"association",
"studies",
"genomics",
"genetic",
"polymorphism",
"phenotypes",
"natural",
"selection",
"genome",
"analysis",
"heredity",
"genetics",
"biology",
"and",
"life",
"sciences",
"population",
"genetics",
"population",
"biology",
"computational",
"biology",
"evolutionary",
"biology",
"quantitative",
"traits",
"evolutionary",
"processes",
"genetic",
"loci",
"human",
"genetics"
] |
2018
|
A population genetic interpretation of GWAS findings for human quantitative traits
|
Identifying effective therapeutic drug combinations that modulate complex signaling pathways in platelets is central to the advancement of effective anti-thrombotic therapies . However , there is no systems model of the platelet that predicts responses to different inhibitor combinations . We developed an approach which goes beyond current inhibitor-inhibitor combination screening to efficiently consider other signaling aspects that may give insights into the behaviour of the platelet as a system . We investigated combinations of platelet inhibitors and activators . We evaluated three distinct strands of information , namely: activator-inhibitor combination screens ( testing a panel of inhibitors against a panel of activators ) ; inhibitor-inhibitor synergy screens; and activator-activator synergy screens . We demonstrated how these analyses may be efficiently performed , both experimentally and computationally , to identify particular combinations of most interest . Robust tests of activator-activator synergy and of inhibitor-inhibitor synergy required combinations to show significant excesses over the double doses of each component . Modeling identified multiple effects of an inhibitor of the P2Y12 ADP receptor , and complementarity between inhibitor-inhibitor synergy effects and activator-inhibitor combination effects . This approach accelerates the mapping of combination effects of compounds to develop combinations that may be therapeutically beneficial . We integrated the three information sources into a unified model that predicted the benefits of a triple drug combination targeting ADP , thromboxane and thrombin signaling .
Cells are subject to diverse stimuli in vivo , and combine these inputs to generate appropriate biological responses . Activators and inhibitors of various targets work together in different configurations to elicit valuable and sometimes unpredictable outcomes , both natural and therapeutically induced . Many therapeutic approaches combine multiple agents acting on different targets , for example in cardiovascular disease[1] , cancer[2–4] , and infection[5] . Ideally , we would have a full systems model of every clinically important signaling process , helping us to predict and define potent combinations . However , in many systems , such a model is largely absent . Accordingly many workers seek to simply study the combination effects without considering additional information regarding the signaling network . Thus , screens for novel agents can take a systematic approach[6 , 7] , but are limited usually to comparing the inhibitor combinations to the effects of single agents , without considering wider aspects of the signaling system . However , the discovery of synergistic effects is not trivial . There is a large set of compounds that target distinct proteins , and considering the pairwise or higher order combinations of all of these is a very substantial task . Accordingly , such screens are frequently performed under a very limited set of experimental conditions . However , in many physiological contexts , cells may be subject to diverse challenges , and it would therefore be ideal for a synergistic combination of drugs to be effective under not just one , but under many alternative conditions . To meet this challenge , systems biology approaches seek to develop integrated computational predictive models of an entire signaling process , and ultimately of a cell , tissue or organism . These models are valuable but often challenging , since their construction requires extensive experimental data , and for this reason they are often developed under relatively limited and controlled settings , such as that of a well characterized cell line . Thus , there is still a requirement to develop more efficient screening methods that by-pass the need for a complete model of a given system , but which capture the essential functional components of that system , as might be relevant in a therapeutic or other practical setting . In order to accelerate the discovery of critical combinations of factors , scientists can either take a bottom-up approach , starting with pairwise combinations and making combinations more complex , or a top-down approach starting with a set of factors and winnowing down the system to the essential components , such as was done to successfully choose 4 transcription factors from 24 that govern the generation of pluripotent stem cells . [8] High intracellular levels of cAMP maintain platelets in a resting state[9] , with prostaglandin I2 ( PGI2 ) and nitric oxide ( NO ) , sustaining the production of cAMP via Gs[10] or limiting its degradation through the cGMP-dependent action of phosphodiesterase III[11] . On the other hand , platelet activators inhibit adenyl cyclase and reduce cAMP via GαI , while βγ subunits of Gi type proteins activate PLC and phosphoinositide 3-kinase ( PI3K ) . The coordinated activity of different types of G proteins is required to modulate platelet behaviour . Platelet activation through G proteins involves Gαi Gαq and Gα12/13[12] , with the thrombin receptor , PAR1 , acting through all three [13–15] and favouring Gαq-mediated calcium mobilization over Gα12/13 signaling when stimulated with thrombin-receptor activating peptide ( TRAP ) [16] . TxA2 receptors couple to Gαq , Gα12 and Gα13 [14 , 17 , 18] . Platelet responses to epinephrine are mediated by the α2A-adrenergic receptors[19] , acting in mice through the Gαi family member Gαz[20] . ADP signalling in platelets , important for sustained aggregation[21] , is via GPCRs P2Y1 ( coupled to Gαq in mice[22] ) , and P2Y12 ( coupled to Gαi2 in mice[20] ) . The activation of GPVI ( the only non-GPCR receptor targeted in our study ) by Collagen or CRP leads to Lyn and Fyn phosphorylation of the FcR gamma-chain[23] , allowing Syk docking[24] and activation of phospholipase C ( PLC ) γ2 [25] and Phosphoinositide 3 kinase ( PI3K ) [26 , 27] . Our goal was to develop efficient and practical methods to identify combinations of platelet inhibitors that would be robust in inhibiting platelets under multiple conditions , and would provide insights into platelet signaling networks . We sought to expand inhibitor combination screening by the incorporation of additional information that might give some insights into the performance of the platelet as a system . The first step in developing our method was to investigate which inhibitors act against which activators[28] . Intuitively combinations of inhibitors are likely to be markedly synergistic when they are acting on parallel pathways . However , it has been shown that under certain feedback conditions , strong synergistic effects will be seen between upstream and downstream points that are located serially along a pathway [7] . Thus , we had no strong expectations of which combinations might show the strongest synergy . We noted that the available consensus that defines the relationships among activators and inhibitors of most signaling systems is frequently based on primary observations that are accumulated in the scientific literature in a piece-meal fashion . Since separate studies may often apply either subtly or grossly different experimental conditions , it is not ideal to simply take the accepted consensus of opinion to pair activators and inhibitors together on the basis of their literature defined targets , but it is of interest to re-evaluate these relationships in a systematic way . The second step in identifying useful combinations was to experimentally evaluate synergistic effects[29 , 30] . Synergy is defined as a functional interaction between two reagents that shows a much greater effect than expected , based on the known effects of the two reagents alone . There are multiple different definitions of what is precisely meant by synergy[31] , and these different definitions may be considered to lie on a spectrum of tests , ranging from weak tests that provide only a suggestion of synergy , and strong tests that provide more robust evidence for such synergy . Typically , the more robust tests rely on the analysis of multiple doses of the two compounds alone and in various combinations . Such synergy studies may rely on analysis of synergies among inhibitors[1 , 6 , 7] . However , synergy studies are not confined to examine synergy among inhibitors , even when inhibition is the primary therapeutic goal . Investigation of synergies among activators[32] can assist in defining the profile of inhibitory effects of single and combination inhibitors , which reduce not only the main effects of the activators , but also provide information regarding their synergistic effects . Since activator-inhibitor relationships , activator-activator synergy and inhibitor-inhibitor synergy each provide insights into the complex network of interacting factors that help in choosing inhibitor combinations , we set out to develop a practical framework integrating all three approaches ( S1 Fig . ) . We integrated this information into a predictive model , and evaluated whether predictions of the model could accelerate the discovery of compound combinations effective at targeting platelet inhibition . This approach predicted a triple combination of compounds that was experimentally validated .
Informed consent was obtained from all subjects for the donation of blood samples for the purpose of platelet function analysis , with study approval obtained from the Royal College of Surgeons in Ireland Research Ethics Committee ( REC679b ) . Experimental methods followed a previous study[33] . Washed platelets were prepared from venous blood of consenting healthy donors drawn via phlebotomy into 15% ( v/v ) acid-citrate-dextrose ( ACD ) anticoagulant ( 38mM citric acid anhydrous , 75 mM sodium citrate , 124mM dextrose ) . Blood was centrifuged at 150 x g for 10 minutes at room temperature and platelet rich plasma ( PRP ) was collected and acidified to pH 6 . 5 with ACD . 1 μM prostaglandin E1 ( PGE1 ) was added prior to centrifuge PRP at 720 x g for 10 minutes . The resulting pellet was resuspended in JNL buffer ( 6 mM dextrose , 130 mM NaCl , 9 mM NaHCO3 , 10 mM sodium citrate , 10 mM Tris base , 3mM KCl , 0 . 81 mM KH2PO4 and 0 . 9 mM MgCl26H2O , pH 7 . 35 ) adjusting the concentration to 3x105 platelets/ μl . Washed platelets were supplemented with 1 . 8 mM CaCl2 immediately prior to the experiment . The ADP release assay used white 96-well plates ( white plates with white wells; Sigma-Aldrich , Ireland ) . Platelets were incubated with inhibitors for 10 minutes at 37°C on orbital slow shake using a Wallac 1420 Multilabel Counter ( Perkin Elmer ) . 10 μl cocktail ( K ) or activators were then added and allowed to activate platelets for 10 minutes in the same conditions used with the inhibitors . 10 μl of the detection reagent Chrono-lume ( Chronolog; Labmedics Limited , UK ) were added and sample luminescence detected after an additional 5 seconds with rapid shaking measuring arbitrary absorbance units ( AAU ) . The compounds used as platelet activators were CRP ( Ca , triple-helical Collagen-related peptide from 0 . 013 to 30 μg/ml; purchased from Dr Richard Farndale , Cambridge , UK ) , U46619 ( Xa , from 0 . 003 to 6 μM; Santa Cruz Biotechnology , Germany ) , TRAP ( Ta , Thrombin Receptor Activator Peptide sequence SFLLRN from 0 . 25 to 16 μM; Sigma-Aldrich , Ireland ) , Epinephrine ( Ea , from 0 . 001 to 30 μM; Chronolog , Labmedics Limited , UK ) , and ADP ( Aa , from 0 . 137 to 100 μM; Chronolog , Labmedics Limited , UK ) . Hill coefficients and response to single agents was evaluated in 4 donors . EC50s and EC90s were determined with GraphPad Prism software , which uses the equation Y = Bottom + ( Top-Bottom ) / ( 1+10 ( LogEC50-% inhibition ) *HillSlope ) . The 2xEC50s were obtained by simply doubling the EC50s . In the case of ADP , to avoid doses higher than 20 μM that might interfere with the assay ( S4 Fig . ) , 10 μ M was used instead of the actual EC50 ( ∼50 μM ) . The letter used to represent each compound denoted the selected dose for each , the letter followed by “2” to denote a dose that is double the selected dose , and the letter followed by “90” to denote a dose that causes the 90% activation ( S1 Table ) . A mother solution of the “activator cocktail” ( K ) , which is all the activators at their selected doses ( 0 . 025 μ M Epinephrine ( Ea ) , 0 . 5 μ M U46619 ( Xa ) , 1 μg/ml CRP ( Ca ) , 4 μM TRAP ( Ta ) , and 10 μM ADP ( Aa ) ) was prepared and serial 1:2 dilutions were used to stimulate platelets . Its EC50 , was found to be 0 . 1636 fold the concentration of the mother solution ( S2 Fig . ) , and this dose was used for cocktail activation in tests of inhibitor synergy . The rationale for choosing this dose was that this was the dose that gave a 50% activation of platelets , which should be relatively sensitive to inhibition by inhibitors or inhibitor pairs: if a higher concentration of the cocktail had been used , it is possible that the platelets would be consistently activated in a way that masked many inhibitory effects or inhibitor combination effects . It is slightly less than the five-fold reduction that would be obtained were the doses to be crudely divided by the number of activators . These doses lie below the individual EC20 values for all five activators ( S2 Fig . ) . To determine inhibitor IC50s , we evaluated ADP release induced by different doses within a range specified in parentheses . Inhibitors used were Wortmannin ( Pi , from 0 . 137 to 100 nM; Sigma-Aldrich , Ireland ) , SQ29548 ( Xi , from 2 . 195 nM to 1 . 6 μM; Enzo Life Sciences , UK ) , BMS200261 ( Ti , from 0 . 000685 to 0 . 5 nM; Sigma-Aldrich , Ireland ) , Yohimbine ( Ei , from 15 . 625 nM to 2 μM; Sigma-Aldrich , Ireland ) , and MRS2395 ( Ai , from 0 . 137 to 100 μM; Sigma-Aldrich , Ireland ) . All were dissolved in water except MRS2395 , which was dissolved in ethanol , where the ethanol proportion was equal to or less than the 0 . 37% of the total volume . Platelets were pre-incubated with the inhibitors and then stimulated with the activator cocktail . Cocktail-stimulated platelets were almost completely insensitive to Wortmannin inhibition and therefore the IC50 for Wortmannin was determined on platelets stimulated with 1 μg/ml of CRP . The 10 consenting healthy donors were all Caucasian between 24 and 42 years of age . Each plate harboured four types of treatments ( single agents , activator/activator combinations , inhibitor/inhibitor combinations , activator/inhibitor combinations ) and two types of controls ( resting and cocktail-activated platelets ) . Two different arrangements of wells were used in order to limit position effects and , since the results for the two plate layouts broadly correlated , a dataset was assembled from 10 consenting healthy volunteers . To account for donor/plate variation , analysis was of the rank within each donor of the observed ADP level . Statistical analysis was performed using STATA version 12 . 0 [34] and the fitting of the final models confirmed using R [35] . Visualisations of data for Fig . 1 and for S3 Fig . ( below ) , were constructed using R[35] . The visualizations were performed using either the basic visualization package or the gplots package in R . The clustering ( S3 Fig . ) was performed using the hclust function of R , which performs hierarchical clustering ( each object is assigned to a cluster , and then the two most similar objects/clusters are joined in one cluster; and so on iteratively until one cluster is created ) . A one-tail Wilcoxon test was used to test the significance of whether activator-activator and inhibitor-inhibitor combinations were superior to either of the double doses of the component reagents . Raw data were converted to logarithms to the base 10 for visualisation . A small number of duplicate treatments within an individual ( ADP for group 1 and Epinephrine for group 2 ) were replaced by their respective means . Main effect terms were held fixed , while interaction terms were added using a forward stepwise multiple regression approach ( adding terms that significantly improved the model , p<0 . 05 ) . The pair-wise interactions were tested by fitting pair-wise interaction terms , along with main effect terms . We present results for synergies of inhibitors ( the two inhibitors together inhibit much more strongly than expected ) or activators ( the two activators activate much more strongly than expected ) ; other significant synergistic interactions were not seen . We defined significant interaction as observation that the double doses of the activators on their own BOTH have significantly less activating effects than the combination in single doses ( two Wilcoxon one-tailed tests with P<0 . 05 for each , Fig . 1 ) . This approach may be beneficial when reagents lack clear dose response relationships[31] . It is equivalent to a limiting case of Loewe additivity , effectively sampling a single point on the isobole when activators have similar potency [30 , 31] . To integrate the three strands of information , we took the significant interactions identified in the double Wilcoxon test for synergy , and the significant activator-inhibitor combination terms identified from the stepwise linear regression modelling . We brought those forward into an integrated model , including the main effects for each activator and inhibitor . The inhibitor-inhibitor and activator-activator testing component of the statistical study design was based on a sequential test , namely to test inhibition combination first against one double dose ( one-tailed test , p < 0 . 05 ) , and then against the second double dose ( second one-tailed test , p<0 . 05 ) . No algorithms are available to calculate the power of this approach . Nevertheless , the study design may be informed by the assumption , when two inhibitors each confer a roughly equivalent effect , that this test is equivalent to a test of the inhibitor combination versus either double dose . Assuming a log ADP intensity of 5 . 2 for a double dose of inhibitor , and 4 . 9 for a dual inhibitor combination ( s . d . = 0 . 2 ) , in order to have 90% power to detect a significant difference ( two-tailed , p< 0 . 05 ) , a sample size of 10 subjects is required . Input , analysis code and output is given in two alternative statistical analysis environments , R and STATA . The same results are obtained using either . The input is the complete analysis dataset presented in the main paper .
Activator-inhibitor combinations are summarized in Fig . 3A , with more detailed plots in Fig . 4 . The expectation was that effects would largely be seen along the diagonal , corresponding to the a priori pairing of activators and inhibitors . In order to make it easier to see to what extent pairings match or depart from that expectation , we adjusted the data for visualisation purposes , where the values represent the mean values in panel A , minus the mean value observed for the single dose activator alone . Two of the combinations strongly match our expectations ( Xa/Xi , and Ta/Ti ) . However , any combinations involving the ADP inhibitor ( Ai ) showed a marked departure from expectation , since its extent of inhibition of ADP activation ( Aa ) was markedly less than that of Ca and Xa ( Fig . 3A and 3C ) . In spite of markedly inhibiting Ca and Xa , Ai did not manage at that same dose to prevent some activation by Aa ( Fig . 3A ) This suggests that it is not acting as a very efficient inhibitor of its intended target , but may be acting via other mechanisms . Overall , epinephrine ( Ea ) had weak activatory effects and its inhibitor yohimbine[36] ( Ei ) had weak inhibitory effects , which may explain why the model did not detect synergies involving this activator-inhibitor pair . It is possible that the doses of epinephrine defined in advance were inappropriate for the particular donors in this study . To evaluate the significance of the observed combination effects , we carried out multiple regression modelling . The regression model was fitted by including a parameter for the main effect for each of the activators and inhibitors . Each additional significant activator-inhibitor combination term ( given a value of 1 if the experiment included both the activator and inhibitor; zero otherwise ) between a particular inhibitor and a particular activator was added as a parameter in a stepwise fashion until no additional significant terms ( p<0 . 05 ) could be added . An initial model that included only activator and inhibitor effects alone explained 68% of the variance ( S2 Table ) . This rose to 73% when specificity of action was considered , by including four additional significant activator-inhibitor combination terms ( S3 Table ) . We considered whether a Boolean representation of activator-inhibitor relationships ( e . g . that inhibitor Ai cancels out entirely the effect of activator Ta ) would model the data adequately . However , a Boolean model of the activator-inhibitor relationships explained less of the variance in the data and provided a significantly poorer fit ( p<10–5; S4 Table ) . Significant inhibition ( Fig . 3B and 3D ) was observed for two activators by the inhibitors normally associated with their receptors ( Ti/Ta and Xi/Xa ) . While GPVI Collagen receptor activation ( Ca ) is thought to be strongly mediated by PI3K [33] , inhibiting PI3K ( with Pi ) had similar effects on Ca as it had on Xa and Ta responses , indicating that Pi is not highly specific for GPVI inhibition , and that its target PI3K may be a convergence point for different signalling routes . Most strikingly , the presumed ADP P2Y12 inhibitor Ai ( MRS2395 ) inhibited other activators ( Ai/Ca; Ai/Ta , and Ai/Xa ) significantly , and more strongly than it inhibited ADP activation . This may be consistent with either a central role for the P2Y12 receptor in mediating signalling via many receptors , or with an alternative target of action of the drug . Regardless of the mechanism of the observed effect , this first strand of evidence highlights the influence of Ai on multiple activators . This suggests that Ai is a promising candidate to include in a set of compounds to inhibit platelets in combination . Significant synergy was defined here as a much greater effect of a combination of two reagents than the double doses of either reagent ( requirement to pass two one-tailed Wilcoxon tests , each with p<0 . 05 ) . While more conservative than other approaches[37] , it avoids statistical difficulties when effect sizes of different reagents are imbalanced , sampled from non-equivalent points on their respective dose response curves , or where reagents do not have standard dose response curves . Activator-activator synergies are summarized in the bottom left triangle of Fig . 3B , and the same observations after adjustment for differences in main effects of activators in the bottom left triangle of Fig . 3D . The detailed results are shown in Fig . 5 . Fig . 3D displays the difference of the activation or inhibition from the most effective double dose of either the first or the second agent within the combination . Two significant activator-activator synergies were identified: activators of the ADP and collagen receptors ( Aa and Ca ) synergised significantly , and activators of the ADP and thromboxane receptors ( Aa and Xa ) synergised significantly . This second strand of evidence suggests that concurrent inhibition of platelets activation elicited by Aa , Ca and Xa may be useful in lowering the activation of platelets in the presence of multiple activators . Again , it particularly points to an important role for the ADP receptor in activation . We tested the effects of inhibitors on the activation of platelets by a cocktail of all five activators , since such a cocktail may be physiologically relevant , and may be more sensitive to inhibitor synergies . The cocktail activation of platelets showed a steep dose response consistent with likely cooperative ( synergistic ) activity ( S2 Fig . ) . We chose a dose of this cocktail that yielded 50% activation ( see Methods ) , intended as a non-saturating combination activator to be used in inhibitor experiments . While it is likely that this cocktail is more dominated by particular activators , it was notable that , while double doses for four of the five inhibitors had difficulty overcoming the activatory effect of this cocktail , eight of the ten inhibitor combinations lowered platelet activation somewhat ( Fig . 3D ) . This indicated that the doses of activators used in the cocktail were showing sensitivity to inhibitor combinations , but much less sensitivity to double doses of single inhibitors . Thus , the dose of cocktail employed in the study appeared to be appropriate for the purpose of detecting synergies among inhibitors , avoiding saturation effects . As before , synergy was defined for each pair of inhibitors whenever the combination of inhibitors had a significantly greater effect than either of the inhibitors in a double concentration ( Wilcoxon p<0 . 05 for both comparisons ) . We observed three significant inhibitor-inhibitor synergies , which involved the pairwise combinations of the inhibitors of Thromboxane Receptor , Thrombin Receptor and PI3K ( Fig . 3B and 3D; Fig . 6; Xi/Ti , Xi/Pi , Pi/Ti ) . This third strand of evidence provides a different perspective from the activator-inhibitor and activator-activator combinations , raising the question of how to reconcile these findings into a single model that makes useful predictions . The goal of anti-platelet therapy is to effectively inhibit platelet activation exposed to multiple challenges . We wished to define what combination of inhibitors would most effectively inhibit platelet activation brought about by several stimuli . In particular , a researcher faced with all the visually displayed information in Fig . 3 would typically find it hard to anticipate what the likely effect of three way combinations might be . Ideally , the different strands of information should be weighted in a sensible way , that is proportional to the degree of evidence supporting each set of data , to predict an outcome of interest to the investigator . To address this , we created an integrated model . The primary data we used in building the model involved pairwise and main effects , but does not provide direct experimental information regarding three-way or higher order synergies . While pairwise synergies are typically the most important [38 , 39] , it is still of interest to investigate further synergy . To combine the three strands of information , we took ( i ) the linear regression model derived from the activator-inhibitor combination analysis , that already included all main effects and four activator-inhibitor combination effects , and added ( ii ) the two significant activator-activator synergy and ( iii ) the three significant inhibitor-inhibitor synergy terms identified above . These parameters were then fitted together in a unified multiple regression model predicting platelet activation . The resulting “integrated model” thus considers simultaneously all the platelet activation data , comprising resting and cocktail activated controls , single doses , and the various combinations of activators and inhibitors ( Fig . 7A; S4 Table ) . As expected , adding the two additional strands of synergy data resulted in a significantly better fit to the data ( p<0 . 0001 , S4 Table ) . Fig . 7A provides a visual representation of the model that can help advance understanding and interpretation of drug combination effects in platelets . We set out to exploit this integrated model to make predictions of the most effective trios of platelet inhibitors . We considered the scenario where a platelet is challenged by all five activators: collagen , epinephrine and activated thrombin , plus ADP and thromboxane release from adjacent platelets , as may occur during coronary arterial platelet plug formation in the presence of a ruptured atherosclerotic plaque . The integrated model ( S4 Table ) was applied to predict the ADP release for each of the 32 ( 25 ) possible three-way combinations of the single dose inhibitors . This enabled us to predict how well each combination could inhibit platelet activation ( S5 Table ) . The most effective predicted combinations all included Ai ( the ADP receptor inhibitor ) . Of these combinations , the most effective trio of inhibitors identified was a combination therapy targeting ADP , thrombin and thromboxane signalling ( Ai , Xi and Ti ) . We experimentally tested whether Ai , Xi and Ti together strongly inhibit the five-activator cocktail . As a comparison , we also considered whether adding a PI3K inhibitor ( Pi ) to Ai and Xi would be as efficient; this acts as a control combination , since the integrated model predicted that it would not result in such a strong inhibition of platelet activation ( S5 Table ) . Fig . 7B indicates that while the Ai/Xi/Ti combination favoured by the model exhibited a marked inhibition of platelet activation , the less favoured Ai/Xi/Pi combination showed much less inhibition ( p = 0 . 0003 ) . This experimental validation of the model indicates that the integration of these three sources of data into a single model can aid in pinpointing higher order effective drug combinations . The model is also useful when trying to determine how much of the pattern of platelet activation in the system remains unexplained , for example by assessing model fit and exploring donor response variability ( See S1 Text ) .
Our method demonstrates that a systematic approach to considering pairwise reagent interactions can lead to the discovery of particular combinations of importance in modulating biological activity , identifying a triple combination of platelet inhibitors that is particularly effective . It is of interest to also integrate our findings with what is known previously of platelet signaling ( Fig . 8 ) , so that we not only identify useful combinations of inhibitors , but also advance understanding of platelet signaling . TXA2R and PAR1 are the only known activators of G12/13 in platelets . PI3K is not a downstream effector of G12/13 and co-activation of both Gi and G12/13 is sufficient to activate platelets[40] . Thus , the synergy of Pi with both Xi and Ti makes sense , as two independent pathways ( G12/13 and PI3K transmitted ) are being targeted in parallel . This suggests that the engagement of both pathways may be required for full activation . By the same logic , since they share a common effector pathway , it is not surprising that there is no significant synergy between Xa and Ta . However , paradoxically , the inhibitors Xi and Ti synergise strongly . This suggests that activation and inhibition states of these two receptors are not simple on-off switches . In endothelial cells TRAP causes the engagement of Gq prior to the engagement of G12/13 [16] . There may be relatively subtle dose dependent effects , such that the spectrum of G12/13 and Gα inhibition by a single versus a double concentration of Ti is not resulting in a balanced increase in the inhibition of both pathways . Alternatively , the difference between the lack of activator synergy and the presence of inhibitor synergy could reflect the presence of more than two conformational states of a receptor being induced by activators and inhibitors . This would be consistent with a multiple state model for the thromboxane receptor studied in a platelet-like cell system [41] where certain inhibitors , including Xi , act as inverse agonists , actually downregulating constitutive activation of the receptor . One explanation for the multiple inhibitory effects seen with Ai ( MRS2395 ) is that it is a “dirty” compound with multiple targets , that is not as efficiently targeting P2Y12 as might be expected . Dirty compounds in principle may have the potential to exhibit multiple synergisms resulting from their diverse targets , but we noted that Ai did not synergize significantly with any of the other four inhibitors . Finally , in our activator-inhibitor screen we observed that while Pi ( Wortmannin ) predictably inhibited Ca ( CRP-induced ) response [26 , 27] , its inhibitory effects were seen across multiple activators , most notably Xa ( U46619-induced ) response , in spite of the fact that the existing literature suggests that TXA2 mediated signalling might not immediately involve PI3K ( Fig . 8 ) . This paradox may potentially be explained by a second wave of signalling and secretion via PI3K following the initial induction of activation [42] . It is also possible that the platelet signalling network is altered in the inhibition experiments by the presence of the three additional activators ( Ea , Ca and Aa ) , thus potentiating the synergy of the two inhibitors . The two most plausible explanations , of alternative receptor states versus alternative network wiring , may not necessarily be mutually exclusive , since alternative receptor states are likely to represent responses to alternative states of the signaling networks either intracellularly or extracellularly . Linear modeling defined the activator-inhibitor effects , and in general such model parameterisation needs to be approached with some care to ensure that statistically sensible parameters correspond to biologically interpretable ones . The linear statistical modeling was then used to integrate the different effects of activator-inhibitor , activator-activator , and inhibitor-inhibitor effects only after synergistic activator-activator and inhibitor-inhibitor effects were predefined in a manner consistent with Loewe isobole analysis , comparing combinations to double doses of both constituents . This avoids some of the dangers of linear modeling in inferring statistically significant synergies under some model which does not correspond robustly to Loewe additivity . Overall , the combined experimental and modeling approach may miss some important interactions that would be detected if we had performed the analysis across the dose response curves of each reagent combination . Given the complexity of platelet signaling , we think it likely that other synergies will emerge at different doses , and with larger sample sizes , or different stimulatory or inhibitory conditions . Nevertheless , we believe our approach is a relatively efficient way of establishing the most critical features of the signaling system , particularly when ensuring that all assays are carried out on the limited material provided by each donor in the study . Statistically , our approach appears relatively robust but clearly is open to further development , in particular moving away from a two-stage analysis ( defining synergy effects separately from activator-inhibitor effects , and then combining these ) . Future models that estimate the synergism simultaneously with the activator-inhibitor effects may increase the efficiency of such studies , and widen the applicability to a wider set of scenarios , for example testing the effects of genetic activatory and inhibitory factors on a phenotype . Integrated modelling of activator-activator , inhibitor-inhibitor and activator-inhibitor combinations may accelerate the discovery of compound and drug combinations that will more effectively target disease states , not only in platelet signalling , but in other potential applications , including cancer therapeutics . Many drugs that are highly successful in the clinic may have a broader mechanism of action than initially hypothesised , often contributing to their clinical efficacy . The systematic approach implemented here provides direct observations of activator-inhibitor relationships that ignores pre-conceived notions regarding the specificity or generality of action of drugs . Thus , in our study , we had prior beliefs concerning the specificity of particular agents in preventing the activation of platelets by certain activators . However , the fact that these pre-conceptions were partly disproved under the particular conditions of our study did not prevent the study design and the computational modelling from identifying a useful triple combination . Clinically used anti-thrombotic regimens provide partial support for the proposed combination identified here , routinely combining inhibition of both ADP and thromboxane signalling[43] . Adding a thrombin receptor inhibitor to these two , as suggested by the integrated model and its experimental validation , is also indicated as a useful three-way combination by a separate study which indicated its apparent synergistic advantages[44] . Clearly , this experimental test of our prediction is relatively limited , considering only two three-way combinations for comparison . Applying modeling to define higher order combinations is likely to be of particular value in experiments with larger numbers of agonists and antagonists , where the number of three-way combinations becomes impractical to screen efficiently . One approach to screening for synergy that has the potential to actually define whether the reagents are acting in serial or in parallel , is to investigate the response profile of synergy derived from investigating the compounds at different concentrations[7] . While our approach cannot resolve whether factors are in serial or in parallel , it does appear to be efficient at identifying interesting combinations . To get a deeper understanding of how the combinations work , they could be studied in combination with analyses of intermediate components in platelet signaling , such as the phosphorylation states of various proteins . Full systems modelling of the dynamics of intermediate signalling factors may more exquisitely and accurately achieve a similar goal to this study , but would need to model the activation states and kinetics of the “hidden” layer of receptors in Fig . 8 , However , this requires collecting quantitative information on the states of these receptors in the presence of multiple combinations of activators and inhibitors . In many clinical contexts such data is difficult to collect , and thus a useful systems model is absent , and may be difficult to develop . Accordingly , synergy modelling integrated with activator-inhibitor combination screens provides a key step in moving beyond the capabilities of current synergy screens[32] . When novel therapeutic inhibitors of blood associated targets are likely to be prescribed in combination with existing therapies , and there are manipulable agonists of the multiple pathways targeted , we advocate initial ex vivo studies to define the combinatorial landscape and make predictions to help in the design of in vivo synergy combination trials in human subjects .
|
Drugs are often used in combinations , but establishing the best combinations is a considerable challenge for basic and clinical research . Anti-platelet therapies reduce thrombosis and heart attacks by lowering the activation of platelet cells . We wanted to find good drug combinations , but a full systems model of the platelet is absent , so we had no good predictions of how particular combinations might behave . Instead , we put together three sources of knowledge . The first concerned what inhibitors act on what activators; the second concerned what pairs of activators synergise together ( having a bigger effect than expected ) ; and the third concerned what pairs of inhibitors synergise together . We implemented an efficient experimental approach to collect this information from experiments on platelets . We developed a statistical model that brought these separate results together . This gave us insights into how platelet inhibitors act . For example , an inhibitor of an ADP receptor showed multiple effects . We also worked out from the model what further ( triple ) combinations of drugs may be most efficient . We predicted , and then tested experimentally , the effects of a triple drug combination . This simultaneously inhibited the platelet’s responses to three stimulants that it encounters during coronary thrombosis , namely ADP , thromboxane and thrombin .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[] |
2015
|
Discovering Anti-platelet Drug Combinations with an Integrated Model of Activator-Inhibitor Relationships, Activator-Activator Synergies and Inhibitor-Inhibitor Synergies
|
HTLV-1 is the causative agent of a severe form of adult T cell leukemia/Lymphoma ( ATL ) , and of a chronic progressive neuromyelopathy designated HTLV-1 associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) . Two important HTLV-1-encoded proteins , Tax-1 and HBZ , play crucial roles in the generation and maintenance of the oncogenic process . Less information is instead available on the molecular and cellular mechanisms leading to HAM/TSP . More importantly , no single specific biomarker has been described that unambiguously define the status of HAM/TSP . Here we report for the first time the finding that HBZ , described until now as an exclusive nuclear protein both in chronically infected and in ATL cells , is instead exclusively localized in the cytoplasm of peripheral blood mononuclear cells ( PBMC ) from patients suffering of HAM/TSP . Interestingly , at the single cell level , HBZ and Tax-1 proteins are never found co-expressed in the same cell , suggesting the existence of mechanisms of expression uncoupling of these two important HTLV-1 viral products in HAM/TSP patients . Cells expressing cytoplasmic HBZ were almost exclusively found in the CD4+ T cell compartment that was not , at least in a representative HAM/TSP patient , expressing the CD25 marker . Less than 1 percent CD8+ T cells were fond positive for HBZ , while B cells and NK cells were found negative for HBZ in HAM/TSP patients . Our results identify the cytoplasmic localization of HBZ in HAM/TSP patient as a possible biomarker of this rather neglected tropical disease , and raise important hypotheses on the role of HBZ in the pathogenesis of the neuromyelopathy associated to HTLV-1 infection .
HTLV-1 is an oncogenic human retrovirus whose infection affects at least ten million people worldwide [1] . HTLV-1 is the pathogenic agent of a severe form of leukemia/lymphoma designated Adult T-cell Leukemia/Lymphoma ( ATL ) characterized by the malignant transformation of CD4+ T cells [2] and of a severe neurological disorder designated HTLV-1-associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) , a chronic progressive neuromyelopathy characterized by spastic paraparesis , sensory dysfunction and sphincter function defects [3 , 4] . Like other retroviruses HTLV-1 produces structural proteins , Gag , Pol and Env , encoded by the plus strand of the viral genome [5] . The HTLV-1 genome expression is mainly influenced by two regulatory proteins , Tax-1 and Rex , encoded by the 3’ region of viral genome between env and 3’ LTR [5] . The viral protein Tax-1 is important for the transcription of the provirus and its oncogenic potential [6 , 7] . The minus strand of the viral genome encodes a transcript whose protein product is designated HTLV-1 bZIP factor ( HBZ ) [8] . It is of note that , while Tax-1 is expressed only in 40% of cells from ATL patients , HBZ transcripts are constantly found in all ATL cells [5 , 9 , 10] . This reflects the fact that HBZ is also important for infectivity and persistence in vivo [11] . HBZ contains a bZIP domain , an activation ( N-terminus ) and a central domain [8] . There are two different isoforms of this protein: a spliced form containing 206 amino acids and an unspliced form with 209 amino acids , leading to proteins that differ in only seven amino acids at their N-terminal AD domains [12 , 13] . The spliced form is more abundant than the unspliced form and is found in almost all ATL patients [14] . HBZ has been described as a nuclear protein , appearing in speckle-like structures [15] . The nuclear localization has been correlated with the presence of several nuclear localization signals present in the basic regions and in the DNA binding domain of the molecule [16] . Localization studies , however have been performed in cell lines often not representative of the real targets of HTVL-1 infection and over-expressing HBZ after transfection with the encoding gene . Experiments using cells transfected with tagged HBZ have shown that HBZ interacts with CREB-2 via its bZIP domain resulting in strong inhibition of the CREB-2/Tax-1 interaction instrumental for the activation of HTLV-1 LTR [8] . Beside the inhibitory effect on the transcription of the viral genome , a series of studies have indicated that HBZ has the capacity to interact with a large array of cellular transcription factors modulating in a positive or negative fashion their biological activity on cell homeostasis [17] . Again , most of these studies have been performed in HBZ-transfected cells raising the possibility that the results obtained might bear limited biochemical and functional relevance . Only recently , by the use of the first reported monoclonal antibody against HBZ , 4D4-F3 , generated in our laboratory , it was possible to investigate for the first time the expression and the biochemical interaction with host factors of endogenous HBZ in HTLV-1 chronically infected cells , in ATL cell lines and , most importantly in fresh PBMCs of an ATL patient [18] . Endogenous HBZ is indeed expressed in speckle-like structures localized in the nucleus , although deprived of aggregates usually seen in HBZ-transfected cells [8 , 19] . A careful quantification of endogenous HBZ let us to show that the viral protein is expressed in the range of 18 . 000–40 . 000 molecules per cell , 20–50 fold less than the amount expressed in HBZ-transfected cells . In chronically infected cells and in ATL , HBZ interacts in vivo with p300 and JunD and co-localizes only partially , and depending on the amount of expressed HBZ , not only with p300 and JunD but also with CBP and CREB2 [18] . The present investigation was set to determine whether we could detect and define by confocal microscopy the subcellular localization of endogenous HBZ in the other major HTLV-1-associated pathology , HAM/TSP . Moreover , the analysis was extended to other cases of ATL and also to infected asymptomatic carriers ( AC ) . Interestingly , and unexpectedly , the analysis of PBMCs from four distinct HAM/TSP patients unequivocally showed that HBZ protein is exclusively localized in the cytoplasm . Nuclear localization of HBZ was instead confirmed in the vast majority of fresh PBMC of ATL patients , whereas we were unable to detect HBZ in cells of AC . In HAM/TSP , the percentage of HBZ-positive PBMC cells ranged between 0 , 4% and 11% . Tax-1 was detected in PBMC of three out of four of these patients , at percentages ranging between 1 to 20% . Interestingly , no co-expression of HBZ and Tax-1 was found in the same cell . Tax-1 was not detected in cells of ATL patients . These results are discussed within the frame of the present knowledge of the pathogenesis of HTLV-1-associated diseases .
In order to define the subcellular localization of endogenous HBZ in HAM/TSP , PBMC of four HAM/TSP patients , namely PH1485 , PH1593 , PH1601 and PH1624 were studied by immunofluorescence and confocal microscopy analysis . In striking contrast with the HBZ nuclear localization in ATL [18] , in all four HAM/TSP patients HBZ localization was confined to the cytoplasm ( Fig 1A ) . Parallel staining with DRAQ5 , a nuclear marker , and with vimentin , a cytoplasmic marker , confirmed the exclusive HBZ cytoplasmic localization . HBZ appeared as discrete dots , similar in shape to the nuclear speckles found in PBMC of ATL patients . The percentage of HBZ positive cells was comprised between 4 to 11% of total PBMC in three of the four patients analyzed with the exception of patient PH1601 having only 0 , 4% HBZ-positive cells ( Table 1 ) . PBMC from HAM/TSP patients were then analyzed for the expression and subcellular localization of Tax-1 protein . Tax-1 was expressed in 1% , 14% and 20% of the PBMC of PH1601 , PH1485 and PH1624 patients , respectively ( Table 1 ) . Tax-1 was undetectable in PH1593 PBMC . In patient PH1485 Tax-1 was localized in dot-like structures in the nucleus ( Fig 1B ) . A proportion of these cells expressed Tax-1 also in the cytoplasm . In patient PH1624 , Tax-1 was localized in the nucleus with a large proportion of these cells expressing Tax-1 also in the cytoplasm . Importantly , no cells were found to co-express HBZ and Tax-1 , as documented in Fig 1C for the PBMC of patient PH1624 displaying a high percentage of both HBZ+ ( 9% ) and Tax-1+ ( 20% ) cells ( see also Table 1 ) . To expand our previous studies in ATL patients , PBMC of two additional patients , PH1393 and PH1505 , were investigated by immunofluorescence and confocal microscopy . In PBMC of both patients , 80% and 83% of the cells were positive for HBZ ( Table 1 ) . HBZ was localized in speckle-like structures in the nucleus ( Fig 2 ) in a very similar fashion as in the previously described PH961 patient [18] . Parallel staining for the nuclear marker DRAQ5 , and for the cytoplasmic marker vimentin , confirmed the exclusive nuclear localization of HBZ . PBMC of the two ATL patients were also analyzed for the presence of Tax-1 protein; none of them showed positivity for this viral protein ( Table 1 ) . Interestingly , PBMC from the three asymptomatic carriers analyzed , PH1614 , PH1619 and PH1621 , did not show any positivity for HBZ ( Fig 3A ) , although they expressed Tax-1 in a small but distinctive proportion of cells ( 11% , 1% and 6% , respectively , Table 1 ) with a preferential , although not exclusive , nuclear localization ( Fig 3B ) . Taken together , these results establish for the first time a striking difference in the subcellular localization of HBZ protein in HAM/TSP patients as compared to ATL patients , showing the unprecedented exclusive cytoplasmic localization of HBZ in HAM/TSP patients . In order to assess whether the cytoplasmic localization of HBZ in PBMC of HAM/TSP patients is a stable feature or dynamic event resulting from a rapid recycling of the protein from the nucleus , PBMC from PH1624 patient were treated with Leptomycin B ( LMB ) , an inhibitor of nuclear export , and analyzed by immunofluorescence and confocal microscopy . Results clearly indicate that the HBZ cytoplasmic localization in PBMC of HAM/TSP is not affected by LMB treatment ( Fig 4A ) . Conversely , as we previously showed [20] this treatment resulted in a prominent nuclear retention of Tax-1 protein in 293T cells transfected with a Tax-1-encoding cDNA ( Fig 4B ) . From these results we conclude that in HAM/TSP patients , HBZ is specifically retained into the cytoplasm and does not shuttle into the nucleus . In order to define the cellular subpopulation expressing the cytoplasmic HBZ protein in HAM/TSP patients , we analyzed in detail the PBMC of patient PH1624 displaying one of the highest number of HBZ-positive cells ( 9% ) . Initial analysis of relevant cell surface markers by immunofluorescence and flow cytometry ( Fig 5 and Table 2 ) showed that this patient expressed CD3 , CD4 and CD8 markers in 88% , 63% and 27% of PBMC , respectively . This phenotype was very similar to the one of PBMC from a normal donor . Interestingly the T cell marker CD25 , known to be expressed in activated as well as in regulatory T ( Treg ) cells was not detected in PH1624 PBMC , whereas was expressed at low amount in 4–5% of normal PBMC . The B cell compartment , as assessed by the presence the CD19 marker , was represented in almost equal proportion of PH1624 ( 10% ) and normal ( 8% ) PBMC . NK cells , as assessed by the CD16 marker , were 4% of PH1624 PBMC and 10% of normal PBMC . HLA class I was expressed in 100% of both PH1624 and normal PBMCs , and HLA class II was expressed in 15% and 18% of PH1485 and normal PBMC , respectively . Subsequently , confocal microscopy analysis was performed . Cytoplasmic HBZ was clearly detected in a significant proportion of CD4+ cells ( Fig 6A , extended field , top panels , and focus on single cell , bottom panels ) . Indeed around 15% of CD4+ T cells were also HBZ+ ( Table 2 ) . Considering that cytoplasmic HBZ was detected in 9% of the total PBMC of PH1624 patient , and that CD4+ T cell represented around 63% of the PBMC in this patient , it derives that virtually all HBZ+ cells are included in CD4+ T cell compartment . Indeed cytoplasmic HBZ+ cells were virtually not detected in the CD8+ PBMC of PH1624 patient ( Fig 6B , top panels , extended fields ) . After careful analysis of more than 100 CD8+ T cells , in fact only one cell was found to co-express HBZ in its cytoplasm ( Fig 6B , middle panels , extended field , and focus on the single positive cell , lower panel ) . Confocal analysis of CD25 expression in PH1624 patient’s PBMC confirmed that cells were negative for this marker ( Fig 7 , upper left panel ) . Similarly , neither CD19+ B cells nor CD16+ NK cells were found to express cytoplasmic HBZ ( Table 2 ) . Thus , in a representative HAM/TSP patient displaying a high percentage of HBZ-positive cells , cytoplasmic HBZ is almost exclusively found in CD4+ T cells and these cells do not co-express the CD25 marker .
Infection by HTLV-1 generally leads to a state of asymptomatic carrier which may last for the entire life . However , in 3–7% of individuals a very severe form of leukemia/lymphoma , ATL , or a chronic progressive neurological disease can develop as result of the infection . Other less frequent diseases are also associated causally with HTLV-1 infection in certain high endemic areas as uveitis in Japan and infective dermatitis in South America , Africa and the Caribbean . The oncogenic progress leading to ATL has been mainly attributed to the viral transactivator Tax-1 that hijacks the basic mechanisms of control of cellular homeostasis [21] . Nevertheless , only 40% of ATL patients can express Tax-1 whereas all of them express HBZ , an event that has been interpreted as an involvement of Tax-1 in the first phases of oncogenic process and of HBZ in the maintaining of leukemic state [11 , 22] . While recent data have unambiguously demonstrated the nuclear localization of endogenous HBZ in ATL cells and in chronically infected cell lines [18] , no data were available on the endogenous expression and subcellular localization of HBZ in patients affected by HAM/TSP . In the present study we demonstrate that a discrete percentage , up to 11% , of PBMC from four HAM/TSP patients express HBZ and that this expression is exclusively confined to the cytoplasm . Cytoplasmic HBZ appears distributed in dots similar to the nuclear speckle-like structures observed in leukemic cells of ATL patients ( [18] and this study ) , sometimes dispersed all over the cytoplasm , sometimes concentrated in a restricted area of it . The relatively low percentage of cells expressing HBZ in otherwise numerically normal PBMC from HAM/TSP patients , and the availability of only small samples of PBMC from the patients under study prevented a biochemical analysis of the molecular basis of HBZ retention in the cytoplasm . Future studies will be concentrated toward this crucial aspect , in conjunction with a more refined characterization of the sub-cytoplasmic compartments where HBZ is located . Within this frame , leukapheresis in selected patients to obtain larger number of cells will certainly be needed for both biochemical and confocal analyses at subcellular level . Nevertheless , several hypotheses can be put forward to help explaining our findings and to orient future studies . For example , two distinct forms of HBZ derived from distinct mRNAs , spliced and unspliced , have been described [12–14] . Spliced and unspliced HBZ display more than 95% sequence homology and diverge only in the first seven N-terminal aminoacids . Unspliced and spliced HBZ should be both present in HAM/TSP and have been found at least at level of mRNA [23] . Although previous studies have suggested that both unspliced and spliced HBZ localize into the nucleus [13 , 15] , particularly in ATL cells , the alternative possibility that one of the two HBZ forms may preferentially distribute in the cytoplasmic region in HAM/TSP cannot be excluded . The 4D4-F3 anti-HBZ monoclonal antibody used in this study was raised against the spliced form of the protein; however , the epitope recognized by the antibody should be present in both spliced and unspliced HBZ , as it maps within the BR1 region , between aa 97–135 [18] . Thus , the 4D4-F3 mAb may not be the appropriate tool to solve this issue . It has been recently described the existence of aminoacid variations in the HBZ protein both in asymptomatic carriers and HAM/TSP patients . These variations were identified in the activation domain and in the nuclear localization signal sequence [24] . Thus , it might be possible that HBZ sequence variation may influence the subcellular localization of the protein . In this regard , it is important to underline that cytoplasmic localization of HBZ in PBMC of HAM/TSP patients was not modified by the treatment of the cells with Leptomycin B , a drug that blocks the CRM1-dependent nuclear-cytoplasmic shuttling of the proteins , strongly indicating that HBZ is a cytoplasmic resident , non migrating protein in HAM/TSP . It should be underlined that we were unable to detect HBZ-positive cells in PBMC of the asymptomatic carriers analyzed in this study . This could be due to the level of HBZ expression under the threshold of detectability of our method or to the real absence of HBZ expression in AC . Compartmentalization of HBZ in specific subcellular cytoplasmic and/or nuclear structures not detectable by our antibody appears unlikely , although cannot be completely excluded a priori . Future studies will clarify this aspect . An additional interesting finding reported in the present study concerns the expression and localization of HTLV-1 Tax-1 at the single cell level . In PBMC of HAM/TSP patients , when detectable , Tax-1 was localized always in the nucleus and in a variable proportion of the cells in both nucleus and cytoplasm . Importantly , it was never expressed in the same cells expressing HBZ . While uncoupling of Tax-1 and HBZ expression is rather common in ATL patients [5 , 9 , 10 , 14] , the mutually exclusive expression of either HBZ or Tax-1 proteins at the single cell level has never been reported . The molecular basis of this event in cells of HAM/TSP patients is unknown at present , and certainly will be the focus of our future investigation . It is tempting to speculate that this finding may be relevant at the functional level and particularly in the immune recognition of HTLV-1 infected cells . The HBZ cytoplasmic localization , possibly in extra-endosolic compartments , may not be appropriate to the generation of peptides that can efficiently bind MHC class I molecules for presentation to , and scrutiny by cytotoxic T cells ( CTLs ) . This may explain the relatively low level of HBZ-specific CTLs in HAM/TSP [25 , 26] in conjunction with the unsatisfactory lytic efficiency of HBZ-specific CTLs with respect to the strong recognition and lytic efficiency of Tax-1-specific CTLs [27] . Additional confocal microscopy analysis of PBMC of a representative HAM/TSP patient PH1624 clearly showed that the major , if not the exclusive , HBZ+ cell subpopulation was represented by CD4+ cells . Indeed around 15% of CD4+ cells expressed cytoplasmic HBZ . Conversely , after counting a large number ( more than 100 cells ) of CD8+ PH1624 cells , only 1 was found to co-express cytoplasmic HBZ . Thus , if from one side this result indicates that CD8+ cells can be infected by HTLV-1 and express HBZ in the HAM/TSP patients , this event is extremely rare as compared to the frequency of HBZ+/CD4+ cells . Absence of HBZ protein expression was also demonstrated in B cells and in NK cells . PBMC of HAM/TSP patient PH1624 did not reveal the presence of CD25+ T cells both by FACS and by confocal analyses . Further refinement by confocal analysis confirmed the absence of CD4+/CD25+ T cells , a subset that includes regulatory T cells ( Treg ) . Although CD4+/CD25+ Tregs can be infected by HTLV-1 and HAM/TSP patients have been shown to have a high number of CD4+/CD25+ Tregs with impaired function ( reviewed in [28] ) as well as HBZ mRNA expression levels comparable to those observed in ATL [29] , the results presented in this paper indicate that in HAM/TSP patients HBZ protein expression can be easily found in CD4+ T cells not displaying the classical phenotype of Treg cells . Additional studies are certainly required to further detail the phenotype and the functional correlate of CD4+ T cells expressing HBZ protein in HAM/TSP patients as well as the possibility that other cells infected by HTLV-1 , such as monocytes/macrophages , may express HBZ protein in their cytoplasm . In conclusion , based on the results presented in this paper we propose that HBZ cytoplasmic localization can be considered as a bona fide biomarker of HTLV-1-derived HAM/TSP pathology . Future studies will be addressed to the assessment of cytoplasmic HBZ localization during HTLV-1 infection and in the follow-up of infected people before they acquire clear signs of pathology , to possibly identify the cytoplasmic HBZ localization not only as a biomarker but also as a predictive element of HAM/TSP development .
We obtained PBMCs from HTLV-1 asymptomatic donors , HAM/TSP patients and ATL patients in the context of a Biomedical Research Program approved by the Committee for the Protection of Persons , Ile-de-France II , Paris ( 2012-10-04 SC ) . All individuals gave informed consent . Patient’s data were analyzed anonymously . Human embryonic kidney 293T cells ( kindly provided by Prof . B . M . Peterlin , UCSF , San Francisco , USA ) were cultured in Dulbecco’s modified Eagle medium ( DMEM ) containing 5 mM L-glutamine and supplemented with 10% fetal calf serum ( FCS ) . Peripheral blood monuclear cells ( PBMC ) from healthy donors , HTLV-1+ asymptomatic carriers , HAM/TSP patients were purified by Ficoll-Paque TM PLUS ( GE-Healthcare Bio-Science , Milan , Italy ) of heparinated blood . PBMC from healthy donors were obtained by the Blood Transfusion Center , Ospedale di Circolo , Fondazione Macchi , Varese , whereas PBMC of HTLV-1-infected patients were obtained through the Biomedical Research Program approved by the Committee for the Protection of Persons , Ile-de-France II , Paris ( 2012-10-04 SC ) . All PBMC preparations were immediately frozen at -80°C and subsequently transferred in liquid nitrogen after 48–96 hours . HTLV-1 infection was confirmed by Western blot on plasma sample and anti-HTLV-1 antibodies in patient’ plasma were titrated through indirect immunofluorescence using the HTLV-1-producing MT-2 cell line ( kindly provided by Dr . B . Macchi , Tor Vergata University , Rome , Italy ) , as described previously [30] . The two ATL patients analyzed in this study , PH1393 and PH1505 , were suffering from a typical acute ATL form characterized by high hyperlymphocytosis with lymphocyte count of 18 . 000/mm3 and 30 . 000/mm3 , respectively , and with 79% and 80% of atypic lymphocytes , as evaluated by optical microscopy . 293T cells cultured on glass coverslips pre-coated with poly-L-lysine were transfected with 0 . 2 μg of plasmid expressing untagged Tax-1 by using FuGENE HD ( 3 μl/μg DNA; Promega , Milan , Italy ) as described previously [20] . Where indicated , the 293T cells or HAM/TSP PBMC were incubated with 20 nM leptomycin B ( LMB; Sigma ) or the vehicle methanol for 3 h at 37°C , 5% CO2 . Frozen vials containing PBMC were thawed by immediate passage from liquid nitrogen to a water bath set at 37°C . Cells were washed with warm RPMI medium and immediately processed for immunofluorescence and flow cytometry analysis or for confocal microscopy , as described [31] . For flow cytometry , the following reagents were used: mouse anti-human HLA class I ( clone B9 . 12 ) ; mouse anti-human HLA class II DR ( clone D1 . 12 ) , both revealed by FITC-labelled rabbit anti-mouse IgG F ( ab’ ) 2 antiserum ( Sigma , Milan , Italy ) ; FITC mouse anti-human CD3 ( clone ( UCHT1 , BD Pharmingen ) ; FITC mouse anti-human CD4 ( clone RPA-T4 , BD Pharmingen ) ; PE-Cy5 mouse anti-human CD8a ( clone RPA-T8; eBioscience , Milan , Italy ) ; PE mouse anti-human CD16 ( clone B73 . 1 , eBioscience , Milan , Italy ) ; FITC mouse anti-human CD19 ( clone HIB19 , BD Pharmingen ) and phyco-erythrin ( PE ) mouse anti-human CD25 ( clone M-A251 , BD Pharmingen ) . For confocal microscopy , appropriate number of cells were cultured on glass coverslips pre-coated with poly-L-lysine ( 0 . 1gr/ml , Sigma ) for five hours . The cells were then washed with 1x PHEM buffer , pH 6 . 9 ( 60 mM PIPES , 25 mM HEPES , 10mM EGTA , 2mM MgCl2 ) three times , fixed in methanol 7 minutes at -20°C , and blocked with 1% BSA in 1x PHEM for 1h at room temperature ( RT ) . Cells were then stained overnight with anti-HBZ 4D4-F3 monoclonal antibody ( mAb ) , anti-Tax-1 mAb ( clone 168 A51-2 from the NIH AIDS Research and Reference Reagent Program ) , anti-vimentin rabbit polyclonal antibody ( Santa Cruz Biotechnology , CA , USA ) , rabbit anti-CD4 monoclonal antibody ( clone EPR6855 , ABCAM ) and anti-CD19 rabbit monoclonal antibody ( clone EPR5906 , ABCAM ) , diluted in PHEM buffer containing 0 . 5% BSA . The slides were then washed five times with cold 1x PHEM and incubated in the dark for 2 h at RT with the following secondary antibodies from Life Technology ( Waltham , MA USA ) : goat anti-mouse IgG1 coupled to Alexa Fluor 546 to detect HBZ , goat anti-mouse IgG2a conjugated to Alexa Fluor 488 to detect Tax-1 , and goat anti-rabbit IgG conjugated to Alexa Fluor 488 or to Alexa Fluor 546 to detect vimentin , CD4 or CD19 . For co-staining with directly labelled antibodies , after extensive washing with 1x PHEM , anti-CD8 rabbit monoclonal antibody directly conjugated to Alexa Fluor 647 ( clone EP1150Y , ABCAM ) and mouse anti-human CD25 monoclonal antibody directly conjugated to Alexa Fluor 488 ( clone BC96 , BioLegend ) were added after the indirect immunofluorescence for two hours at room temperature . Similarly , after indirect immunofluorescence , the nuclei were stained by incubating the cells with DRAQ5Fluorescent Probe ( Thermo Scientific , Waltham , MA USA ) , for 30 min at room temperature . After washing , the slides were mounted on coverslips with the Fluor Save reagent ( Calbiochem , Vimodrone ( MI ) , Italy ) and examined by a confocal laser scanning microscope ( Leica TCS SP5; HCX PL APO objective lenses , 63x original magnification , numerical aperture 1 . 25 ) . Images were acquired and analyzed by LAS AF lite Image ( Leica Microsystem , Milan , Italy ) and/or Fiji ( Image J ) softwares .
|
Currently , more than 10 million people worldwide are infected with HTLV-1 , the first discovered human oncogenic retrovirus . Up to 7% of infected individuals experience during their life a severe form of T cell malignancy or a chronic progressive inflammatory disease of the nervous system designated HTLV-1-associated myelopathy/tropical spastic paraparesis ( HAM/TSP ) . At present , there is no resolutive therapy for both of these diseases . In HAM/TSP patients , besides classical neurological signs and the degree of proviral load , no specific virus-related biomarker has been defined that unambiguously distinguishes infected cells of HAM/TSP from those of asymptomatic carriers or ATL patients . Here for the first time , we present evidence that an HTLV-1 protein , designated HBZ , previously found expressed only in the nucleus , is indeed exclusively localized in the cytoplasm of peripheral blood mononuclear cells of HAM/TSP patients and almost exclusively in the CD4+ T cell compartment without the need that these cells co-express the Treg-associated marker CD25 . This finding establishes an association between development of the inflammatory HAM/TSP disease and presence of a viral product in the cytoplasm , opening new ways to understand the molecular basis of the HTLV-1-mediated pathogenesis of this severe form of neuromyelopathy .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"light",
"microscopy",
"cloning",
"retroviruses",
"viruses",
"cytoplasmic",
"staining",
"rna",
"viruses",
"microscopy",
"confocal",
"microscopy",
"cytotoxic",
"t",
"cells",
"molecular",
"biology",
"techniques",
"immunologic",
"techniques",
"cellular",
"structures",
"and",
"organelles",
"research",
"and",
"analysis",
"methods",
"specimen",
"preparation",
"and",
"treatment",
"staining",
"white",
"blood",
"cells",
"animal",
"cells",
"medical",
"microbiology",
"htlv-1",
"microbial",
"pathogens",
"t",
"cells",
"immunoassays",
"molecular",
"biology",
"immunofluorescence",
"cytoplasm",
"cell",
"biology",
"viral",
"pathogens",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"organisms"
] |
2017
|
Cytoplasmic Localization of HTLV-1 HBZ Protein: A Biomarker of HTLV-1-Associated Myelopathy/Tropical Spastic Paraparesis (HAM/TSP)
|
The p400 E1A-associated protein , which mediates H2A . Z incorporation at specific promoters , plays a major role in cell fate decisions: it promotes cell cycle progression and inhibits induction of apoptosis or senescence . Here , we show that p400 expression is required for the correct control of ROS metabolism . Depletion of p400 indeed increases intracellular ROS levels and causes the appearance of DNA damage , indicating that p400 maintains oxidative stress below a threshold at which DNA damages occur . Suppression of the DNA damage response using a siRNA against ATM inhibits the effects of p400 on cell cycle progression , apoptosis , or senescence , demonstrating the importance of ATM–dependent DDR pathways in cell fates control by p400 . Finally , we show that these effects of p400 are dependent on direct transcriptional regulation of specific promoters and may also involve a positive feedback loop between oxidative stress and DNA breaks since we found that persistent DNA breaks are sufficient to increase ROS levels . Altogether , our results uncover an unexpected link between p400 and ROS metabolism and allow deciphering the molecular mechanisms largely responsible for cell proliferation control by p400 .
Cell fate decisions largely rely on the activation or the repression of specific genetic programs . Proteins , which regulate these genetic programs , are involved in the accurate control of cell fate . Among these proteins , chromatin modifying-enzymes are proposed to play a special role because they can set up epigenetic imprints in chromatin and thus mediate long term and transmissible effects on chromatin function . In mammals , one such protein is the p400 ATPase which is an ATPase of the SWI/SNF family conserved from yeast to human ( it is called SWR1 in yeast and Domino in drosophila ) [1]–[3] . It belongs to a multimolecular complex , which contains other enzymes such as the helicases Tip49a and Tip49b and , at least in mammals and in drosophila , the histone acetyl transferase Tip60 [1] , [4]–[6] . p400 can mediate exchange of histone H2A variants , such as H2A . Z in yeast and mammals and H2Av ( which is a drosophila-specific variant related to both H2A . Z and H2A . X ) in drosophila [4]–[8] . Through this activity , p400 participates in various processes such as DNA double strand breaks ( DSBs ) repair and transcription: in drosophila , Domino exchanges phosphorylated H2Av by unphosphorylated H2Av following completion of DNA repair , leading to the suppression of DNA DSB signalling [5] . Transcriptional regulation by p400 largely relies on H2A . Z incorporation at specific promoters [9] . H2A . Z incorporation can lead both to positive or negative outcome for transcription: whereas removal of H2A . Z is often required for transcription to occur , H2A . Z can also “poise” genes for activation , preventing the propagation of neighbouring repressive heterochromatin [10] . In agreement with this dual effect of H2A . Z in transcription , p400 mediates transcriptional repression of the p21 gene in the absence of DNA damage [11] , [12] but it is also required for transcriptional activation of estrogen-responsive genes upon hormone treatment [13] , both effects being mediated through H2A . Z incorporation [7] . Many results underline the role of p400 and p400-associated proteins in cell fate decisions control . First , p400 was characterized as a protein associated with the viral transforming protein E1A from adenovirus [1] . Moreover , association with p400 was found to be required for E1A to promote cell transformation as well as apoptosis [1] , [14] , indicating that p400 is important for E1A-mediated cell proliferation and cell transformation control . p400 prevents cell cycle arrest in human osteosarcoma-derived cells [12] , inhibits apoptosis in colon carcinoma-derived cells [15] and blocks senescence induction in non transformed human fibroblasts [11] or mouse embryonic fibroblasts [16] . Also , depletion of p400 or of associated proteins ( such as Tip60 ) results in a decrease cell proliferation rate of embryonic stem cells [17] . Altogether , these data point to a critical role of p400 in allowing cell proliferation . The function of p400 in preventing cell cycle arrest or senescence is proposed to be mediated through the direct transcriptional regulation of p21 expression by localized H2A . Z incorporation [7] . However , we show here that p400 depletion can induce oxidative stress suggesting that it may also indirectly activate p21 expression through the activation of DNA damage pathways . By inhibiting these pathways , we show that this indirect mechanism largely account for p21 regulation by p400 as well as for downstream control of cell fate ( such as senescence , cell cycle progression or apoptosis ) . Altogether , our results allow us to decipher the molecular mechanism which accounts for most of the effects of p400 on cell proliferation .
In order to identify genes regulated by p400 , we performed a genome wide analysis of genes affected upon p400 knockdown . We transfected U2OS osteosarcoma cells in duplicates with two control siRNAs and two previously described siRNAs directed against p400 [12] . Silencing efficiency was checked by real time PCR and western blotting ( Figure S1A and S1B ) . Moreover , the p21 mRNA was induced by p400 depletion ( Figure S3B ) , as already described [12] . Forty-eight hours following transfection , we prepared total RNA that we analysed on Affymetrix gene microarrays containing 19 , 734 gene probe sets . We then compared gene expression upon transfection of each anti-p400 siRNA against each control siRNA independently . After normalization , statistical analysis and thresholding ( see Materials and Methods ) , we considered genes as regulated by p400 when they are similarly affected in three out of the four conditions ( p400-1 vs Ctrl-1 , p400-2 vs Ctrl-1 , p400-1 vs Ctrl-2 and p400-2 vs Ctrl-2 ) . 878 genes were identified and , as expected , the p400 mRNA was found decreased upon p400 siRNA transfection whereas the p21 mRNA was found activated . The complete list of modified genes , sorted by the mean of fold change , is shown in Figure S2 . In addition , we validated our microarrays results by analysing several genes by real time PCR ( Figure S3 ) . A Gene Ontology ( GO ) analysis ( see Figure 1 for the eight top singular annotations ) indicates that p400 mainly regulates genes linked to cell proliferation ( the first three singular annotations are composed of genes involved in cell cycle , cell division and mitosis ) . Such regulation is expected and is likely to be indirect , since p400 knockdown affects cell cycle progression [12] . The next singular annotation represents genes involved in the oxido-reduction balance , suggesting a link between p400 and the control of oxidative stress . To test this hypothesis , we directly measured the intracellular levels of ROS ( Reactive Oxygen Species , a major intracellular inducer of oxidative stress ) using fluorescent probes by flow cytometry upon siRNA transfection . We found that the knockdown of p400 leads to an increase of ROS levels ( measured by calculating the mean fluorescence from 25 , 000 cells ) detectable 48 hours and 72 hours following siRNA transfection ( Figure 2A ) . This increase is similar to what is observed in cells treated with H2O2 , an oxidant molecule ( Figure S4A ) , or in cells presenting mutations in the succinate dehydrogenase enzyme , a protein directly involved in the control of the respiratory chain in mitochondria [18] . Moreover , this increase was also observed using 2 other independent p400 siRNAs ( Figure 2B , see Figure S1B for RT-PCR monitoring p400 expression silencing ) , ruling out the possibility of any off-target effects . Thus , altogether , these results indicate that p400 is required to decrease ROS levels in U2OS cells . To confirm the importance of p400 in controlling ROS levels , we used MEFs originating from mice in which the p400 gene has been genetically targeted [19] . We prepared wild type and heterozygous MEFs . Both MEFs grow at identical rates and presented similar basal ROS levels ( data not shown ) . We next treated these MEFs with H2O2 to analyse their response to an oxidative stress increase . We found that the increase in ROS levels following 15 min H2O2 treatment was similar in both genotypes ( Figure 2C ) . To test their ability to cope with this ROS increase , we washed H2O2 and we analysed the decrease in ROS levels . In wild type cells , the decrease in ROS levels was very rapid since they went down to basal levels in less than two hours ( Figure 2C ) . In MEFs in which one allele of p400 had been targeted , the decrease was slower and ROS levels was still largely above basal levels two hours following H2O2 removal ( Figure 2C ) . Thus , the loss of one p400 allele leads to a defective management of ROS levels following a burst of oxidative stress , demonstrating that the role of p400 in controlling ROS levels is not limited to human cells neither to tumoral cells . Taken together , Figure 2A–2C data indicate that normal p400 expression is required to control ROS levels . Deregulated ROS production can lead to DNA damage . We then reasoned that p400 could maintain ROS levels below a threshold which could be detrimental to DNA integrity . To test this possibility , we investigated whether knockdown of p400 in U2OS cells leads to the generation of DNA damage . We performed a neutral comet assay , which directly reveals the amount of DNA strand breaks ( single and double strand breaks ) and alkali-labile sites . We found that the number of cells with detectable comet tails ( thus , with detectable DNA breaks ) was increased upon p400 knockdown ( Figure 2D for typical cells and Figure 2E for the quantification; see Figure S5 for more detailed distribution of comet tail moments ) . Importantly , very similar results were obtained using another independent p400 siRNA ( Figure S6 ) . We thus conclude from this experiment that p400 expression is required for genetic integrity and that p400 prevents both increased ROS production and accumulation of DNA damage . In order to test whether DNA damage induced by the absence of p400 are due to oxidative stress , we used N-Acetyl Cysteine ( NAC ) , a widely used anti-oxidant reagent . As expected , NAC treatment efficiently reversed the increase in ROS levels induced by p400 knockdown ( Figure 3A ) . Importantly , NAC treatment also reversed the increase in comet tail moments upon p400 knockdown ( Figure 3B , see also Figure S6 with an independent p400 siRNA ) . Thus , these data indicate that p400 prevents oxidative stress–induced DNA damage . We next tested whether p400-induced DNA damage can lead to the activation of a DNA Damage Response ( DDR ) and then to a cellular response . We found that knockdown of p400 in U2OS cells is sufficient to induce a significant increase of cells harbouring γH2AX foci ( Figure 3C and 3D ) , a widely used marker of DNA damage signalling . It also leads to an increase in autophosphorylation of the sensor kinase ATM ( reflecting ATM activation ) , measured either by immunofluorescence or by western blot ( Figure 4A and 4B ) . In addition , phosphorylation of downstream substrates of ATM , such as the p53 tumour suppressor , was induced ( Figure 4B ) . Thus , p400 knockdown results in the activation of ATM-dependent DNA damage response pathways . Strikingly , NAC treatment partially reversed the appearance of γH2AX foci ( Figure 3C and 3D ) thereby indicating that DNA damage signalling induced by p400 siRNA is mediated through increased oxidative stress . Thus , altogether , Figure 3 and Figure 4 results indicate that p400 prevents the induction of DNA damage and of ATM-dependent DNA damage signalling by oxidative stress . We next intended to investigate the contribution of these ATM-dependent DNA damage response pathways to cell proliferation control by p400 . As already reported [11] , [12] , we found that p400 knockdown induces the activation of the gene encoding the p21 cell cycle-dependent kinase inhibitor ( Figure 4C ) , as well as the concomitant accumulation of cells in the G1 phase of the cell cycle ( Figure 4D ) . Importantly , the two fold induction of p21 mRNA induction , as well as the extent of cell cycle arrest we observed here , is within the range of what has been found previously in U2OS cells by others and us [11] , [12] . p400-mediated repression of the p21 promoter was proposed to be direct and to rely on the targeted incorporation of H2A . Z on the p21 promoter [7] . However , it may also be an indirect consequence of ATM-dependent DDR pathways repression by p400 since the p21 promoter is a direct target of the DNA damage-activated p53 tumour suppressor . To assess the relative contribution of these two mechanisms ( direct repression through H2A . Z incorporation ( as demonstrated by [7] ) and control of ATM-dependent DNA damage pathways ( as we observed here ) ) , we intended to inhibit these DNA damage pathways using a siRNA directed against ATM . Transfection of the ATM siRNA efficiently inhibits ATM expression and does not affect silencing of p400 by the p400 siRNA ( Figure S1A ) . The ATM siRNA by itself does induce only a slight decrease , if any , of p21 mRNA expression ( Figure 4C ) . However , it decreases the activation of p21 mRNA and protein expression induced by p400 depletion ( Figure 4C ) . Moreover , it partially reversed the concomitant cell cycle arrest ( Figure 4D ) . Importantly , activation of p21 mRNA ( Figure 4E ) and cell cycle arrest ( Figure 4F ) induced by Nutlin-3 , an inhibitor of the p53/Mdm2 interaction [20] , is not affected by ATM knockdown , indicating that ATM is not generally required for the p53-dependent activation of p21 mRNA expression and cell cycle arrest and then , that the effect of ATM is upstream of p53 activation . Thus , taken together , these data indicate that p400 represses p21 expression , at least in part , through the control of DNA damage pathways . Note that the residual activation of p21 mRNA and G1 accumulation induced by p400 depletion in the presence of ATM siRNA , is probably due to directs effects of p400 on the p21 promoter [7] . Strikingly , other proposed roles of p400 in cell proliferation ( repressor of apoptosis in HCT116 cells [15] and repressor of senescence in IMR90 cells [11] ) could also be dependent on its ability to modulate ATM-dependent DNA damage pathways . First , we checked whether p400 also controls ATM-dependent DDR pathways in these cells by transfecting them with a p400 siRNA , which decreased p400 mRNA or protein levels as expected ( Figure S7 ) . We observed by immunofluorescence staining that p400 knockdown induces ATM phosphorylation both in IMR90 cells ( Figure 5A ) and HCT116 cells ( Figure 5B ) . Thus the ability of p400 to prevent activation of DDR pathways is not restricted to U2OS cells . We next investigated the involvement of ATM-dependent DDR pathways activation in cell fate control by p400 . To this aim , we co-transfected cells with the siRNAs against p400 and ATM in HCT116 and IMR90 cells . In these two cell types , transfection of the ATM siRNA efficiently inhibits ATM expression and does not affect silencing of p400 by the p400 siRNA ( Figure S7 ) . As already demonstrated [11] , transfection of p400 siRNA leads to senescence of IMR90 cells , as observed by the appearance of the so-called SAHF ( Senescence-Associated Heterochromatin Foci ) ( Figure 5C ) , the induction of Senescence–Associated β-Galactosidase activity ( Figure 5D ) and induction of p16 mRNA expression ( Figure 5E ) . Depletion of ATM does not have any effect by itself on senescence induction . However , depletion of ATM completely reverses the induction of senescence by the transfection of p400 siRNA ( Figure 5C–5E ) , indicating that ATM expression is required for p400 knockdown to induce senescence . Similarly , we found that ATM expression is also required for p400 knock-down-induced apoptosis in HCT116 cells ( Figure 5F ) . Thus taken together , these data indicate that cell fate control by p400 is largely dependent on its ability to prevent activation of ATM-dependent DNA damage pathways . We next addressed the mechanism by which p400 knockdown leads to ROS production . Strikingly more and more evidence suggests that persistent DNA damage ( due for example to defective DNA repair pathways ) induces an oxidative stress [21]–[24] . Since p400 depleted cells present an increase in DNA breaks ( see Figure 2D and 2E ) , we tested whether this increase in DNA damage could participate in ROS production . To test whether DNA breaks can induce ROS production in U2OS cells , we used a cell line derived from U2OS cells in which we can induce the nuclear localisation of the AsiS1 restriction enzyme by 4-hydroxy-tamoxifen ( OHTam ) treatment . When localized in the nucleus , this restriction enzyme generates a large number ( about 200 ) of pure double strand breaks ( DSBs ) [25] ( in contrast to more widely used methods , such as ionizing radiations , which directly produce free radicals ) . OHTam treatment generated nuclear accumulation of the restriction enzyme and efficiently induced DNA DSBs , as indicated by the appearance of γH2AX foci ( Figure 6A ) . DNA breaks induction is very rapid since increased γH2AX staining can be detected as early as 15 min following OHTam addition ( data not shown ) . Moreover , very high levels of γH2AX staining are still observed 48 hours following OHTam addition demonstrating that the DNA DSBs persist up to 48 hours of treatment ( most likely because restriction sites are permanently repaired and re-cleaved ) ( Figure 6A ) . We then treated , or not , these cells with OHTam for 4 or 48 hours and we measured ROS levels . Whereas no change could be detected 4 hours following OHTam , ROS levels were strongly increased after 48 hours of OHTam treatment , reaching a level similar to the one measured 48 hours after p400-targeting siRNA transfection ( Figure 6B ) . Importantly , no increase in ROS production could be observed in parental U2OS cells treated with OHTam ( data not shown ) , indicating that ROS production is indeed due to the generation of DSBs by the restriction enzyme . These results indicate that in U2OS , whereas the presence of DNA DSBs per se is not able to induce a detectable increase in ROS levels ( as indicated by the results obtained 4 hours following OHTam treatment ) , their persistence for up to 48 hours is sufficient to induce such an increase . Thus , the presence of persistent DNA damage in cells depleted by p400 could participate in the increase in oxidative stress . However , such a mechanism does not fully explain our results since we found that DNA damage induction by p400 is partially blocked by anti oxidant treatment ( indicating that DNA breaks induction is , at least partly , a consequence and not a cause of oxidative stress increase ) . Moreover , we found that p400 expression is important to restore normal ROS levels upon exogenous oxidative stress increase at time points ( 4 hours , see Figure 2C ) at which DNA damages do not increase ROS levels ( see Figure 6B ) . Thus , we investigate whether some of the effects of p400 on ROS levels control could be transcriptional . Specific inspection of the microarrays results indicates that p400 indeed regulates many genes whose products are known to be involved in the control of ROS levels , in such way that it could favour an increase in ROS levels ( such as Hsp70 [26]–[28] , FANCA [29] , [30] or Lamin B1 [31] ) . To investigate the importance of transcriptional control in the effects of p400 on ROS levels , we focused on Hsp70 and FANCA . Indeed , knocking-down Hsp70 expression induces oxidative stress in many cell type , and oxidative stress management is defective in cells from the Fanconi anemia group A [29] , [30] . Moreover , we observed , in our microarrays analyses , that FANCA mRNA as well as two mRNAs coding for Hsp70 protein ( Hspa1a and Hspa1b ) ( collectively referred as “Hsp70” thereafter since they encode identical proteins [32] ) are decreased upon p400 knockdown ( Figure S2 ) . We first checked whether Hsp70 and FANCA are bona fide target genes of p400 . We transfected U2OS cells with two independent p400 siRNAs and analysed Hsp70 and FANCA mRNA expression by RT-QPCR . We confirmed that Hsp70 and FANCA mRNA levels are decreased in cells transfected by p400 siRNA , indicating that p400 positively regulates their expression ( Figure S3C ) . Moreover , we also found that p400 knockdown decreases Hsp70 and FANCA protein expression ( Figure 7A ) . We next tested whether these two genes can be direct transcriptional targets of p400: indeed , p400 is known to favour transcription through binding to specific promoters , such as some controlling genes involved in maintenance of embryonic stem cells [17] or estrogen-responsive genes [13] . Thus , to test whether p400 can directly regulate the promoter of Hspa1a ( a gene coding for Hsp70 protein ) and of FANCA , we performed Chromatin ImmunoPrecipitation ( ChIP ) experiments with p400 antibodies using chromatin from U2OS cells . As expected , we could detect p400 binding to the p21 promoter ( a known direct target of p400 [11] ) ( Figure 7B ) . This binding is specific , since an unrelated sequence derived from the ribosomal phosphoprotein P0 promoter was only marginally enriched in the p400 immunoprecipitates ( IP ) . Strikingly , we found that sequences derived from the Hspa1a gene promoter and from the FANCA promoter are also enriched in the p400 IP ( with an efficiency comparable to the p21 promoter ) ( Figure 7B ) , indicating that p400 physically binds to these two promoters . Taken together , these results indicate that the Hspa1a and FANCA promoters are directly regulated by p400 . Direct regulation of promoter activity by p400 often involves incorporation of the H2A . Z variant . To test the involvement of H2A . Z incorporation in p400-mediated regulation of these promoters , we depleted H2A . Z expression using a specific siRNA ( see Figure S1C for RT-PCR and western blot showing the efficiency of H2A . Z depletion ) . We found that this depletion leads to a decrease in Hsp70 mRNA expression ( Figure 7C ) but not of FANCA mRNA expression ( Figure 7D ) , suggesting that p400 regulates these two promoters by independent mechanisms . We next tested whether transcriptional regulation can be important for p400 to control intracellular ROS levels . To this aim , we transfected together with the p400 siRNA , an expression vector for Hsp70 and FANCA to prevent the decrease of their mRNA . Such an experiment is feasible since transfection efficiency for plasmids routinely reached from 50 to 80% in U2OS cells ( data not shown ) . In addition , RT-PCR measurement of mRNA levels indicated that , in cells transfected by the expression vector , the p400 siRNA efficiency was unchanged ( data not shown ) . We next measured ROS levels 48 hours following transfection . We found that ectopic expression of Hsp70 or FANCA have , by themselves , no effect on ROS levels in U2OS cells ( Figure 8A ) . However , they significantly inhibited the increase in ROS levels induced by p400 knockdown ( Student t-test p<10−5 on three independent experiments ) . This result indicates that , if the down regulation of Hsp70 or FANCA is prevented , p400 siRNA transfection induces ROS production less efficiently . Thus , regulation of Hsp70 and FANCA expression is required for p400 to control ROS production . Finally , we transfected U2OS cells with siRNAs directed against Hsp70 and FANCA to test whether we can recapitulate the effects of the p400 siRNA on ROS production . Both siRNAs inhibited their targets as shown by reverse transcription followed by Q-PCR or western blotting experiments ( Figure S1D and S1E ) . We next measured their effects on ROS production and we found that , whereas inhibition of Hsp70 has no effect by itself , transfection of FANCA siRNA induces ROS accumulation ( Figure 8B ) . Thus , this result indicates that FANCA is certainly a critical target by which p400 controls oxidative stress . In agreement with this hypothesis , H2A . Z inhibition , which does not affect FANCA expression ( Figure 7D ) , does not induce any increase in ROS levels ( Figure S8 ) . Defects in Hsp70 expression , although probably not causal in oxidative stress induction , likely participates in the defective response to oxidative stress in p400-depleted cells , since activation of Hsp70 mRNA expression upon acute oxidative stress in MEFs is abrogated upon loss of one p400 allele ( Figure S9 ) . Taken together , these results indicate that the p400-mediated control of ROS levels and cell proliferation is brought about , at least in part , through the transcriptional regulation of specific promoters , including the FANCA and Hsp70 promoters .
In this manuscript , we first demonstrate that p400 plays a major role in the control of ROS metabolism since a decrease of p400 level is sufficient to induce a ROS imbalance in U2OS cells . Consequently , p400 expression is required to maintain ROS levels below a threshold at which DNA damage are induced and the DNA damage response ( DDR ) activated , and is important for ROS homeostasis upon a burst of oxidative stress . We further provide important information on the mechanism by which p400 exerts its anti-oxidant functions: First , we demonstrate that part of the mechanism is transcriptional: indeed , some genes known to be involved in the control of ROS levels are positively regulated by p400 such as the genes coding for FANCA , Hsp70 and Lamin B1 . Moreover , the regulation of some of these proteins ( FANCA and Hsp70 ) is required for p400 to control ROS levels . It is likely that the initial increase in ROS production upon p400 knock-down is related to transcriptional defects in the expression of genes involved in ROS metabolism such as FANCA . Indeed , p400 depletion leads to a decrease of FANCA levels to an extend at which FANCA knockdown induces ROS ( Figure 8B ) . It is thus likely that the transcriptional defects , in the absence of p400 , lead to a deficiency in ROS removal and thus , due to the continuous ROS production in living cells , in an increase in ROS levels and oxidative stress . This increase will then lead to the induction of DNA damage , since we found that DNA damages induction is largely reversed by anti-oxidant treatment ( Figure 3B ) . Interestingly , our results suggest that the continuous presence of DSBs per se is sufficient to induce ROS levels . The mechanism by which persistent DSBs induces an oxidative stress is not known but may be related to an elevated cellular metabolism to achieve DNA repair . In agreement with this hypothesis , the existence of a link between defects in DNA damage repair pathways and increased ROS production is now becoming more and more clear [21]–[24] . Whatever the mechanism , the presence of persistent DSBs in the absence of p400 may contribute to the observed increase in ROS levels . These DSBs can be continuously created or may persist because unrepaired . Strikingly , some p400 homologues in other species ( SWR1 in yeast and domino in drosophila ) are known to participate in DNA DSB repair . Moreover , we identify here FANCA , a protein important for DNA repair , as a direct target gene of p400 . Finally , the GO analysis of our microarrays data of p400 target genes has identified DNA repair as a major singular annotation ( See Figure 1 ) . Therefore , it is tempting to speculate that when p400 levels are low , unrepaired DSBs accumulate and the persistence of these DNA breaks leads to ROS increase . Such a mechanism cannot fully explain by itself our results since , as noted above , antioxidant treatment largely reversed DNA breaks ( which would not be the case if DNA breaks would be responsible for increased ROS production ) and since p400Mut/+ MEFs exhibit a defective ROS response compared to wild type MEFs at time points at which DNA breaks do not induce ROS accumulation . Taken together , our results indicate that the transcriptional defects in p400-depleted cells decrease ROS metabolism leading to an initial ROS increase . This initial ROS increase leads to the appearance of persistent DNA breaks , which in turn favour continuous ROS production , which are incorrectly managed in the absence of p400 , resulting in the apparition of a positive feedback loop inducing oxidative stress and the downstream cellular response ( see our model in Figure 8C ) . p400 has been shown to play a critical role in cell fate decision since it strongly favours cell proliferation by preventing senescence induction [11] , cell cycle arrest [12] and apoptosis induction [15] . Our data here indicate that inhibition of these anti-proliferative states by p400 largely relies on the same molecular mechanism , which is the control of intracellular ROS levels: upon p400 depletion , ROS levels are increased leading to endogenous oxidative stress . This stress is high enough to induce DNA damage and activation of the ATM-dependent DNA Damage Response . Depending on the cell type , activation of this DDR following p400 knockdown will lead to induction of senescence ( normal cells such as IMR90 cells ) , cell cycle arrest or apoptosis ( tumoral cells ) . Importantly , although cell fate control by p400 was originally shown to be largely dependent on the p53 tumour suppressor [11] , [12] , it is now clear that the effects of p400 on senescence or apoptosis can also be p53-independent [15] , [16]: these latter effects could be mediated by p53-independent DDR pathways . Also , our findings could explain why p400 and its associated protein Tip60 have been described as antagonists [12] , [15] . Indeed , Tip60 is largely required for DNA damage Response [33] , at least in part through ATM acetylation [34] , [35]: inactivation of Tip60 would abolish any effects due to DDR activation following p400 depletion , as we described for p21 activation and accumulation of cells in G1 [12] or for apoptosis induction [15] . Of note , Gevry et al . proposed that p400 represses p21 promoter through the targeted incorporation of H2A . Z variant and that this repression is lost upon DNA damage induction [7] . Although we were able to confirm that p400 binds to the p21 promoter ( Figure 7B ) and that H2A . Z is enriched on the p21 promoter ( data not shown ) , we show here that p400 can repress p21 transcription indirectly: indeed , in U2OS cells , p400 knockdown induces activation of the ATM-dependent DDR pathway and phosphorylation of p53 . Moreover , induction of ROS to levels comparable to those observed upon p400 depletion is sufficient to induce activation of p21 mRNA expression ( Figure S4B ) . To distinguish between direct ( through regulated H2A . Z incorporation ) or indirect ( through the control of the DDR pathways ) repression of p21 promoter by p400 , we inhibited the DDR pathway and we found that we can partially relieve the effects of p400 knockdown . Thus , the integrity of the DDR pathway is required for full p21 activation following p400 knockdown . In contrast , the DDR pathway is not required for full p21 activation by p53 in the absence of DNA damage ( by the Mdm2 inhibitor Nutlin-3 , see Figure 4E ) . Taken together , these data demonstrate that regulation of the p21 promoter by p400 involves , at least in part , control of the DDR pathways upstream of p53 transcriptional activity . Interestingly , such a mechanism is consistent with our previous finding that p400 depletion does not further activate p21 expression upon full induction of DDR pathways through genotoxic treatments [12] . Note however that , because ATM inhibition does not totally relieve p21 activation by p400 knockdown , we cannot rule out the possibility that H2A . Z incorporation participates in the repression of p21 expression by p400 in U2OS cells . Here we identify Hspa1a ( a gene encoding Hsp70 protein ) and FANCA as direct target genes of p400 . Hspa1a and FANCA can thus be added to known target genes of p400 which include some E2F and c-myc target genes [36] , [37] , the p21 cell cycle inhibitor [11] , [12] , estradiol-receptor target genes [13] and genes required for pluripotency maintenance in ES cells [17] . Hsp70 plays a major role in adaptation to various types of stresses [32] . p400 , through the regulation of Hsp70 expression , may also be generally involved in stress response . Indeed , cells deficient for both Hsp70 proteins are more susceptible to UV , osmotic stress , ischemia , and heat [32] . It would be of particular interest to investigate whether p400 itself is a target of stress-induced pathways . Strikingly , cancer cells are subjected to various types of stress and increased Hsp70 levels is a common feature of cancer cells [38] . This increase is believed to help cancer cells to cope with the various stresses they encounter , including therapy-induced stress . Therefore , Hsp70 is an increasingly popular potential therapeutic target . The results we present here suggest that deregulation of p400 function may also favour cancer progression and resistance to anticancer treatments through the control of stress-response pathways . In agreement with such a hypothesis , we recently showed that the siRNA-mediated decrease of p400 levels favours the response to 5-fluorouracil of colon cancer cells [15] . Our data highlight the importance of studying p400 expression in human cancer and confirms that p400 could be a promising therapeutic target .
The p400 antibody was purchased from Abcam ( Paris , France ) , the anti-HDACs ( which recognizes HDAC1 , 2 and 3 ) from Transduction Laboratories ( Lexington , KY ) , the anti-HA from Covance ( Madison , WI ) , the anti-γH2AX from Upstate Biotechnologies ( Millipore , Inc . Billerica , MA ) , the anti-GAPDH antibody from Chemicon International , Inc ( Temecula , CA ) , the anti-ATM and the anti-phospho-ser15 p53 from Calbiochem ( EMD Chemicals , Inc . Darmstadt , Germany ) , the anti-phospho ATM from Cell Signalling technology , Inc ( Boston , MA ) , the anti-H2A . Z and anti-FANCA from Abcam Inc ( Cambridge , MA ) , the anti-Hsp70 from StressGen ( Ann Harbor , MI ) and the anti-HA ( Y-11 ) and anti-β-actin ( C-2 ) from Santa Cruz Biotechnology Inc ( Santa Cruz , CA ) . All secondary antibodies were purchased from Amersham ( Piscataway , NJ ) . Hsp70 expressing plasmids and the corresponding empty vector were kind gifts from Dr Claire Vourc'h . FANCA expressing plasmid was a kind gift of Dr Filippo Rosselli ( IGR , Villejuif , France ) . All siRNAs were purchased from Eurogentec . The control siRNA does not recognize any human mRNA . The sequences of the top strands of the various siRNAs were as follows: Ctrl: CAUGUCAUGUGUCACAUCU-dTdT p400-1: UGAAGAAGGUUCCCAAGAA-dTdT p400-2: CAUCCACAUAUACAGGCUU-dTdT p400-3: CGACACAUUGGAUACAGAA-dTdT Hsp70: GCGAGAGGGUGUCAGCCAA-dTdT ATM: GCCUCCAGGCAGAAAAAGA-dTdT H2A . Z-1: GUAGUGGGUUUUGAUUGAG-dTdT H2A . Z-2: AAAGGACAACAGAAGACUG-dTdT FANCA: AAGCTGTCTTCCCTGTTAGAGTT-dTdT The efficiency of siRNAs silencing was checked in each experiments by reverse transcription followed by real time PCR as described [12] . The following primer pairs were used to amplify cDNAs following reverse transcription experiments ( from 5′ to 3′ ) : p400: CTGCTGCGAAGAAGCTCGTT and CAATTCTTTCCCTCTCCTGC ATM: ACCACACAGGAGAATATGGA and CTCTGCAGTAATGTATTACACA p21: GTCAGAACCGGCTGGGGATG and TGAGCGAGGCACAAGGGTAC p16: CTGCCCAACGCACCGAATAG and ACCACCAGCGTGTCCAGGAA Hsp70: ACCAAGCAGACGCAGATCTTC and TCGGCCAAGGTGTTGGCGTCC H2AZ: CCTTTTCTCTGCCTTGCTTG and CGGTGAGGTACTCCAGGATG FANCA: CCAGCGTGATGTTATATCGG and CAAGGAATCCCTCGTCCTAC GAPDH: GAAGGTGAAGGTCGGAGTCA and GAAGATGGTGATGGGATTTC The following primer pairs were used to amplify promoters following ChIP experiments ( from 5′ to 3′ ) : P21: GTGGCTCTGATTGGCTTTCTG and CTGAAAACAGGCAGCCCAAGG Hspa1a: CCGACCCTTCCTGTCAATTA and TTCCTTGGACCAATCAGAGG FANCA: GTCGTGGCCATGTTGGTC and CTTCAGGACCAACCCCAGT P0: GGCGACCTGGAAGTCCAACT and CCATCAGCACCACAGCCTTC Culture products were purchased from Invitrogen ( Carlsbad , CA ) . The colorectal carcinoma cell line HCT116 , the human normal lung fibroblasts IMR90 and the osteosarcoma cell line U2OS were purchased from the ATCC collection . HCT116 and U2OS cells were cultured in Dulbecco's modified Eagle's medium ( DMEM ) and IMR90 in Modified Eagle's medium ( MEM ) supplemented with antibiotics , 10% FCS and non-essential amino acids ( for IMR90 ) . The construction of U2OS-HA-AsiSI-ER cells is described in [25] . MEF cells were prepared from p400Mut/+ mice [19] using E11 . 5 embryos . Dissected tissues were subjected to trypsinization and dissociation using syringe before plating cells in Petri dishes containing DMEM medium supplemented with 10% FCS , Penicillin/Streptomycin cocktail and 10 µM β-mercaptoethanol . MEF cells were then genotyped using RedExtract-N-Amp Tissue PCR kit ( Sigma , Saint Quentin Fallavier , France ) and used in experiment or maintained in culture for a maximum of five passages . H2O2 , N-Acetyl-Cysteine ( NAC ) and Nutlin-3 were purchased from Sigma and were added on cells at 10 mM , 10 mM and 20 µM , respectively . When needed , U2OS-HA-AsiSI-ER cells were treated with 300 nM 4-OHTam for 4 h or 24 h . For transfection , cells were electroporated with siRNAs or plasmids using an electroporation device ( Amaxa AG , Köln , Germany ) , according to manufacturer's specifications . Nuclear extracts or total cell lysates were prepared as previously described [15] . 10 to 50 µg of proteins per lane were separated by NuPAGE Novex 3-8% Tris-acetate gel ( Invitrogen ) . Proteins in the gel were transferred on a PVDF membrane . Primary antibodies as well as peroxidase-conjugated secondary antibodies were used according to standard western blot procedure and peroxidase was detected by using the Lumi-LightPLUS Western Blotting Substrate ( Roche Diagnostics , Meylan , France ) . Total RNA was extracted using an RNeasy mini kit ( QIAGEN ) . 2 µg of each purified RNA preparation was reverse-transcribed and cDNAs were analysed by Q-PCR using specific primers ( see above ) . Analysis on Affymetrix DNA microarrays ( GeneChip Human Gene 1 . 0 ST Array ) was carried out using U2OS cells previously transfected with siRNA targeting p400 ( p400-1 , p400-2; previously described in [12] ) , or with control siRNA ( C1 , C2 ) . p400-silencing efficiency was checked by real time PCR and western blotting ( Figure S1A and S1B ) . 48 h after siRNA transfection , 100 ng of total RNA for each condition was subjected to cleanup , reverse transcription , amplification and labelling according to the manufacturer's instructions ( GeneChip Whole Transcript Sense Target Labelling Assay; Affymetrix ) . Following quantification , raw data normalization and statistical analysis was done using GeneSpring GX 10 . 0 analysis software ( Agilent technologies Inc , Santa Clara , CA ) . Briefly , after normalization using the RMA algorithm , a T-test statistical analysis was carried out to select genes whose expression levels significantly change in knockdown p400 condition compared to controls ( p-value <0 . 05 ) . The Fold Change ( FC ) for each gene was calculated for each siRNA p400 relative to each siRNA control . Qualified genes are those found regulated with stringent criteria , i . e . when their expression was modified more than 1 . 25 fold , in the same way , in at least three out of four comparisons ( p400-1 vs C1 , p400-2 vs C1 , p400-1 vs C2 and p400-2 vs C2 ) . Validation was done by RT-QPCR using FANCA , Hsp70 , TP53INP1 and p21 primers ( Figure S3 ) . Gene co-occurrence analysis was done using GeneCoDis2 online software ( http://genecodis . dacya . ucm . es/analysis/ ) [39] , [40] . Q-PCR analysis was performed on a CFX96 Real-time system device ( Biorad ) using the platinium SYBR Green qPCR SuperMix ( Invitrogen ) , according to the manufacturer's instructions , and specific primers ( see above ) . All experiments included a standard curve and all samples were analyzed in triplicates . Cells were submitted to the SA- β-Gal assay according to the manufacturer's instructions ( 96-well Cellular Senescence Assay Kit , Cell Biolabs ) . Fluorescence was read with a fluorescence plate reader ( Fluoroskan Ascent , Thermolabsystems , Courtaboeuf , France ) at 365 nm ( Excitation ) /502 nm ( Emission ) . For apoptosis analysis , cells were harvested and treated using an AnnexinV-FITC/7-AAD kit ( Beckman Coulter , Marseille , France ) according to the manufacturer's instructions . Cells were then analyzed by flow cytometry using FACScalibur apparatus ( Becton Dickinson ) . For cell cycle analysis , cells were harvested , fixed with ethanol and treated 30 min with Propidium iodide . Cells were then analyzed by flow cytometry using FACScalibur and cell cycle distribution was calculated using the ModFit LT V3 . 0 software ( Verity Software House , Inc . Topsham , ME ) . Immunofluorescence of cells on coverslips , observations and acquisition of native images were performed as previously described [41] . Quantification of fluorescence levels was done on approximately 100 cells/slide using home-developed macros in ImageJ software ( NIH , Bethesda , MA ) to normalize background , thresholds and signals . The presence of DNA damages was assayed by alkaline comet assay . U2OS cells were transfected with siRNA . After 48 h , cells were harvested , mixed with low-melting-point agarose and layered onto agarose-coated glass slides . Slides were maintained in the dark at 4°C until electrophoresis . Slides were submerged in lysis buffer for 1 h and incubated for 30 min in alkaline electrophoresis buffer . After electrophoresis , slides were neutralized and stained with ethidium bromide . Average Comet Tail Moment was scored for 100 cells/slide by using the CometScore-v 1 . 5 software . Generation of ROS was studied by flow cytometry using carboxy difluorodihydroFDA probe ( Invitrogen ) . Cells seeded in 6-well plates were washed with PBS and incubated with FDA probe ( 5 µg/ml ) in HBSS solution for 15 min at 37°C . Plates were then placed on ice and trypsinized . Cells were resuspended in PBS and immediately analysed by flow cytometry using FACScalibur apparatus . The mean fluorescence intensity of 25 , 000 cells was analyzed in each sample and corrected for autofluorescence from unlabeled cells . Experiments were performed as previously described in [12] . Briefly , a crosslink was done using formaldehyde 1% followed by lysis of the cells , sonication of the DNA and immunoprecipitation of protein/DNA complexes with specific antibodies or without antibody as negative control . Crosslink was then reversed by adding NaCl and DNA was purified with the GFX PCR kit ( Amersham ) and analyzed by Q-PCR using specific primers ( see above ) .
|
External or internal causes can lead to the generation of oxidative stress in mammalian cells . This oxidative stress is detrimental to cell life since it can induce protein damages or , even worse , DNA damages . Thus , cells have to control strictly oxidative stress levels . In this manuscript , we show that the p400 ATPase , a chaperone of specific histone H2A variants , is important for this control in mammals and therefore prevents DNA damage induction . Moreover , we demonstrate that the known roles of p400 in cell proliferation are dependent upon its effect on oxidative stress . Finally , we identify the mechanisms by which p400 modulates oxidative stress levels . Altogether , our study uncovers a new role of mammalian p400 and demonstrates its functional importance .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"molecular",
"biology/histone",
"modification",
"cell",
"biology/gene",
"expression",
"cell",
"biology/cellular",
"death",
"and",
"stress",
"responses"
] |
2010
|
The E1A-Associated p400 Protein Modulates Cell Fate Decisions by the Regulation of ROS Homeostasis
|
Prion diseases are fatal , neurodegenerative disorders in humans and animals and are characterized by the accumulation of an abnormally folded isoform of the cellular prion protein ( PrPC ) , denoted PrPSc , which represents the major component of infectious scrapie prions . Characterization of the mechanism of conversion of PrPC into PrPSc and identification of the intracellular site where it occurs are among the most important questions in prion biology . Despite numerous efforts , both of these questions remain unsolved . We have quantitatively analyzed the distribution of PrPC and PrPSc and measured PrPSc levels in different infected neuronal cell lines in which protein trafficking has been selectively impaired . Our data exclude roles for both early and late endosomes and identify the endosomal recycling compartment as the likely site of prion conversion . These findings represent a fundamental step towards understanding the cellular mechanism of prion conversion and will allow the development of new therapeutic approaches for prion diseases .
Conversion of the cellular prion protein ( PrPC ) into a conformationally altered pathogenic form , denoted PrP scrapie ( PrPSc ) is the central event in the pathogenesis of transmissible prion diseases [1] . In the most accredited model of prion formation and replication , a direct interaction between the pathogenic PrPSc template and the endogenous PrPC substrate is proposed to drive the formation of nascent infectious prions [1] . Despite decades of research , the mechanism of prion conversion , the intracellular site where this process occurs and how it leads to neurological dysfunction remain unknown [2] . A number of studies have already attempted to identify subcellular location ( s ) where prion conversion occurs , mostly by analyzing PrPC and PrPSc subcellular distribution and trafficking in infected cell lines [3] , [4] , primary neurons [5] , [6] , [7] and in the brains of infected animals [8] , [9] , [10] , [11] , [12] using different techniques . However , these results remain controversial and do not provide clear evidence for the involvement of any specific compartment . PrPC has been shown to localize to different compartments depending on the cell type . While Pimpinelli et al . reported a predominant localization in late endosomes of neuroblastoma-derived ( N2a ) and hypothalamic gonadotropin releasing ( GT1-7 ) cell lines [13] , other studies have reported that in primary neurons and in N2a cells PrPC is internalized and recycled back to the cell surface , with very little being localized in lysosomes [6] , [7] . Furthermore , a chimeric protein fused with GFP ( GFP-PrPC ) expressed in SN56 cells derived from septal cholinergic neurons , has been detected in the Golgi , early endosomes ( EEs ) , and in the endosomal recycling compartment ( ERC ) [14] . In hippocampal neurons , PrPC is found principally at the plasma membrane [5] and on vesicles resembling early endocytic or recycling vesicles [12] . Less information is available about the intracellular localization of PrPSc . This is mainly due to the lack of specific antibodies and the need for protein denaturation by guanidine-hydrochloride ( Gnd ) in order to reveal PrPSc epitopes [15] . Earlier work reported that the majority of PrPSc is intracellular [15] , sequestered within lysosomes of scrapie-infected N2a cells [3] , [16] , [17] with little localization at the cell surface [18] . In infected brains , PrPSc has been reported to accumulate at the plasma membrane and occasionally in late endosome/lysosome-like structures [10] . More recent reports describe accumulation of PrPSc either in the perinuclear Golgi region of neurons in scrapie-infected transgenic mice [11] , in the late endosomal compartment in infected GT1-7 and N2a cells [13] and at the cell surface and on early endocytic and recycling vesicles of hippocampal neurons [12] . Furthermore exogenous Alexa-labeled PrPSc was shown to be internalized into vesicles positive for late endosomal/lysosomal markers in SN56 cells and hamster cortical neurons [14] . Several studies indicate that PrPSc is formed after PrPC has reached the plasma membrane [3] , [19] , [20] , [21] . Furthermore , the endocytic pathway has been proposed to be important for the conversion of PrPC to PrPSc , based on the observation that release of nascent PrP from the cell surface using phosphatidylinositol-specific phospholypase C , or inhibition of endocytosis using a temperature block prevented PrPSc synthesis [2] , [3] . Moreover , PrPSc is cleaved at its N-terminus by endogenous proteases in acidic compartments immediately after its generation [3] , [16] , suggesting that its conversion to a protease-resistant state occurs prior to its exposure to proteases within an endo-lysosomal compartment . Furthermore , the expression of a dominant-negative version of the GTPase , Rab4a , which inhibits recycling to the plasma membrane , increases the production of PrPSc in infected N2a cells supporting the hypothesis that PrPSc formation does not require cell-surface recycling and occurs in an intracellular compartment [4] . Although no specific compartment has been identified , altogether these data provide good evidence that PrPSc may be generated either at the cell surface or more likely along the endocytic pathway . Proteins have been shown to enter the cell through many different routes ( for review see [22]; however , regardless of the internalization pathway used , cargo is first delivered to early endosomes . At this level cargo to be recycled is returned to the cell surface either by a fast-recycling pathway , directly from early endosomes , or it is transported first to the endosomal recycling compartment and then to the cell surface . In contrast , cargo destined for degradation is sorted to multivesicular late endosomes and finally to lysosomes , where degradation occurs ( for review see [23] , [24] and see below ) . Importantly , Rab GTPases , which reside in different subcellular compartments , have been identified as central regulators of intracellular transport , controlling specific fusion between vesicles and different compartments ( for review see [25] ) . The aim of the current study was to identify the subcellular compartment ( s ) of PrPSc production . To this end we set up a quantitative image analysis system and monitored PrPC and PrPSc localization and PK-resistant PrP levels under different experimental conditions in three different neuronal cell lines ( N2a , GT1 and CAD cells ) that represent established cell models for prion infection and replication [26] , [27] . We show here that at the steady-state high amounts of PrPSc reside in the endosomal recycling compartment ( ERC ) . Then by selectively perturbing PrP trafficking through the endosomal compartments we excluded roles for both early and late endosomes in prion conversion and provide evidence that this event occurs in the endosomal recycling compartment .
Because the subcellular distribution of PrP forms may yield clues as to the site of prion conversion , we quantitatively analyzed the intracellular distribution of PrPC and PrPSc in infected GT1 cells ( ScGT1 ) by using specific antibodies and a high resolution wide-field microscope ( Marianas , Intelligent Imaging Innovations ) together with different imaging software packages ( see Methods ) . Based on colocalization with different organelle markers , we found ∼20% of PrPC localized in the Golgi ( Figure 1A , upper panels and Figure 1E ) , ∼15% in early endosomes ( EEs ) ( Figure 1C , upper panels and Figure 1E ) , ∼15% in the endosomal recycling compartment ( ERC ) ( Figure 1D , upper panels and Figure 1E ) and only ∼3% in late endosomes ( LEs ) ( Figure 1B , upper panels and Figure 1E ) . The majority of the protein ( ∼50% ) was localized at the cell surface ( data not shown ) . Interestingly , we observed the same distribution of PrPC in uninfected GT1 cells , indicating that PrPSc infection did not alter PrPC trafficking in GT1 cells ( data not shown ) . Interestingly , in two other infected neuronal cell lines , ScCAD and ScN2a , PrPC was almost exclusively localized at the cell surface ( data not shown ) . Due to the lack of PrPSc - specific antibodies [15] in order to visualize PrPSc , a denaturation step with guanidine hydrochloride ( Gnd ) is required . However , this treatment does not allow one to easily distinguish PrPSc from PrPC . In contrast to previous studies , we took advantage of the features of our imaging system to discriminate PrPSc from PrPC by adjusting the signal threshold , and recording only the higher signal intensities characteristic of PrPSc staining after Gnd treatment in infected cells ( see Methods , Figure S1 and [28] ) . Because we were concerned that with this stringent approach we would detect mainly the brighter PrPSc signal which could derive from large aggregates , we performed velocity gradients in order to analyze the state of PrP aggregation , similar to what has been shown before in brain [29] . Interestingly , we did not observe any difference in the distribution of PrP on the gradients between uninfected cells containing only PrPC and infected cells containing both PrPC and PrPSc , thus arguing against the presence of large PrPSc aggregates in our cell model ( data not shown ) in contrast to what was described before in infected brains [29] , [30] . These data , although indirectly , support the use of this thresholding procedure to reveal ( with a reasonable approximation ) the intracellular distribution of all PrPSc in infected cells as recently shown [28] , [31] . By this approach we observed that unlike PrPC , only ∼10% of PrPSc was localized at the cell surface ( data not shown ) , while ∼90% was found to be intracellular as previously described [15] , [18] . However , in contrast to earlier reports [11] , [13] , [17] we did not observe any localization of PrPSc in the Golgi compartment ( Figure 1A lower panel ) the majority of the intracellular protein being localized throughout the endocytic pathway . While only ∼10% of the protein was localized in LEs ( Figure 1B lower panels and Figure 1E ) and EEs , ( Figure 1C lower panels and Figure 1E ) more than 25% was found in the ERC ( Figure 1D lower panels and Figure 1E ) . Interestingly , a similar distribution of PrPSc was observed in ScCAD and ScN2a cells ( Figure S2 ) . Therefore , the greater amount of PrPSc in the ERC when compared to other subcellular sites could indicate a potential involvement of this compartment in the conversion process . Alternatively , the ERC could just represent the compartment through which PrPSc recycles after being converted elsewhere ( eg . at the cell surface , in EEs or in LEs , which have previously been proposed as sites of conversion ) [17] , [32] , [33] . To directly examine this question we selectively perturbed the trafficking through the different endosomal compartments using pharmacological inhibitors or mutant proteins affecting the different pathways . To examine the role of the late endocytic pathway in prion conversion , we treated ScGT1 cells with U18666A . This compound triggers cholesterol accumulation in LEs and lysosomes by an as of yet unknown mechanism [34] . As a consequence , it effectively inhibits trafficking from EEs to LEs , because Annexin II , which is coordinating this step is redistributed into cholesterol laden LEs . In addition , exit from LEs to the Golgi and lysosomes is inhibited and protein degradation is impaired [35] , [36] . After treating ScGT1 cells with U18666A for 6 days , we observed a complete disappearance of Proteinase K ( PK ) resistant PrPSc , as shown by limited proteolysis using the PK assay ( Figure 2A and see the explanation in Methods ) . We also observed an increase in total PrPC levels as expected because of the known effect of U18666A on protein degradation [35] ( Figure 2A ) . Therefore , under these conditions the reduction in PrPSc was likely due to inhibition of PrPSc production rather than its increased degradation . Interestingly , PrPSc reduction following U18666A treatment was previously reported in ScN2a cells and was attributed to the redistribution of PrPC outside of cholesterol and glycosphingolipids enriched membrane microdomains , known as detergent-resistant domains ( DRMs ) or lipid rafts [37] . This is conceivable considering the effect of this drug on cholesterol and sphingolipid trafficking [38] . However , although we could reproduce the results in ScN2a cells ( Figure S3A ) , we did not observe any effect of U18666A on the association of PrPC with DRMs neither in ScN2a ( Figure S3B ) nor in ScGT1 cells ( Figure 2B ) , thus refuting the above hypothesis . In contrast , we observed that in both cell lines U18666A treatment altered the subcellular distribution of PrPC , which became highly enriched ( ∼40% ) in EEs after the treatment ( compare control cells with treated cells in Figure 2C and Figure S3C , and see quantification of EEA-1/PrP colocalization in Figure 2D ) . Therefore , these data suggest that U18666A affects PrP conversion by altering its intracellular trafficking . Furthermore , they also indicate that PrPC must exit the EE compartment in order to be converted to PrPSc . To identify the pathway involved in prion conversion we had to determine to what extent the U18666A treatment affected endocytic pathways in ScGT1 cells . To this aim we analysed the trafficking of two molecules widely used to characterize trafficking through endocytic compartments: dextran , which traffics through EEs and reaches LEs , and transferrin ( Tfn ) which recycles back to the surface through EEs and the ERC . By following fluid phase uptake of fluorescently tagged dextran ( Figure S4 ) we confirmed that U18666A also inhibits the traffic from EEs to LEs in ScGT1 cells ( Figure S4B ) as previously reported for other cell types [35] , [36] . Furthermore , by staining cholesterol with filipin , we also show that U18666A causes enlargement of LEs , which accumulate cholesterol ( see large organelles in Figure S4B , costained with filipin and LBPA ) . In addition to this well documented effect , we found that in U18666A treated cells fluorescently tagged Tfn was not able to reach the ERC and after 15 minutes of internalization was still arrested in EEs both in ScGT1 ( Figure 3B ) and ScN2a cells ( Figure S3D ) . Nonetheless , we observed that , similar to control ScGT1 cells , Tfn could not be detected inside the cells after a 45 minutes chase period ( compare panels c and d in Figure 3 ) , indicating that under these conditions Tfn recycling to the surface via EEs was unaffected . These data therefore show that prolonged treatment with U18666A did not alter arrival to EEs nor recycling from EEs to the PM but inhibited trafficking pathways both between EEs and LEs and between EEs and the ERC . Since endogenous PrPC accumulates in EEs in the presence of U18666A ( Figures 2C and 2D ) , overall these results indicate that EEs ( and recycling from EEs to the PM ) are not involved in PrPSc production , while exit of PrP from EEs towards either the ERC or LEs is required for conversion to occur . Therefore we decided to selectively inhibit these two pathways and to analyze the effect on scrapie production . In order to directly test whether LEs/lysosomes are involved in PrPSc production , we used an RNAi approach to downregulate Alix , a protein that is required for the biogenesis of LEs [39] ( Figure S5A ) . We obtained a clear reduction of protein levels ( ∼80% ) after transfection of Alix siRNA in ScGT1 and ScCAD cells ( Figure 4A and Figure S5B ) , but not in ScN2a cells ( data not shown ) . Consistent with previous findings in HeLa cells [39] the number of LEs and lysosomes in transfected ScGT1 cells was drastically reduced ( ∼10 fold ) ( compare top and bottom panels in Figure S5A and quantifications ) . However , despite the drastic reduction in the number of LEs , the levels of total PrP and PrPSc were unaltered in Alix-depleted ScGT1 cells . Instead , we observed a slight increase in PrPSc as well as total PrP , likely due to the reduction in protein degradation in cells with fewer LEs/lysosomes ( Figure 4A ) . Interestingly , we could also reproduce these data in ScCAD cells ( Figure S5B ) , indicating that the effect of Alix depletion on PrPSc was not limited to a single infected cell line . Overall , these data demonstrate that PrP transport to LEs is not required for PrPSc production and rule out the involvement of LEs in scrapie conversion . Interestingly , we observed that Alix depletion did not affect the intracellular localization of PrPC and PrPSc , which were found to be enriched in the ERC similar to control cells ( Figure 4B , compare colocalization with Tfn in control and treated cells for PrPC ( −Gnd ) and PrPSc ( +Gnd ) and see quantification in the right panel ) , thus , pointing to a role for this compartment in prion conversion . In U18666A-treated cells Tfn delivery to the ERC was inhibited ( Figure 3B ) but its recycling to the surface from EEs was unaffected ( Figures 3C and 3D ) . Under the same conditions , PrP was enriched in EEs ( Figure 2B ) . These data suggest that ERC-independent recycling to the surface via EEs is not likely to be relevant for PrPSc production . Consistent with this hypothesis expression of a dominant-negative Rab4 mutant ( GFP-Rab4N121I ) , which impairs recycling from EEs [40] , did not affect PrPSc levels in either ScGT1 or ScN2a cells ( Figure S6 ) . Interestingly , it was previously reported that in ScN2a cells the same Rab4 dominant-negative mutant increased scrapie levels [4] . Although in our hands the levels of PrPSc remained unaltered in ScN2a cells transfected with Rab4 dominant-negative , both of these sets of data concur in showing that recycling from EEs to the PM is not involved in scrapie production . To directly test whether PrPC must reach the ERC in order to be converted to PrPSc we analyzed the effect of Rab22a overexpression on PrP sorting and PrPSc production . Rab22a has been shown to regulate Tfn sorting from EEs to the ERC [41] and to be involved in sorting of MHCI from the ERC into tubular carriers destined for the cell surface [42] . In CHO cells overexpression of GFP-Rab22a causes a characteristic enlargement of EEs [41] , probably due to enhanced homotypic fusion of Rab22-containing vesicles . This prevents segregation of the domains required for fusion with other compartments , resulting in delayed transport out of EEs . As a consequence , Tfn accumulates in this compartment [41] . Importantly , Rab22 overexpression in CHO cells does not impair protein delivery from EEs to LEs . In contrast to CHO cells , when GFP-Rab22a is overexpressed in HeLa cells it localized to the ERC and to tubular carriers and does not affect either the distribution or the recycling of Tfn [42] . In order to see whether in our infected cell models Rab22a was acting as a specific effector of trafficking between EEs and the ERC , we analyzed the subcellular distribution of GFP-Rab22a and the effect of its overexpression on Tfn trafficking in our three infected cell models . Similar to CHO cells , in transfected ScGT1 cells we observed enlarged EEA-1 positive EEs decorated with GFP-Rab22a ( Figure 5A ) . We also found that the transfected cells were able to internalize Tfn normally ( Figure S7A ) . However , in contrast to control cells , Tfn was not transported to the ERC after internalization , but remained inside EEs ( ∼80% ) ( Figure S7A ) . Similarly PrP was enriched in the EEs of Rab22a-transfected ScGT1 cells ( Figure 5A ) as shown by quantification of its co-localization with EEA-1 ( see graph in Figure 5A ) . These data therefore indicate that , like Tfn , PrP was not able to reach the ERC in Rab22a-transfected ScGT1 cells . Interestingly , by performing PK assay we observed that while total PrP levels were unchanged , there was ∼50% decrease in PrPSc levels in GFP-Rab22a-expressing cells compared to control cells transfected with GFP alone ( Figure 5B , middle and right panels ) . This partial decrease in PrPSc levels corresponded to the transfection efficiency , which was approximately 50% ( data not shown ) . In agreement with the biochemical data , when GFP-Rab22a transfected cells were analyzed by immunofluorescence after Gnd treatment , no PrPSc signal could be observed in GFP-Rab22a-expressing cells ( Figure 5C ) . Interestingly , like in ScGT1 cells , GFP-Rab22a was mainly localized in enlarged EEA-1 positive endosomes in transfected ScCAD cells ( Figure S8A , upper panels ) . However , a substantial amount of GFP-Rab22a was also found in perinuclear tubular structures ( Figure S8A upper panels , arrowheads ) and vesicles lacking EEA-1 ( Figure S8A upper panels , arrows ) . Despite this heterogenous distribution , sorting of Tfn from EEs to the ERC was impaired in GFP-Rab22a expressing CAD cells similar to ScGT1 cells ( Figure S8A lower panels ) . Therefore , as expected , Rab22a expression also caused redistribution of PrPC to EE and decreased PrPSc production in this cell line similar to what was observed in ScGT1 cells ( Figures S8B , S8C and S8D ) . In contrast , in ScN2a cells , Rab22a behaved similarly to HeLa cells and did not inhibit the trafficking of Tfn from EEs to the ERC . Consequently , overexpression of Rab22a in ScN2a cells did not have any effect on the levels of PrPSc ( data not shown ) . Although these data indicate that PrPC must reach the ERC in order to be converted , we had to rule out the possiblity that overexpressed Rab22a was titrating out other cellular sorting factors . To this end , we used a complementary approach and downregulated endogenous Rab22a using siRNA . Given the role of Rab22 in cargo sorting towards the ERC [41] , depletion of Rab22a should cause a delay in transport from EEs , resulting in the same effect as Rab22 overexpression . As expected , when endogenous Rab22a was depleted from ScGT1 cells a similar decrease in PrPSc levels was obtained ( Figure S7B ) . Altogether , these data indicate that a block of PrPC trafficking from EEs to the ERC reduces levels of PrPSc in infected cells of different neuronal origin . Nonetheless , they do not allow us to discriminate whether the reduction in PrPSc levels is due to impaired production or to enhanced degradation of PrPSc . In particular , it is possible that impairment of PrP recycling in GFP-Rab22a expressing cells could divert PrP towards a degradation pathway . To rule out this hypothesis we monitored the distribution of PrPSc and Alexa-546 dextran in ScGT1 cells expressing GFP-Rab22a three days post-transfection , when the reduction of PrPSc was not complete and when we could still detect PrPSc in the transfected cells . Under these conditions dextran was mainly found in LEs after the 3 h chase period ( Figure 6A , lower panels ) , confirming that Rab22a did not affect the routing of dextran from EEs to LEs [41] . Conversely , only a minority of PrPSc was localized in LE under these conditions similar to control cells ( Figure 6A , upper panels , arrows ) . Therefore , these data indicate that in GFP-Rab22a-expressing cells PrP was not diverted toward LEs . Instead , a significant amount of PrPSc was distributed in GFP-Rab22a-positive EEs , similar to what was observed for PrPC ( Figure 6A , upper panels , arrowheads and Figure 5A ) . Altogether , these results indirectly suggest that the observed decrease in PrPSc levels in Rab22-expressing cells is not due to increased delivery and degradation of PrPSc in LEs , but rather is due to impaired scrapie production . To further test this hypothesis and to discriminate between increased degradation and decreased production of PrPSc , we compared the levels of PrPSc in control cells and in cells transfected with GFP-Rab22a after treating them with ammonium chloride ( NH4Cl ) , which impairs lysosomal degradation [43] . We reasoned that if the reduction of PrPSc by Rab22a was due to increased lysosomal degradation , a block of the lysosomal function with NH4Cl should revert its effect and result in unchanged or increased PrPSc levels . On the the other hand , if the decrease of PrPSc levels by Rab22a-expression was not due to increased PrPSc degradation but rather to inhibition of PrPSc synthesis , NH4Cl treatment should not interfere and we should still observe a decrease in PrPSc levels . As expected from a block of lysosomal degradation we observed an increase in total cellular PrP levels in NH4Cl-treated cells ( in both control cells transfected with GFP and in cells transfected with GFP-Rab22a ) when compared to untreated cells ( Figure 6B right panels ) . Similarly , PrPSc levels were also increased in control cells ( expressing GFP alone ) treated with NH4Cl ( Figure 6B right panels ) . However , in contrast to total PrP , PrPSc levels were reduced in GFP-Rab22a expressing cells , independently of the presence of NH4Cl ( Figure 6B right panels ) . These results therefore suggest that the observed reduction of PrPSc levels in GFP-Rab22a expressing cells is due to impaired PrPSc production rather than to increased degradation . Overall , the results described in the previous paragraph indicate that the recycling of PrP through the ERC is required for scrapie production . However , they do not allow us to discriminate whether PrPC to PrPSc conversion occurs within this compartment or upon return to the cell surface . In order to directly analyze the involvement of the ERC in prion conversion , we overexpressed in ScGT1 cells wild type and dominant-negative forms of Rab11 , which have been shown to modulate recycling through ERC . Similar to what has been shown in other cell lines [44] , [45] , we found both Rab11wt-GFP and Rab11S25N-GFP in the Golgi , in the ERC , and occasionally in peripheral vesicular structures that were negative for EEA-1 ( Figure S9 and data not shown ) . In contrast to previous reports in CHO and BHK cells [44] , overexpression of Rab11S25N-GFP in ScGT1 cells did not interfere with the transport of Tfn from EEs to the ERC ( Figure S9 ) but impaired its recycling from the ERC to the PM ( Figure 7A ) . Indeed , in contrast to control cells in which the majority of Tfn was chased out of the cells after 45 min , in cells overexpressing Rab11S25N-GFP exit of Tfn from the ERC was delayed . While a significant amount of the protein was still detected inside the cells after a 45 min chase period ( Figure 7A upper panels ) , complete clearance of Tfn occurred only after 90 min ( data not shown ) . Interestingly , compared to control cells , a higher amount of PrPSc was observed to colocalize with Tfn in the ERC of transfected cells ( Figure 7A lower panels ) indicating that PrP was also retained in this compartment . Significantly , in these cells the levels of PrPSc were slightly increased ( Figure 7B ) . We also attempted to analyze PrPSc levels in ScCAD and ScN2a cells expressing Rab11S25N-GFP . However , due to increased mortality of Rab11S25N-GFP-expressing cells we were not able to perform a quantitative analysis of PK-resistant PrP levels . Nonetheless , immunofluorescence analysis in the transfected cells revealed the presence of PrPSc in these cells ( data not shown ) , thus supporting the result obtained in ScGT1 cells . Overall these data strongly suggest that PrPSc conversion occurs in the ERC and imply that recycling from ERC to the PM is not relevant to this process .
Conversion of PrPC to PrPSc is the key event in prion pathogenesis . Prion conversion is thought to occur at a site where the two protein forms meet and are allowed to physically interact . To date there is no direct evidence for the involvement of any specific intracellular compartment in this event as several compartments have been proposed to have a role in different cell systems [11] , [12] , [13] , [16] , [17] , [32] . Based on analyses of the subcellular localization of different PrP forms , lysosomes have been proposed as a possible location for the conversion process [13] , [17] , [32] . However , contrasting data about PrPC and PrPSc localization exist [6] , [7] ( and see introduction ) , underlying the need for a more systematic and quantitative approach aimed at uncovering the site of conversion . The inconsistencies in the reported localization of different PrP forms are most likely related to the lack of PrPSc antibodies that can distinguish between the two prion forms and the need for protein denaturation by guanidine hydrochloride ( Gnd ) to reveal PrPSc epitopes making such analyses quite difficult . Because characterization of the intracellular localization of different PrP forms can provide important clues about the compartment where PrPC to PrPSc conversion occurs , we reassessed the subcellular distribution of PrPC and PrPSc in three different neuronal cell lines infected with different prion strains ( ScGT1 infected with RML , ScCAD infected with 139A and ScN2a infected with 22L ) , which have been widely used as cellular models for prion infection [26] , [27] . To clearly distinguish between PrPC and PrPSc we employed advanced imaging technology complemented by quantitative image analysis , allowing us to define the relative amounts of PrPC and PrPSc in each subcellular compartment after treating the cells with Gnd in immunofluorescence experiments ( Figure S1 , Methods and [28] ) . A similar approach , based on thresholding of the lower PrPC-derived fluorescence in order to extract only the higher fluorescence signal from PrPSc was recently reported by Veith and collegues in N2a cells [46] . In agreement with the results from this group and in contrast to previous reports in ScN2a and ScGT1-7 cells in which standard immunofluorescence approaches were utilized [13] , [17] , [32] , we found only a small amount of PrPSc in LEs , arguing against the involvement of this compartment in PrPSc production . Instead , by quantitative fluorescence analysis we observed a preferential localization of PrPSc in the endosomal recycling compartment of all three cells lines tested ( Figure 1 and Figure S2 ) . In support of these findings , a similar localization was recently observed using cryo-immunogold electron microscopy on hippocampal sections from mice infected with the RML prion strain [12] . These observations prompted us to further assess the involvement of the endocytic pathways and specifically that of the endosomal recycling compartment in PrPSc conversion . To this aim we selectively inhibited PrP trafficking through the different endocytic compartments using both pharmacological and reverse genetic approaches in infected cells ( see Figure 8 ) and analyzed PrPSc levels under the different experimental conditions . We demonstrated that EEs are not involved in PrPSc production . Indeed , when we blocked PrP exit from EEs the levels of PrPSc were drastically reduced and an accumulation of PrPC in EEs was observed ( Figures 2A , 2C , 5A , 5B , S8B , S8C and S8D ) . Besides ruling out the involvement of EEs in the conversion process , these data indicate that PrPC must exit EEs in order to be converted . Furthermore , in line with previous observations [4] recycling from EEs to the cell surface does not seem to play an important role in PrPSc production ( Figure S6 ) . Therefore we analyzed the sorting from EE to LE and/or to the recycling compartment . By specifically reducing the number of LE by Alix depletion ( Figure 4 and Figure S5 ) we demonstrated that LE are not involved in PrPSc production . In contrast , PrP sorting from EEs to the ERC seems to be the crucial event in the conversion process . In particular , we found that PrPSc levels are drastically reduced when trafficking from EEs to the ERC is specifically impaired ( Figure 5 and Figure S8 ) . We clearly demonstrate that this is not a cell-type-specific effect . Indeed , we observed a decrease in PrPSc when we overexpressed Rab22a in both ScGT1 ( Figure 5 ) and ScCAD ( Figure S8 ) cells , where Rab22a has a clear effect in inhibiting transport from EEs to ERC . In contrast , no reduction in PrPSc was observed when Rab22a was overexpressed in ScN2a cells , where Rab22a does not control this pathway ( data not shown ) . In addition to PrPSc , other glycosylphosphatidylinositol ( GPI ) -anchored proteins ( GPI-APs ) have also been shown to be retained in the ERC and this retention is GPI- , sphingolipid- and cholesterol-dependent [47] , [48] . Furthermore , retention of GPI-APs in the ERC can be physiologically relevant , as in the case of the Folate Receptor for which loss of retention in this compartment ( by removing its GPI anchor or by depleting cholesterol ) severely impairs folate uptake [49] . In the case of the prion protein , either removal of the GPI anchor , its replacement with a transmembrane domain or cellular cholesterol depletion impairs scrapie formation in cell lines [50] , [51] , [52] , [53] even though the anchor seems to be dispensable for the conversion process itself [54] , [55] , [56] . In support of this hypothesis are the conclusions of a recent study by McNally and colleagues suggesting that the GPI anchor of PrP is important for the persistent infection of cells in vitro [57] . These authors found that neural stem cells derived from PrP null mice expressing only anchorless PrP cannot be persistently infected , although production of PK resistant PrP was detected in the first 96 hours after infection [57] . In agreement with these in vitro results , when a PrP molecule , fused to the Fc portion of human IgG1 heavy chain and lacking the GPI anchor , is expressed in Prnp0/0 mice it is not convertible and delays the onset of the disease when expressed together with wild type PrP [58] . Moreover , intracerebral inoculation of prions into Prnp0/0 mice expressing an anchorless form of PrP results in a reduced titre of infectivity , different anatomical localization of amyloid plaques and no obvious clinical signs of disease compared to wild type mice [59] . This suggests that the GPI anchor , although not essential for the conversion process , plays an important role in its efficiency in vivo . Thus , it is possible that the GPI anchor is necessary for mediating the targeting of PrP to the correct intracellular site ( and/or to a propitious membrane domain ) in order to sustain the conversion process . In this scenario PrP retention in the ERC might be necessary in order to concentrate prion proteins in a cholesterol enriched membrane domain for a sufficient amount of time to promote conversion . Indeed , by inhibiting PrP exit from the ERC using the Rab11 dominant-negative mutant , we found an increase in PrPSc levels , supporting the hypothesis that prion accumulation in the ERC stimulates PrPSc production . Furthermore , the observation that cholesterol redistribution from other cellular compartments to LEs , induced by U18666A treatment , impairs PrP trafficking from EEs to the ERC ( Figure 2C ) suggests that cholesterol is not only important for GPI-AP retention within the ERC , but might also be required for its sorting from EEs to the ERC . Given that protein sorting from EEs to the ERC is Rab22a-dependent [41] , [42] it is possible that , similarly to Rab4 [60] , cholesterol levels also influence Rab22a function . However , this attractive possibility remains to be explored . The mechanism of PrPC to PrPSc conversion remains unknown . It has been proposed that the conversion process requires interaction between PrPC and PrPSc and is assisted by other protein ( s ) [1] . However , despite years of effort this protein has not been identified , partially due to the secondary responses of cells to prion infection that render analysis more difficult [61] , [62] . Our results suggest that if such a protein exists it encounters PrP in the ERC , therefore narrowing down potential candidates . Alternatively , the PrPC to PrPSc conversion may not be mediated by a specific protein but instead may depend on the local lipid environment . The observation that cholesterol depletion reduces PrPSc levels has highlighted the importance of the lipid environment for scrapie production [50] . Cholesterol and glycosphingolipids are enriched in membrane microdomains , known as detergent-resistant domains ( DRMs ) or lipid rafts [63] . They have been implicated in various processes such as signal transduction , endocytosis and cholesterol trafficking [64] . It has been proposed that PrPC to PrPSc conversion occurs within lipid rafts , since both proteins were found to reside in such domains [65] . Recent evidence suggests that lipid rafts are heterogeneous both in terms of their protein and lipid content , and can be localized to different regions of the cell [64] . Furthermore , it has been shown that the endocytic route of GPI-APs is cell-type dependent and correlates with their residence time in DRMs . In CHO and BHK cells , transport of GPI-APs to the ERC or to late endosomes , respectively , is accompanied by differential association to DRMs [66] . Interestingly , PrP and Thy-1 , another GPI-AP , were found to reside in distinct lipid domains in rat and mouse brain [67] . Moreover , it has been shown that Thy-1 associates with rafts only transiently [68] and that raft partitioning , or raft residence time , can be modulated by subtle changes in lipid composition [69] . Therefore , a cholesterol-dependent retention mechanism in the ERC could facilitate the efficient conversion of native prion protein into the scrapie form , by segregating it into distinct membrane domains or aggregates . Interestingly , ERC membranes have been shown to be enriched in cholesterol in different cells [24] , [70] , [71] . Furthermore , PrP association with specific lipid domains in different cellular contexts may influence its cellular transport and thereby determine the differential cellular susceptibility to prion infection . In conclusion , our data , based on observations in three different cell models of prion infection , indicate that the ERC is a likely candidate for the intracellular site of prion conversion . Furthermore , the fact that these results were consistent for two different prion strains ( RML and 22L ) suggests that this is a general mechanism . Although we cannot completely exclude contributions by other compartments to prion conversion , recent evidence , showing a localization of PrPSc in the recycling endosomes of primary hippocampal neurons derived from infected brains [12] supports our hypothesis , strengthening the case for the role of the ERC in prion conversion in infected mouse models . These results open the door to more targeted approaches to study the factors involved in this central event of the disease and to develop better therapeutic strategies .
GT1-1 cells ( gift of Dr . Mellon P . , University of California , San Diego , USA ) were infected with RML prion strain ( gift of Dr . Korth K . , Heinrich Heine University Düsseldorf , Germany ) and ScN2a cells infected with 22L prion strain ( gift of Dr . Korth K . , Heinrich Heine University Düsseldorf , Germany ) were cultured in DMEM with addition of 10% FCS ( Invitrogen ) . ScCAD cells infected with 139A prion strain ( gift of Dr . Laude H . , Institut National de la Recherche Agronomique , Jouy-en-Josas , France ) were cultured in DMEM:F12 ( Invitrogen ) with addition of 10% FBS . U18666A was purchased from Calbiochem . All other chemicals were purchased from Sigma . SAF32 and SAF61 monoclonal antibodies were purchased from SPI-BIO , while POM1 monoclonal antibodies were purchased from Dr . A . Aguzzi ( Institute of Neuropathology , University Hospital Zurich , Zurich , Switzerland ) . Anti-tubulin monoclonal antibodies were from Sigma . Anti-flotillin-1 monoclonal antibodies were from BD Transduction Laboratories . Anti-GFP antibodies and all the fluorescently labeled secondary antibodies were purchased from Invitrogen ( Molecular Probes ) , as well as Lysosensor , fluorescently tagged transferrin and dextran . Anti-EEA-1 antibodies were kind gift of Dr . M . Zerial ( Max Planck Institute of Molecular Cell Biology and Genetics , Dresden , Germany ) and anti-LBPA antibodies were gift from Dr . J . Gruenberg ( University of Geneva , Geneva , Switzerland ) . Anti-Alix antibodies were gift from Dr . R . Sadoul ( Neurodégénérescence et Plasticité , E0108 , INSERM/Université Joseph Fourier , Grenoble , France ) . Anti-Rab22a antibodies and GFP-Rab22a construct were kind gift of Dr . L . Mayorga ( Laboratorio de Biología Celular y Molecular , Instituto de Histología y Embriología ( IHEM-CONICET ) , Facultad de Ciencias Médicas , Universidad Nacional de Cuyo , Mendoza , Argentina ) . Rab11wt-GFP and Rab11S25N-GFP were gift of Dr . S . Mayor ( National Centre for Biological Science ( TIFR ) , Bellary Road , Bangalore 560 065 , India ) . Alix siRNA oligos were published previously [39] and purchased from Dharmacon . All the other siRNA used in the study were predesigned ON TARGETplus SMARTpool from Dharmacon . U18666A was reconstituted according to producer's instructions and used in DMEM+10% FCS at 5 µM concentration for ScGT1-1 cells or 1 µM concentration for ScCAD and ScN2a cells . During the 6 day-treatment , medium containing U18666A was changed every 2 days . To analyze PrP levels on western blot cells were lysed and lysates were either treated or not with Proteinase K as described . PrP was revealed by SAF61 antibodies . For immunofluorescence analysis cells were washed after the 6 day treatment , and processed as described below in Immunofluorescence analysis . ScGT1-1 cells were transfected at 50% confluence using FuGENE6 ( Roche Diagnostic ) for DNA constructs according to manufacturers protocol . Transfection of siRNA in ScGT1-1 was done using HiPerFect ( Qiagen ) . To downregulate Alix and Rab22a , 250 nM oligo and 500 nM oligo respectively was used with 10 µl of HiPerFect per 60 mm dish , while 3 µl of HiPerFect was used per well for cells grown in 24-well plate . Hyperfect was mixed with siRNA in DMEM without FBS , incubated for 10 min at room temperature and added to the cells . Transfection of ScCAD and ScN2a cells with both DNA constructs and siRNA was done using Lipofectamine 2000 ( Invitrogen ) , according to producer's protocol . In order to detect the effect on PrPSc levels , both silencing and overexpression of proteins were required during 6-day period . Therefore in all the experiments siRNA and plasmids , except of pEGFP were transfected twice ( 3 days post-transfection a second transfection was performed ) during 6 days . ScGT1-1 , ScCAD and ScN2a cells ( 200000 ) were grown on coverslips for 2 days . For steady state localization , 50 µg/ml of Alexa-labeled transferrin was added to the medium for 15 minutes at 37°C . In pulse and chase experiment 50 µg/ml of transferrin was added to the medium for 15 min at 37°C . The cells were extensively washed with PBS and Alexa-labeled transferrin was chased out of the cells with the excess ( 5 mg/ml ) of unlabeled transferrin for indicated time period . To follow dextran endocytosis by immunofluorescence , cells were first incubated in 10% FBS at 4°C for 30 min followed by 30 min incubation with 3 mg/ml of Alexa-dextran at 37°C . Cells were then extensively washed with PBS and chased in the medium deprived from dextran for additional 3 h at 37°C . For immunofluorescence analysis , cells were fixed with 2% paraformaldehyde for 30 min unless differently indicated and permeabilized with 0 , 1% of Triton X-100/PBS . To analyze PrPSc cells were additionally incubated in 6 M guanidine-hydrochloride for 10 min after permeabilization . Cells were then blocked in 2% BSA for 30 min unless differently indicated , following by 30 min-incubation with primary and secondary antibodies respectively . For immunofluorescence analysis of Alexa-labeled dextran cells were fixed in 4% PFA for 4 h and further processed as described . When filipin staining was used , cells were fixed with 4% PFA for 60 min and blocked with 0 , 2% BSA/PBS . Filipin ( 250 µg/ml ) was added to blocking solution and additional 30 min-incubation was performed after incubation with secondary antibodies . Immunofluorescence was analyzed by high-resolution wide-field microscope Marianas ( Intelligent Imaging Innovations ) using a 63× oil objective . When PrPSc was analyzed , the auto-scaling ( min/max ) of signal detection was used to record only maximal signal intensities . Briefly , the exposure time ( 100 ms ) used to detect PrPSc fluorescence was insufficient to detect significant signal coming from PrPC . The camera settings were then adjusted to record only the range of PrPSc-derived fluorescence signal . The images were deconvolved using constrained iterative algorithm in Slidebook software ( Intelligent Imaging Innovations ) . Colocalization was quantified by intensity correlation coefficient-based ( ICCB ) analysis using Imaris software ( Bitplane ) or JACoP ( http://rsb . info . nih . gov/ij/plugins/track/jacop . htlm ) . Statistical analysis of the correlation of the intensity values of either green and red pixels or blue and red pixels in dual-channel image was performed using Pearson's and Menders's coeficient and Van Steensel's approach [72] . The amount of total fluorescent signal in one channel overlapping with the total fluorescent signal in the other channel was presented . To quantify the number of LBPA and Lysosoensor positive vesicles in control cells or cells transfected with Alix siRNA , the vesicles were counted by particle analysis using Image J software ( http://rsb . info . nih . gov/ij ) . Control ScGT1 and U18666A-treated cells grown to confluence in 150 mm dishes were harvested in cold PBS and resuspended in 1 ml lyses buffer ( 1% TX-100 , 10 mM Tris-HCl pH 7 . 5 , 150 mM NaCl , 5 mM EDTA ) , left on ice for 20 minutes and passed 10 times through 22-gauge needles . Lysates were mixed with an equal volume of 85% sucrose ( w/v ) in 10 mM Tris-HCl pH 7 . 5 , 150 mM NaCl , 5 mM EDTA , placed at the bottom of a discontinuous sucrose gradient ( 30–5% ) in the same buffer and ultracentrifuged at 200 , 000 g for 17 hours at 4°C in an ultracentrifuge ( SW41 rotor from Beckman Instruments , Fullerton , CA , USA ) . Twelve fractions were harvested from the top of the gradient . A white light-scattering band identified in fraction 5 at the interface between 5 and 30% sucrose , contained DRM domains . Samples were TCA precipitated and proteins were analyzed by western blotting . Cells were grown to confluence in 60 mm dishes and lysed in 500 µl of Lyses buffer ( 0 , 5% triton X-100 , 0 , 5% DOC , 100 mM NaCl , 10 mM Tris-HCl pH 8 ) . To analyze PrPSc by western blot , lysates ( 500 µg of protein ) were treated with 20 µg/ml of Proteinase K ( PK ) for 30 min on 37°C and protein content was pelleted by centrifugation at 14000 rpm and 4°C for 1 hr . Pellets were resuspended in Laemmli buffer and proteins were analyzed by western blotting . In contrast to PrPC , which is completely degraded by PK , only partial degradation of PrPSc occurs in this condition [73] . Other proteins , including total PrP were analyzed by western blotting from 25 µg of total lysate . T-test was used for statistical analysis of the data . The differences were considered significant when p<0 . 05 .
|
The misfolded form ( PrPSc or prion ) of the naturally occuring prion protein ( PrPC or cellular PrP ) is responsible for neurodegenerative diseases such as Creutzfeldt-Jakob disease ( CJD ) , bovine spongiform encephalopathy ( BSE ) ( also known as ‘mad cow disease’ ) and a new variant of CJD ( vCJD ) , which is thought to be caused by ingestion of cattle-derived foodstuffs contaminated with prions . These diseases are characterized by the accumulation of protein deposits in the central nervous system ( CNS ) . However , unlike other neurodegenerative diseases , prion diseases are infectious and prions are able to propagate in a chain reaction by imposing their malconformed state onto the properly folded cellular proteins . Understanding where the conversion of PrPC into PrPSc occurs in cells has been an unsolved question until now . By analysing the intracellular localization of PrPC and PrPSc and measuring the levels of PrPSc produced in infected neuronal cell lines under conditions in which intracellular trafficking of the protein is impaired , we found that prion conversion occurs in the endosomal recycling compartment ( ERC ) where it transits after being internalized from the cell surface . This study will help to clarify the cellular mechanism of the disease and it opens the way to new therapeutic strategies aimed at the conversion compartment .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"cell",
"biology/membranes",
"and",
"sorting",
"infectious",
"diseases/prion",
"diseases"
] |
2009
|
Identification of an Intracellular Site of Prion Conversion
|
Adult Clonorchis sinensis live in the bile duct and cause clonorchiasis . It is known that the C . sinensis metacercariae excyst in the duodenum and migrate up to the bile duct through the common bile duct . However , no direct evidence is available on the in vivo migration of newly excysted C . sinensis juveniles ( CsNEJs ) . Advanced imaging technologies now allow the in vivo migration and localization to be visualized . In the present study , we sought to determine how sensitively CsNEJs respond to bile and how fast they migrate to the intrahepatic bile duct using PET-CT . CsNEJs were radiolabeled with 18F-fluorodeoxyglucose ( 18F-FDG ) . Rabbits with a gallbladder contraction response to cholecystokinin-8 ( CCK-8 ) injection were pre-screened using cholescintigraphy . In these rabbits , gallbladders contracted by 50% in volume at an average of 11 . 5 min post-injection . The four rabbits examined were kept anesthetized and a catheter inserted into the mid duodenum . Gallbladder contraction was stimulated by injecting CCK-8 ( 20 ng/kg every minute ) over the experiment . Anatomical images were acquired by CT initially and dynamic PET was then carried out for 90 min with a 3-min acquisition per frame . Twelve minutes after CCK-8 injection , about 3 , 000 18F-FDG-labeled CsNEJs were inoculated into the mid duodenum through the catheter . Photon signals were detected in the liver 7–9 min after CsNEJs inoculation , and these then increased in the whole liver with stronger intensity in the central area , presenting that the CsNEJs were arriving at the intrahepatic bile ducts . In the duodenum , CsNEJs immediately sense bile and migrate quickly with bile-chemotaxis to reach the intrahepatic bile ducts by way of the ampulla of Vater .
Human Clonorchis sinensis infections are endemic in East Asia countries , such as China , Vietnam , and Korea , where 15–20 million people are estimated to be infected [1] . In South Korea , clonorchiasis is currently the most prevalent parasitic infection and estimated to infect 1 . 3 million people [2] . C . sinensis infected patients suffer from abdominal pain , hepatomegaly , obstructive jaundice , indigestion , and complications of cholecystitis , cholelithiasis , and cholangiocarcinoma [3] , [4] . Furthermore , recently , C . sinensis was categorized as a Group 1 biological carcinogen by the International Agency for Research on Cancer [5] . Humans are the final host and become infected by eating freshwater fish containing C . sinensis metacercariae . Ingested metacercariae excyst in the duodenum due to trypsin stimulation [6] , and the newly excysted C . sinensis juveniles ( CsNEJs ) migrate to the intrahepatic bile duct . The migration route of CsNEJs has been previously examined in experimental animals . In rabbit experiments , the common bile duct was first ligated surgically then C . sinensis metacercariae were administered to the rabbits through a gastric tube . One month later , adult C . sinensis were searched for in the bile ducts , but were not found . Based on this finding it was suggested that CsNEJs migrate through the common bile duct to the intrahepatic bile ducts [7] , and this has been taken to be the migration route of C . sinensis in mammalian hosts [3] . Parasites such as C . sinensis have specific in vivo migration routes in their hosts , which could be targeted for development of therapeutic and preventive interventions against parasitic diseases . Furthermore , in vivo imaging technologies have been recently developed for the clinical diagnoses of a wide range of diseases , and these techniques have a potential to monitor the movements of CsNEJs . Molecular imaging has emerged as a discipline at the intersection of molecular biology and in vivo imaging . It enables cellular functions to be visualized and molecular processes to be followed in living organisms in a non-invasive manner . Recently , studies on the visualization of live parasite in hosts have been conducted . Using transgenic Plasmodium parasites , pre-erythrocytic development was visualized; Plasmodium sporozoites entered hepatic cells , developed in a large schizont , and released merozoites in liver [8] , [9] . However , these techniques are not applicable to trematodes , because stable transgenic flukes are difficult to be generated . In mammalian hosts , adult forms of trematodes consume large amounts of glucose to generate and supply energy by running the glycolytic pathway [10] . Adult schistosomes import exogenous glucose , equivalent to their dry body weight every 4 hours from host blood by using glucose transporters in their tegumental membranes [11] , [12] . In C . sinensis , glucose transporter and Na+/glucose co-transporter are expressed abundantly in the adult stage but less so in the metacercarial stage as presented in the C . sinensis transcriptome [13] . Adult C . sinensis worms uptake glucose to produce energy in the anaerobic environment of the bile duct [14] . Therefore , we expected that C . sinensis could be labeled with 2-deoxy-2-[18F]fluoro-D-glucose ( 18F-FDG ) , a glucose analogue used for the radiolabeling and diagnostic imaging of cancer cells [15] . Thus , by ex vivo labeling CsNEJs with 18F-FDG , we hoped their migration in the final host could be traced in vivo by positron emission tomography-computed tomography ( PET-CT ) . In vivo imaging techniques have strong merits for the noninvasive tracing on pathogens moving within tissues of living animals , as they involve minimal manipulation and/or euthanasia of animals , and allow repetitive tracking in same animals . Furthermore , as was found in the present study , these techniques make it possible to monitor the distribution and migration of CsNEJs in vivo from the duodenum to the liver or distal bowel . This study was carried out to determine how CsNEJs find their way and how rapidly they migrate to the intrahepatic bile duct by using in vitro 18F-FDG radiolabeling and PET-CT in a rabbit model .
Topmouth gudgeons ( Pseudorasbora parva ) , the second intermediate host of C . sinensis , were purchased at a fish market in Shenyang , Liaoning Province , People's Republic of China . Fishes were ground then digested in artificial gastric juice ( 8 g of pepsin 1∶10 , 000 ( MP Biochemicals Co . , Solon , OH , USA ) and 8 ml of concentrated HCl in 1 liter of water ) for 2 hr at 37°C [10] . To remove particulate matters , the digested soup was filtered through a sieve of 212 µm mesh . C . sinensis metacercariae ( 135–145 µm×90–100 µm ) were then filtered out using seives of 106 and 53 µm meshes and washed thoroughly several times with 0 . 85% saline . C . sinensis metacercariae were collected under a dissecting microscope and stored in phosphate-buffered saline at 4°C until required [10] . The metacercarial cyst wall of C . sinensis is thick and can hinder glucose diffusion . Thus to maximize radiolabeling efficiency , metacercariae were excysted and juvenile worms were liberated from cysts . The C . sinensis metacercariae were excysted by treating them with 0 . 05% trypsin at 37°C for 5 minutes ( Gibco , Grand Island , NY , USA ) in 1× Locke's solution ( 150 mM NaCl , 5 mM KCl , 1 . 8 mM CaCl2 , 1 . 9 mM NaHCO3 ) , a maintaining medium of CsNEJs [16] . CsNEJs were washed 5 times with 1× Locke's solution , and used immediately . CsNEJs were divided into two groups of 10–270 juveniles each; one was of CsNEJs that excysted just before radiolabeling and the other was of the CsNEJs fasted for 24 hours . The two CsNEJ groups were radio-labeled with 18F-FDG by incubating them in 1× Locke's solution containing 74 MBq 18F-FDG at 37°C for 15 , 30 , or 60 min . After washing 3 times with 1× Locke's solution , radioactivity was measured for 10 min using a PET ( GEMINI TF , Philips Healthcare , Cleveland , OH , USA ) . Numbers of CsNEJs were counted and labeling efficiency was calculated as counts per minute ( cpm ) divided by number of the CsNEJs . Radio-labeling efficiencies of the CsNEJs in both groups were measured 3 times and significant differences were determined using the student's t-test . Rabbits ( New Zealand White , male , 2 . 2–2 . 5 kg ) were purchased from Samtako Bio Korea Inc . ( Osan , Korea ) . Rabbits were cared for and handled according to guidelines issued by Chung-Ang University College of Medicine Animal Facility ( an accredited facility ) in accordance with AAALAC International Animal Care policy . Animal experiments were approved by the institutional review board of the Chung-Ang University animal facility ( CAUMD 09-0024 ) . Gallbladder contraction and emptying time induced by cholecystokinin-8 ( CCK–8 ) varied from rabbit to rabbit . To select rabbits that responded sensitively to CCK-8 , cholescintigraphy and 99mTc-mebrofenin ( 3-bromo-2 , 4 , 6-trimethylphenyl carbamoylmethyl iminodiacetic acid ) were used . Briefly , rabbits were fasted for 12 hrs and anesthetized with a 0 . 47 mg/kg Rompun ( xylazine hydrochloride; Bayer Korea , Seoul , Korea ) and 12 . 5 mg/kg Zoletil 50 ( Zolazepam and Tiletamine; Virvac Korea , Seoul ) , intramuscular injection . 99mTc-mebrofenin ( 74 MBq ) in 0 . 5 ml volume was then administered via an ear vein to each anesthetized rabbit . When full of 99mTc-mebrofenin , gallbladders were stimulated to contract by injecting CCK–8 intravenously at 20 ng/kg every 1 min . A dynamic image was taken every 1 min for 1 hour for each rabbit . All images were obtained with a rotating dual-headed gamma camera equipped with a low-energy , high-resolution collimator ( Vertex TM , Philips Healthcare , Cleveland , OH , USA ) using a 256×256-pixel matrix at an energy range of 20% at 140 keV . Fresh CsNEJs ( n = ∼3 , 000 ) were radio-labeled with 18F-FDG by incubating them in a maintaining medium containing 74 MBq 18F-FDG at 37°C for 15 min . CsNEJs were washed 3 times with 1× Locke's solution and then placed in 500 µl of 1× Locke's solution . The procedure was conducted as follows ( Figure 1 ) . A rabbit sensitive to CCK-8 was anesthetized with 0 . 47 mg/kg Rompun and 12 . 5 mg/kg Zoletil 50 by intramuscular injection and placed in restraints in a supine position on a plastic board . A catheter ( 5F Simmons II , Cook Co . , Bloomington , IN , USA ) , equipped with a guidewire ( 0 . 035″ Radifocus® , Terumo , Tokyo ) , was inserted through the animal's mouth and its end positioned in the mid duodenum under guidance ( Axiom Artis; Siemens , Erlangen , Germany ) . The rabbit was then moved with the catheter in situ and placed in PET-CT bed . To stimulate gallbladder contraction and bile juice release , 20 ng/kg of CCK–8 was injected intravenously every minute over this experiment [17] . After 12 minutes of CCK-8 injection , 18F-FDG-labeled CsNEJs in 500 µl of 1× Locke's solution were introduced into the mid duodenum through the catheter; residual CsNEJs in the catheter were flushed into the duodenum with 0 . 5 ml of 1× Locke's solution . One transmission CT image was obtained before the introduction of the 18F-FDG-labeled CsNEJs and a dynamic PET scan then was performed over 90 min with a 3-min acquisition per frame . Finally , one static PET image was scanned for 10 min . This procedure is depicted schematically as a flow-chart in Figure 1 . All photon data were collected using a dedicated PET-CT scanner . PET images were reconstructed after applying CT-based attenuation and scattering corrections using the ordered subset expectation maximization algorithm ( 2 interations , 16 subsets ) with the point spread function . Image analysis was performed on a dedicated workstation using Extended Brilliance Workspace ( ver . 3 . 5 . 2 . 2260 , Philips Healthcare ) . A region of interest ( ROI ) was set on the whole liver in dynamic axial images while referencing corresponding coronal images and radiating photons were counted over each frame . PET images were subsequently visually evaluated for the presence of focal 18F-FDG uptake by radiolabeled CsNEJs . Migration of the CsNEJs to the intrahepatic bile ducts was estimated by semi-quantitatively analyzing photon counts from rabbit liver . To confirm migration of the CsNEJs to the intrahepatic bile ducts , adult C . sinensis were recovered from the liver of the CsNEJ-inoculated rabbits . Four weeks after image scanning , rabbits were euthanized and C . sinensis adult worms were recovered from the bile ducts by carefully squeezing liver slices . For pathologic section slides , the liver was fixed in 10% neutral formalin , processed along a routine procedure and stained with hematoxylin and eosin . As a negative control , 18F-FDG-labeled CsNEJs were inoculated into two rabbits not injected with CCK-8 . In these rabbits , bile is not released from the ampulla of Vater , neither attract the CsNEJs to the bile duct .
Fresh CsNEJs were labeled with 10 , 760 , 7 , 726 and 13 , 842 cpm/worm after incubation in radiolabeling media for 15 , 30 , and 60 min , and fasted CsNEJs were labeled with 11 , 115 , 8 , 043 , and 12 , 318 cpm/worm when incubated for 15 , 30 , and 60 min , respectively ( Figures 2A & B ) . Labeling efficiencies were similar in the two groups at all time points . For downstream experiments , fresh CsNEJs were radiolabeled with 18F-FDG at 37°C for 15 min . To determine an appropriate time point to inoculate the 18F-FDG-labeled CsNEJs in the duodenum after CCK-8 injection , gallbladder contraction and 50% bile emptying times were determined using 99mTc-mebrofenin and cholescintigraphy . After 99mTc-mebrofenin injection , radioactivity increased immediately in the gallbladder to reach a peak at about 15 min , which was maintained for over 60 min ( Figure 3A ) . When rabbits were injected intravenously with CCK-8 , 99mTc-mebrofenin was rapidly released from the gallbladder and flowed down the small intestine ( Figure 3B ) . Of the 16 rabbits tested for gallbladder contraction , 6 responded sensitively to CCK-8 . On average , it took 11 . 5 min to evacuate 50% of the gallbladder volume after the first CCK-8 injection . The rabbits responding to CCK-8 were allowed one week to recover and were then included in the in vivo imaging experiments . Under x-ray visualization and anesthesia , the end of a catheter was located in the mid duodenum ( Figure S1 ) . The rabbit was then positioned in the PET-CT bed; anesthesia was maintained with intravenous CCK-8 at a dose of 20 ng/kg every minute during PET-CT scanning . One abdominal CT image was obtained initially and then dynamic PET scanning was started . Three minutes after the initial PET scanning , the 18F-FDG-labeled CsNEJs were inoculated into the mid duodenum ( Figure 1 ) . Dynamic and static PET scans were carried out using PET-CT on migrating 18F-FDG-labeled CsNEJs in 6 rabbits , which included 2 controls . Signals emitted from the 18F-FDG-labeled CsNEJs were detected in the intestine of the 4 experimental rabbits by PET , and thus , we were able to trace CsNEJ migration by in vivo imaging . When the 18F-FDG-labeled CsNEJs were injected through the catheter , signals were detected at end of the catheter in the duodenum and along the small intestine driven by peristalsis along the distal portion of the intestine ( Figure S2 ) . Signals of CsNEJs appeared in the liver as early as 7–9 min after inoculating the 18F-FDG-labeled CsNEJs into the duodenum ( Figure 4A ) . As time elapsed , some photon spots emerged in the liver region and enlarged whereas others faded . These spots appeared to be randomly and evenly distributed in the liver regardless of lobe structure ( Figures 4A–D ) , and gradually increased in number to plateau at about 21 min after inoculation of the radiolabeled CsNEJs ( Figures 4 & 5 ) . Spots suggestive of CsNEJs moving through the common bile duct were not observed in PET-CT images . In static PET-CT images taken finally over 10 min , CsNEJs appeared to aggregate in central region of the liver ( Figures 4E & F ) . Of the CsNEJs inoculated into the duodenum , some migrated up to the bile ducts and others down to the lower bowel driven by peristalsis ( Figure S2 ) . In rabbits not injected with CCK-8 ( the negative control group ) , signals of 18F-FDG-labeled CsNEJs were only observed in the small intestine in dynamic and static PET images . At 4 weeks after the CsNEJs inoculation into the duodenum , adult C . sinensis worms were found to inhabit and to have provoked pathologic changes in the bile ducts . On average 1 , 077±806 adults were recovered from the biliary tracts of the rabbits ( Figure S3 ) .
In vivo the migration route of C . sinensis was indirectly determined by ligating the common bile ducts of hosts . Recently , live Schistosoma mansoni adults in mice were labeled with protease-activated fluorochrome or 18F-FDG and visualized , localized , and quantified using fluorescence molecular tomography or PET [18] , [19] . In the present study , we applied the methodologies and investigated PET-CT as a new in vivo imaging method for monitoring the migration of CsNEJs and their localization in the rabbit liver . The rabbits are highly susceptible to and retain the C . sinensis infections long time to evaluate impact of the infection on the hepatobiliary system . The rabbits have the biliary system similar to that of human . Distribution of C . sinensis in the liver of the experimental rabbits was proportional to volume of the liver lobes [20]–[22] . We , therefore , expected the rabbit as a reliable experimental animal model to study bile-chemotactic migration of the CsNEJs , suggesting that findings obtained from the rabbits are applicable to human . Trematodes import glucose through glucose transporter , and a large number of glucose transporters have registered in the C . sinensis transcriptome database [13] . 18F-FDG is a glucose analog tagged with isotope 18F , and is transported into cytoplasm by glucose transporters in cell membrane . In the cytoplasm , FDG is phosphorylated to FDG-6-phosphate by hexokinase , and FDG-6-phosphate is neither metabolized further nor able to diffuse out of cells . Thus , FDG-6-phosphate is trapped and accumulates in cells as the dephosphorylation of FDG-6-phosphate by glucose-6-phosphatase in cytoplasm is a slow process [15] , [23] . We expected that fasted CsNEJs would uptake more FDG than fresh CsNEJs because CsNEJs should have consumed their reserve energy source , primarily glucose , during fasting in glucose-free 1× Locke's solution . However , FDG uptakes in both groups were similar , suggesting that FDG moved quickly into the tegument of CsNEJs through glucose transporter by facilitative diffusion , as was observed for schistosomes [24] , [25] . During our studies , we have observed that CsNEJs move toward bile dose-dependently by chemotaxis in in vitro assays ( unpublished data ) . Based on our data and the notion that C . sinensis juveniles migrate up through the common bile duct , it was essential that bile juice is released from the gall bladder to attract CsNEJs into the common bile duct . Technetium labeled hepatobiliary radiopharmaceuticals has greatly facilitated studies on gallbladder function [26] . Since CCK-stimulated cholescintigraphy was first reported in 1979 , gallbladder emptying function has been measured by using standard cholagogic stimulus agents by biliary excretion scintiography [27] , [28] . Cholescintigraphy with 99mTc-iminodiacetic acid has been used to diagnose diseases in the biliary system , such as , bile duct obstruction , cholelithiasis , cholecystitis , and biliary fistula [29]–[31] . The gallbladder normally fills with hepatic bile during fasting and empties its contents into the duodenum in response to stimulation by CCK , either released endogenously following a meal or administered exogenously [32] . However , gallbladder emptying response to exogenous CCK varies among patients and experimental animals . In this study , gallbladder contraction and bile juice release was achieved by repeatedly injecting CCK-8 . By cholescintigraphy , 99mTc-mebrofenin was found to be released rapidly from gallbladders after CCK-8 administration . Thus , this scheme enabled us to study in vivo bile-chemotactic behavior of CsNEJs in rabbits . Using CsNEJ radiolabeling and bile excretion from gallbladder , images of CsNEJs migrating to the intrahepatic bile ducts in rabbits were obtained by PET-CT . The radiolabeled CsNEJs were inoculated into the mid duodenum , which is supposed to be an excystation site for C . sinensis metacercariae [3] , [6] . We visualized 18F-FDG-labeled CsNEJs migrating to the liver in experimental rabbits using PET-CT . The first signals of CsNEJs arriving at the liver from the duodenum were detected by dynamic PET as early as 7–9 min after inoculating CsNEJs into duodena . At 21 minutes post-inoculation , photon signals emitted from CsNEJs in liver appeared to have stabilized though their intensities undulated , which suggested most CsNEJs responsive to bile immediately migrated up to the intrahepatic bile duct . Imaging was ended with a final static PET-CT image because signals were of greater intensity than on dynamic PET images , suggesting that some CsNEJs were late to arrive and accumulated in the intrahepatic bile ducts [3] . In in vitro assays , CsNEJs showing rapid bile-taxis were promptly re-activated and moved rapidly and continuously toward bile added to assay chambers , and slow responders responded slowly ( unpublished data ) . The artificial manipulation of CsNEJs employed in this study , that is , in vitro excystation and radiolabeling , and inoculation into the duodenum , may have reduced adaptation to body temperature , chemotactic response to bile , migration to the bile duct , and survival in bile juice . To compensate for this , in the present study , 3–5 times more 18F-FDG-CsNEJs than normally usual experiment was inoculated via catheter into the duodenum . We believe that slow responders arrived late at the intrahepatic bile ducts after PET scans , and increased numbers of C . sinensis adult worms recovered from the bile ducts [3] . When filet of the fresh water fish was minced by teeth and ingested by mammalian animals including human , the C . sinensis metacercariae could be released from the filet in the stomach after 1–2 hour , and then passed down to the duodenum . Considering immediate excystation of the C . sinensis metacercariae in contact with trypsin [6] , human infection may take place within 2–3 hours after eating raw filet of the fresh water fish . We searched for photonic signals from the common bile ducts in dynamic and static PET-CT images of experimental rabbits , but found no signal . The common bile duct is narrow and CsNEJs either passed rapidly or steadily in file , and thus , only small number of juveniles ( not enough to create a PET-CT image ) was captured in a given frame . Furthermore , anatomically the common bile duct is located in the deep abdomen under the liver , which hinders emitted photons . Collectively , we report for the first time that CsNEJs were efficiently radiolabeled in vitro with 18F-FDG , and that CsNEJs migrate quickly with bile-chemotaxis to the intrahepatic bile duct as visualized in rabbits by PET-CT .
|
Clonorchis sinensis adults habituating in the bile duct cause clonorchiasis endemic in East Asian countries , in which about 15–20 million people are supposedly infected . It has previously been reported that C . sinensis metacercariae excyst in the duodenum and that the juvenile flukes migrate to the bile duct through the ampulla of Vater in 4–7 hours . Recently advanced imaging technologies have enabled visualization of movements and localizations of parasites in mammalian hosts . From present study , we found the following: newly excysted C . sinensis juveniles ( CsNEJs ) were efficiently in vitro radiolabeled with 18F-FDG since CsNEJs have glucose transporters; CCK-8-induced gallbladder contraction was various rabbit to rabbit; CsNEJs promptly recognized bile and migrated up the duodenum to reach the intrahepatic bile ducts by way of the ampulla of Vater and the common bile duct as early as 7–9 minutes after inoculation . Some CsNEJs responding slowly to the bile delayed arriving at the distal bile capillaries . It was visualized for the first time that the CsNEJs migrate quickly within 10–20 minutes from the duodenum to the intrahepatic bile duct . These findings provide fundamental information on the migration of parasites living in the biliary passages of mammals .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"biology",
"microbiology",
"parasitology"
] |
2011
|
Tracing of the Bile-Chemotactic Migration of Juvenile Clonorchis sinensis in Rabbits by PET-CT
|
Human T-lymphotropic virus type 1 ( HTLV-1 ) and type 2 ( HTLV-2 ) both cause lifelong persistent infections , but differ in their clinical outcomes . HTLV-1 infection causes a chronic or acute T-lymphocytic malignancy in up to 5% of infected individuals whereas HTLV-2 has not been unequivocally linked to a T-cell malignancy . Virus-driven clonal proliferation of infected cells both in vitro and in vivo has been demonstrated in HTLV-1 infection . However , T-cell clonality in HTLV-2 infection has not been rigorously characterized . In this study we used a high-throughput approach in conjunction with flow cytometric sorting to identify and quantify HTLV-2-infected T-cell clones in 28 individuals with natural infection . We show that while genome-wide integration site preferences in vivo were similar to those found in HTLV-1 infection , expansion of HTLV-2-infected clones did not demonstrate the same significant association with the genomic environment of the integrated provirus . The proviral load in HTLV-2 is almost confined to CD8+ T-cells and is composed of a small number of often highly expanded clones . The HTLV-2 load correlated significantly with the degree of dispersion of the clone frequency distribution , which was highly stable over ∼8 years . These results suggest that there are significant differences in the selection forces that control the clonal expansion of virus-infected cells in HTLV-1 and HTLV-2 infection . In addition , our data demonstrate that strong virus-driven proliferation per se does not predispose to malignant transformation in oncoretroviral infections .
The retroviruses HTLV-1 and HTLV-2 diverged from each other more than one million years ago [1] before becoming established in humans . They are similar in several crucial respects , with homologous genome structures that encode a number of regulatory proteins , including the pro-proliferative gene tax [2] , [3] . Both viruses are transmitted by transfer of infected lymphocytes via breast feeding , blood transfusion and sexual contact [4] . However , their geographical distributions are quite different . HTLV-1 is endemic in particular regions of Japan , sub-Saharan Africa , the Caribbean and South America [5] , whereas HTLV-2 is primarily endemic in indigenous populations in Africa and the Americas , although it can also be found among injection drug users in Europe and the United States [4] . HTLV-1 causes both inflammatory and lymphoproliferative diseases . In contrast , HTLV-2 causes little disease . By following a large cohort of HTLV-1/2-infected and seronegative individuals for almost two decades , the HTLV Outcomes study ( HOST ) detected myelopathy and other neurologic abnormalities among HTLV-2-infected subjects [6] , [7] , a finding supported by other studies [8] . HTLV-2 was also associated with an increase in both all-cause and cancer-related mortality [9] , as well as persistently elevated lymphocyte and platelet counts , suggesting chronic low-level inflammation [10] . However , no mechanistic inferences can yet be drawn . An important distinction between HTLV-1 and HTLV-2 lies in their host cell predilection . Although they use the same cellular receptors [11] , HTLV-1 preferentially infects CD4+ T-cells , whereas HTLV-2 favours CD8+ T-cells [12] , [13] . The biological basis for this difference is not clear . In vitro evidence suggests that the relative cell surface density of two host receptors , heparan sulphate proteoglycans and glucose-transporter 1 [14] , determines host cell tropism . However , in vivo studies suggest that both T-cell lineages are efficiently infected by both viruses , and that subsequent proliferation of CD4+ or CD8+ T-cells driven by HTLV-1 or HTLV-2 , respectively , leads to differential expansion of the two T-cell subsets [15] . It is known that HTLV-2 , like HTLV-1 , can immortalize human lymphocytes in culture [16] , [17] . Both HTLV-1 [18] , [19] and HTLV-2 [20] , [21] have also been shown to cause selective proliferation of certain infected T-cell clones in vivo . Although the molecular pathways by which the viral proteins drive cellular proliferation are well described [22] , the mechanistic basis of selective clonal proliferation is not understood . We have recently shown that the genomic integration site and transcriptional orientation of the provirus relative to the nearest host gene play important roles in determining selective HTLV-1 clonal abundance in vivo [23] , [24] . However , the total number of infected clones in a single host has not been accurately determined until recently . It was previously estimated that the number of clones in a typical HTLV-1-infected host was of the order of 100 [25] , and that individuals with the inflammatory disease HAM/TSP had a smaller number of more abundant clones , i . e . a more oligoclonal distribution . However we have shown [23] , [24] ( Laydon et al . , manuscript submitted ) that the total number of clones is in fact 100-fold to 1 , 000-fold greater ( median 28 , 000 ) , and that patients with HAM/TSP differ from asymptomatic carriers in that they have a larger number of clones rather than a more oligoclonal distribution . In HTLV-2 infection , neither the number nor the absolute or relative abundance of infected T-cell clones has been quantified . It has been suggested that the greater in vitro IL-2 dependency of HTLV-2-infected cells might lead to decreased clonal proliferation in vivo , which might in turn explain the difference in oncogenic potential between the two viruses [26] . In this work , we investigated natural HTLV-2 infection by quantifying the viral burden in CD4+ and CD8+ T-cells , comparing the clonal distribution of HTLV-2-infected peripheral blood mononuclear cells ( PBMCs ) to that observed in HTLV-1 , and examining the genomic environment of integrated HTLV-2 proviruses . For these purposes , we adapted our recently described high-throughput method for the identification , mapping and quantification of retroviral integration sites , which we have used previously to study host factors associated with clonal abundance , proviral expression and disease progression by tracking infected clones using the genomic coordinates of retroviral integration sites [24] , [27]–[29] ( Hodson et al . , manuscript submitted ) . This method also allows us to calculate , using the Gini index [30] , the degree of oligoclonality and the relative in vivo clonal expansion of infected clones .
To determine the relative contributions of CD4+ and CD8+ T-cells to the HTLV-2 proviral load , PBMCs from 28 HTLV-2-seropositive carriers were sorted by flow cytometry into separate CD3+ CD4+CD8− and CD3+CD4−CD8+ populations . Integration sites were then mapped and quantified in DNA extracted from both sorted and unsorted cells using our previously described [23] method . We assumed that HTLV-2 infects a single-positive CD4+CD8− or CD4−CD8+ T-cell , and attributed the integration sites found in the unsorted sample to either CD4+ or CD8+ cells based on the sorted sample in which they were found more frequently . Across all samples , 50% of sites ( representing >99% of proviruses ) could be attributed in this manner . Although the number of CD4+ cells isolated by flow sorting exceeded the number of CD8+ cells in the majority of samples , HTLV-2 was identified chiefly in the CD8+ fraction , whether quantified as the number of sequencing reads or the number of proviruses ( Supplementary Figure S1 ) . In 15 out of 16 patients ( those with sufficiently high numbers of detected proviruses ) the HTLV-2 load was almost wholly confined to CD8+ cells ( mean = 96 . 3% , median = 98 . 7% ) ; a mean of only 0 . 3% ( median = 0% ) was positively attributed to CD4+ cells ( Figure 1 ) . In the remaining individual , the majority of the HTLV-2 load was found in CD4+ cells . We analysed proviral integration sites in PBMCs isolated from 28 HTLV-2-infected individuals and 16 HTLV-1-infected individuals without malignant disease . At the nucleotide level , the consensus sequence flanking HTLV-2 genomic integration sites was very similar to that reported for HTLV-1 infection [31] , with bias towards GT at positions -3 and -2 , respectively , and AC at positions 8 and 9 , respectively , across the 6 base repeat ( Supplementary Figure S2 ) . The chromosomal distribution of integration sites was similar for HTLV-1 and HTLV-2 ( Figure 2A ) : In each case the frequency of integration sites detected in certain chromosomes in vivo was remarkably greater ( e . g . chromosome 13 ) or lower ( e . g . chromosome 10 ) than expected by chance . HTLV-1 and HTLV-2 were also similar with respect to features of the genomic environment flanking the provirus . In particular , activating and repressive histone marks were similarly enriched at integration sites compared to random expectation for both viruses ( Figure 2B ) . Previously , we found that HTLV-1 integration was significantly more frequent than expected on a random basis in proximity to ChIP-seq-verified binding sites for certain transcription factors and chromatin modifying proteins , most notably STAT-1 , p53 and HDACs [24] . In vivo , proviruses also lie near these binding sites more frequently than expected by chance . In the present study , we reproduced this observation in an independent cohort of HTLV-1-infected individuals , and observed a similar integration targeting preference for HTLV-2 ( Figure 2C ) . Although the magnitude of bias toward these genomic sites was very similar for both viruses , statistical significance was lower in the case of HTLV-2 , most likely owing to the lower number of total integration sites . In an earlier study , we showed that HTLV-1 infection is characterized by very large numbers of clones ( over 4 , 000 unique integration sites [UIS] have been observed in 10 µg of PBMC-derived DNA ) and that much of the load ( in non-malignant cases ) is composed of low abundance clones [23] . We confirmed this observation here for HTLV-1 , but the clonal distribution in HTLV-2 infection showed several marked differences ( Figure 3A ) . Significantly lower numbers of unique integration sites were identified in samples from HTLV-2-infected individuals ( median 16 UIS ) than those from HTLV-1-infected individuals ( median 766 UIS; Figure 3B ) . Using the recently developed biodiversity estimator ‘DivE’ ( Laydon et al . , manuscript submitted ) , HTLV-1-infected subjects were estimated to carry a median of 31 , 710 distinct clones in the blood ( consistent with previous estimates for HTLV-1 ) , whereas HTLV-2-infected subjects were estimated to carry a median of only 976 clones ( p<0 . 001 , Mann-Whitney test; Figure 3C ) . The distribution of proviral load across identified clones also differed significantly between HTLV-1 and HTLV-2 . To compare the two viruses , we used the oligoclonality index [23] , a parameter based on the Gini index , as a measure of dispersion describing the magnitude of unevenness of a frequency distribution . The oligoclonality index ranges between 0 and 1 , where a value of 0 represents a distribution in which each clone constitutes an equal share of the proviral load , and 1 represents an upper bound where the load is effectively made up by a single clone . A median oligoclonality index of 0 . 34 was observed in non-malignant HTLV-1 carriers , consistent with our previous findings . In contrast , the oligoclonality index was remarkably variable between HTLV-2-infected individuals , and on average significantly higher than in HTLV-1-infected individuals ( median 0 . 73 , p<0 . 001 , Mann-Whitney test; Figure 3D ) . Furthermore , the proportion of singletons ( clones identified only once ) was significantly lower in HTLV-2 infection ( median = 1 . 96% ) than in HTLV-1 infection ( median = 37 . 17% , p<0 . 001 , Mann-Whitney test; Figure 3E ) . The difference in clonal distribution between HTLV-1 and HTLV-2 was also apparent when measuring the absolute abundance of each clone as copies per 10 , 000 PBMCs ( Figure 4A ) . In particular , highly expanded clones ( i . e . those that each made up more than 0 . 1% of PBMCs ) represented 20% of all HTLV-2 clones but only a fraction of all HTLV-1 clones ( 0 . 18% ) . We reported previously that the genomic environment flanking the integration site appears to play a role in determining the equilibrium abundance of a given clone in vivo [23] , [24] . The positive effect of integration within 10 kb of a RefSeq transcription start site on the abundance of HTLV-1 clones was observed here again ( p = 0 . 04 , chi-squared test for trend ) , but there was no correlation between the abundance of HTLV-2 clones and the proximity of a RefSeq gene ( Figure 4B ) . Whereas there was no significant correlation between the oligoclonality index and the proviral load in HTLV-1 infection ( p = 0 . 681 , rho = 0 . 112 , Spearman's test ) , we observed a highly significant positive correlation ( p = 0 . 0015 , rho = 0 . 599 , Spearman's test ) between these parameters in HTLV-2 infection ( Figure 4C ) . In contrast , the proviral load in HTLV-1 infection was more strongly correlated with the total number of clones ( p<0 . 001 , rho = 0 . 785 , Spearman's test ) compared to that in HTLV-2 ( p = 0 . 0019 , rho = 0 . 578 , Spearman's test; Figure 4D ) . To test whether the expanded clones observed in HTLV-2 infection were long-lived , we analysed integration site content in samples collected at an earlier time point ( range = 7 . 5 to 14 . 4 years , median = 9 . 9 years ) from 10 of the HTLV-2-infected individuals in our cohort . The vast majority of the load ( median = 96% ) at the later time-point was represented by clones already present at the earlier time-point ( Figure 5A ) . These clones did not change significantly in terms of their relative abundance; those representing >1% of the load at the later time-point were significantly more likely to make up >1% of the load at the earlier time-point and vice versa ( p<0 . 001 , OR = 7 . 73 , Fisher's exact test; Figure 5B ) .
Both HTLV-1 and HTLV-2 infect the susceptible host by the same routes , and propagate within the host by the same two non-mutually-exclusive routes: the infectious route , which results in proviral integration at a new genomic site; and the mitotic route , where the provirus is replicated passively when the infected cell undergoes DNA replication and mitosis . It therefore benefits these viruses to drive proliferation of the infected cell . Indeed , the Tax proteins from both HTLV-1 and HTLV-2 have been shown in vitro to accelerate progression through the cell cycle , inhibit apoptosis and transform cells [3] , [32] , [33] . Consistent with these in vitro observations , we showed previously using metabolic labelling that cells spontaneously expressing the HTLV-1 Tax protein ex vivo proliferate faster in vivo [34] . HTLV-1 infection causes a chronic or acute T-lymphocytic malignancy in up to 5% of infected individuals [35] , [36]; however , HTLV-2 is not unequivocally linked to a T-cell malignancy . Compared with the HTLV-1 Tax protein , HTLV-2 Tax is more dependent on IL-2 for the transformation of cells in culture [37] . This observation led to the suggestion that HTLV-2 would cause less in vivo proliferation of infected cells than HTLV-1 , which in turn would decrease the oncogenic potential of the virus [26] . However , two lines of evidence go against this model . The first is the recent finding that HTLV-2 Tax has a greater in vitro immortalization capacity than HTLV-1 Tax in primary human T cells [38] , [39] . The second is the finding reported here that HTLV-2 infection in vivo results in a small number of highly expanded T-cell clones ( Figure 3 ) . Although non-malignant HTLV-1 infection can result in the preferential expansion of certain clones , including clones that contain the provirus at genomic sites with particular characteristics [23] , [24] , HTLV-2 infection is capable of driving infected T-cells to proliferate selectively , generating clones which are often more highly expanded than those observed in non-malignant HTLV-1 infection ( Figure 4A ) . The resulting clone frequency distribution in HTLV-2 infection is more similar to that observed in Adult T-cell Leukemia/Lymphoma ( ATLL ) patients than in non-malignant HTLV-1 infection ( compare , for example , Figure 3D here with Figure 2B in [23] ) . That is , HTLV-1 infection is characterized by a large number of distinct clones in the circulation , while HTLV-2 infection is confined to a small number of highly expanded clones ( Figure 3 ) The host genomic environment flanking HTLV-2 integration sites in vivo closely resembles that of HTLV-1 integration sites . Similarities are evident at the nucleotide and the chromosome levels , and when examining particular genomic features known to be more frequent in proximity to HTLV-1 integration sites than expected by chance [24] ( Figure 2 ) . This observation is likely to result from the similarity between the HTLV-1 and HTLV-2 integrases , leading to shared targeting preferences during initial infection and integration . Since the samples analysed here ( for both viruses ) were drawn from patients infected for many years , this result suggests that there are also similar selection forces acting upon the infected cells in vivo in the two respective infections; the major selection force that differs between infected individuals is likely to be the acquired immune response , in particular cytotoxic CD8+ T-lymphocyte ( CTL ) activity [40] . The selection forces that act upon HTLV-1-infected clones have been the subject of many previous studies [41] . Two principal opposing forces govern the abundance of each clone in vivo: the ability of the clone to proliferate ( e . g . through Tax-mediated cell proliferation or through antigen-mediated activation ) , and the susceptibility of the clone to elimination by CTL-mediated cell killing . We recently demonstrated that the genomic environment at the proviral integration site is associated with the clonal expansion in HTLV-1-infected individuals and with the tendency of a given clone to express the HTLV-1 Tax protein [23] , [24] . Further , cells that spontaneously expressed the HTLV-1 Tax protein belonged more frequently to low-abundance clones in vivo compared with non-Tax-expressing cells , suggesting that the expression of this dominant T-cell immunogen [42] , [43] limits proliferation in vivo . We suggest that this limited proliferation of Tax-expressing cells is due to counter-selection by the abundant , chronically activated Tax-specific CTLs . The immune response to HTLV-2 proteins is less well understood . Oliveira and colleagues showed that high frequencies of CTLs specific for HTLV-2 Tax can be found in the circulation of HTLV-2 carriers [44] . Thus , while the genomic site preferences for HTLV-1 were mirrored in HTLV-2 infection , it is surprising that the integration site plays a lesser role as a determinant of clonal expansion in HTLV-2-infected individuals ( Figure 4B ) . Although abundant clones ( absolute abundance >10 cells per 10 , 000 PBMCs ) represent only a small fraction of all infected clones in HTLV-1 infection , they represented approximately 20% of all HTLV-2 clones in this study ( Figure 4A ) . There are two possible explanations for this discrepancy: either HTLV-2 clones are not controlled as efficiently as HTLV-1 clones by the immune response , or there is an unidentified driver ( in addition to the virus itself ) that determines the proliferation of HTLV-2 clones . One potential additional driver is antigenic stimulation of the infected cells . Regardless of the forces that drive this vigorous clonal proliferation of HTLV-2-infected cells , the observed correlation between the oligoclonality index and proviral load in HTLV-2 infection ( a correlation not observed in non-malignant HTLV-1 infection , see Figure 4C ) suggests that clonal proliferation plays a greater role in determining the viral burden of HTLV-2 than it does in HTLV-1 . Conversely , in HTLV-1 infection , the total number of clones is more important as a determinant of viral burden ( Figure 4D ) , consistent with previous observations [23] , [28] . We conclude that the proviral load in HTLV-1 infection – and therefore the risk of both inflammatory and malignant disease – is determined primarily by the extent of infectious spread of the virus , and that oligoclonal proliferation per se , contrary to previous belief , does not contribute to HTLV-1-associated diseases . It remains an important question whether infectious spread is mainly confined to the early stages of infection or whether it persists indefinitely , with continual formation and destruction of many low-abundance clones . Work is now in progress to quantify the ratio of infectious spread to mitotic spread in these two infections . Given the observations that HTLV-2-infected clones proliferate to a greater extent than many HTLV-1-infected clones in vivo , and that HTLV-2 shows transformation potential in vitro [3] , [45] , it is puzzling that HTLV-2 is not associated with a T-cell malignancy . One potential explanation is that the expansion of HTLV-2 clones in vivo is short-lived , and that major clones succeed each other over time . To test this possibility , integration sites and clonal distribution in PBMCs taken at an earlier time-point in the infection were compared to those identified ∼10 years later in 10 HTLV-2-infected individuals . Although there was no significant difference in the oligoclonality index or the total number of clones after correcting for the different numbers of proviruses detected ( not shown ) , the bulk of the load at the later time-point ( median = 96% , Figure 5A ) belonged to clones already present at the earlier time-point , and that these clones principally maintained their expanded state over that period of time ( Figure 5B ) . A similar observation was made in HTLV-1 infection by Gillet et al [23] . These observations reinforce the conclusion reached above that oligoclonal proliferation of infected T-cells in vivo does not in itself predispose to malignant disease in these retroviral infections . To determine the proportion of the proviral load carried by CD4+ and CD8+ T-cells , we analysed the relative contribution of each identified clone to the load to calculate the cumulative contribution of each cell type . Using this method we found that HTLV-2 was primarily restricted to CD8+ T-cell clones ( mean = 96 . 3%; Figure 1 ) . The main limitation of this method is the sampling probability - i . e . the chance of redetection . If a clone is detected in unsorted PBMCs but not in the sorted sample it is not possible to attribute the clone to either CD4+ or CD8+ cells , or to distinguish the lack of redetection by chance from the possibility that the load is present in a different cell type ( e . g . B cells [46] ) . However , since more abundant clones are more likely to be redetected in repeated experiments ( N . Gillet and H . Niederer , unpublished observations ) , the fact that 70% of high-abundance clones ( each constituting >1% of the load ) were redetected compared with only 40% of low-abundance clones ( each constituting <1% of load ) in one of the cell-sorted populations suggests that low clone abundance , rather than an untested cell type , was responsible for the small fraction of the load not identified within either the CD4+ or CD8+ T-cell compartments . It remains unclear what controls the proliferation of HTLV-2 clones in vivo , and what mechanisms underlie the difference in oncogenic potential of the two viruses . Possible factors include differences between HTLV-1 and HTLV-2 in the actions of the respective Tax protein [47] or the antisense proteins HBZ ( HTLV-1 ) and APH-2 ( HTLV-2 ) [48] . Also , CD4+ and CD8+ T cells may differ in their susceptibility to malignant transformation . A useful insight may be found by examining the clonal distribution of CD8+ cells infected with HTLV-1 . A comparison between the clonal distribution of HTLV-1 and HTLV-2 in CD8+ cells may enable a distinction between effects due to infected cell phenotype and effects due to the differences in viral genome . This project is currently underway . In summary , we report a comprehensive analysis of integration site preferences and clonal distribution in HTLV-2 infection . By comparison with similar data from HTLV-1-infected individuals , our results suggest an important distinction between virus-driven cell proliferation and virus-driven malignancy , and strengthen the conclusion that oligoclonal proliferation per se does not predispose to malignant transformation .
UK blood samples were obtained through the Communicable Diseases Tissue Bank at Imperial College , approved by the UK National Research Ethics Service ( NRES reference 09/H0606/106 ) . Samples , with data linkage , were donated by HTLV-1 or HTLV-2-infected subjects attending the National Centre for Human Retrovirology , St Mary's Hospital , Imperial College Healthcare NHS Trust , London after giving written informed consent . HOST Study approved by the University of California San Francisco Committee on Human Research . Cryopreserved PBMCs from 16 HTLV-1-infected and 28 HTLV-2-infected individuals were used in this study ( Supplementary Table S1 ) . Twenty-six HTLV-2-infected subjects were recruited to the HOST cohort , a long-term study of outcomes of HTLV-1 and HTLV-2 infection [49] . Two HTLV-2 and all 16 HTLV-1-infected subjects were Communicable Diseases Tissue Bank donors . Proviral load data on the HTLV-1-infected individuals were reported previously [50] . Genomic integration sites in 6 of the 16 HTLV-1-infected individuals were also studied previously , albeit at a distinct time-point in each case [23] . All DNA extractions were carried out using a DNeasy Blood & Tissue Kit ( Qiagen ) according to the manufacturer's protocol . HTLV-2 proviral load was quantified as reported elsewhere [51] . We used the proviral load for each patient at the nearest available time-point , because the proviral load was not always known for each at the same time-point at which clonality was analysed , and because HTLV-2 proviral load is reported to be stable over time [52] . Consistent with this assumption , our findings were not qualitatively altered by using a proviral load measurement from a different time-point . Cells were thawed and washed , then surface-stained with directly-conjugated monoclonal antibodies specific for CD3 , CD4 and CD8 . Flow cytometric sorting was conducted to high purity ( >98% ) using a custom-modified FACSAria II ( BD Biosciences ) . Lymphocytes were pre-gated on CD3 , then sorted as CD4+CD8− and CD4−CD8+ populations . Data were analysed with FACSDiva v6 software ( BD Biosciences ) . For each sample , DNA was extracted from both sorted populations and from a separate aliquot of unsorted PBMCs . Ligation-mediated polymerase chain reaction ( LM-PCR ) primer binding site sequences were determined from Sanger-sequenced PCR amplicons ( HTLV-1 primers: 5LTRfw –CTCGCATCTCTCCTTCACG , 5LTRrev – CTGGTGGAAATCGTAACTGGA; HTLV-2 primers: H2LTRfw – GACTCACCTTGGGGATCCAT , H2LTRrev – TTAGCCAAATGGGCGTTTTA ) . Identification of integration sites was performed via LM-PCR followed by high-throughput sequencing as described previously [23] , using the HTLV-2-specific forward primers: H2B3 – AAGGGCTAGGGCTTCCTGAACCTC and H2B5 – CTATAGGCAGGCCCGCCCCAGGAG ( or variants thereof according to defined LTR polymorphisms ) . Prepared libraries were mixed and sequenced using either a Genome Analyzer II or a HiSeq System ( Illumina ) . The resulting sequences were aligned to the human genome reference ( hg18 , excluding haplotypes , randoms and mitochondrial DNA ) and HTLV-1/2 upstream sequence using the eland_pair implementation of CASAVA 1 . 8 . 2 ( Illumina ) . Integration sites were quantified by enumerating unique shear sites as described previously [23] . Bins of absolute clonal abundance were determined by the number of copies per 10 , 000 PBMCs ( i . e . relative clonal abundance multiplied by proviral load ) . In flow-sorted samples , clones were attributed exclusively either to the CD4+ or CD8+ cell population . Six clones were initially detected in both CD4+ and CD8+ cells; in these cases the clone was ascribed to the cell type with the greater number of proviruses of that clone . In silico sites were derived by random selection of 190 , 000 sites from the human genome ( hg18 ) . To eliminate any potential bias due to alignment limitation , the DNA sequences at those sites were generated using Galaxy [53] , [54] and back-aligned to the human genome using the same pipeline . Annotations to the human genome ( hg18 ) were retrieved as described previously ( see Table S3 in [24] ) and compared to integration sites using the R package hiAnnotator ( http://malnirav . github . com/hiAnnotator ) , kindly provided by Nirav Malani and Frederic Bushman ( University of Pennsylvania , Philadelphia , USA ) . The DivE estimator was used to estimate the total number of clones , in addition to those observed . DivE involves fitting many mathematical models to individual-based rarefaction curves ( Laydon et al . , manuscript submitted ) . Here , individual-based rarefaction curves depict the expected number of clones against the number of infected cells sampled . Four numerical criteria are used to score each model in terms of how consistently it reproduces the total observed rarefaction curve from nested subsamples thereof . Using the geometric mean , estimates from the best-performing models are aggregated to produce the final estimate . Samples with near-linear rarefaction curve ( i . e . where the curvature was less than 0 . 1 ( Laydon et al . , manuscript submitted ) were excluded from the analysis . Such curves imply a ( biologically impossible ) linear relationship between the number of infected cells and the number of clones , which is indicative of severe under-sampling . Prohibitively small sample sizes of less than 150 proviruses were also excluded from the analysis . DivE requires an estimate of the number of cells in the blood Nblood , for which we assume: ( i ) a circulating blood volume of 5L; ( ii ) a PBMC count of 3×109 L−1; and ( iii ) each infected T-cell carries a single copy of the provirus [55] . The number of PBMCs is thus assumed to be 5×3×109 . Proviral load ( PVL ) is defined as the number of viral copies per 100 PBMCs . Therefore , Nblood is given by PVL×5×3×109 . Statistical analysis was carried out using R version 2 . 15 . 2 ( http://www . R-project . org/ ) . The Gini coefficient was calculated using the reldist R package ( [56]; http://CRAN . R-project . org/package=reldist ) . Two-tailed non-parametric tests including the Mann-Whitney and Fisher's exact test were used for all comparisons .
|
The two human retroviruses HTLV-1 and HTLV-2 are similar in their structure , replication cycle and the manner through which they spread between and within individuals . They differ in their preferred host T-cell type and in their possible clinical outcomes . HTLV-2 has not been linked with a specific disease , whereas HTLV-1 infection can cause leukemia and profound neuropathology . It is well established that HTLV-1-infected cells undergo clonal expansion in infected individuals , but little is known about clonality in HTLV-2 infection . In this work , we demonstrate that the extent of HTLV-2-infected cell expansion significantly exceeds that of HTLV-1-infected cells in healthy carriers , approximating instead to that observed in patients with HTLV-1-associated leukemia . Furthermore , we show that HTLV-2 characteristically resides in a small number of expanded clones that persist over time , and that the degree of oligoclonality significantly correlates with viral burden in HTLV-2-infected individuals . These results highlight the distinction between in vivo clonal proliferation and malignant transformation , and suggest that the infected cell type may be a more important determinant of clinical outcome in retroviral infections .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"virology",
"biology",
"microbiology"
] |
2014
|
Clonality of HTLV-2 in Natural Infection
|
The regulatory roles of sphingolipids in diverse cell functions have been characterized extensively . However , the dynamics and interactions among the different sphingolipid species are difficult to assess , because de novo biosynthesis , metabolic inter-conversions , and the retrieval of sphingolipids from membranes form a complex , highly regulated pathway system . Here we analyze the heat stress response of this system in the yeast Saccharomyces cerevisiae and demonstrate how the cell dynamically adjusts its enzyme profile so that it is appropriate for operation under stress conditions before changes in gene expression become effective . The analysis uses metabolic time series data , a complex mathematical model , and a custom-tailored optimization strategy . The results demonstrate that all enzyme activities rapidly increase in an immediate response to the elevated temperature . After just a few minutes , different functional clusters of enzymes follow distinct activity patterns . Interestingly , starting after about six minutes , both de novo biosynthesis and all exit routes from central sphingolipid metabolism become blocked , and the remaining metabolic activity consists entirely of an internal redistribution among different sphingoid base and ceramide pools . After about 30 minutes , heat stress is still in effect and the enzyme activity profile is still significantly changed . Importantly , however , the metabolites have regained concentrations that are essentially the same as those under optimal conditions .
Any substantial increase in temperature has a direct effect on the macromolecules in a cell . Among them , proteins and lipids are most strongly affected . Nucleic acids can denature upon exposure to heat , but this process requires much higher temperatures of about 75°C–100°C [9] , which are outside the realm of tolerable heat stress . Heat affects proteins in three ways . First , high temperature can modulate their synthesis from gene expression . In this context , Castells-Roca and colleagues investigated transcription rates and the stability of various mRNAs in S . cerevisiae following a temperature shift from 25°C to 37°C , and concluded that both were affected [10] . Second , processes of protein inactivation are temperature dependent . And third , heat can change a protein's folding state , which in turn may affect its function , as well as its removal by the proteasome . In particular , if the protein is an enzyme , its activity is influenced directly by its ambient temperature , according to an empirical relationship commonly called the Arrhenius effect or the Q10 effect . Lipids are major constituents of membranes , and although the effects of heat are not completely understood , it appears that changes in temperature have an impact on membrane stiffness and fluidity [11] . Jenkins and coworkers [12] were among the first to connect sphingolipids to heat stress responses in yeast , demonstrating that these lipids play several particularly important roles ( see also [13]–[17] ) . They subdivided the heat stress response into two phases . During the first phase , the cell needs to gain thermotolerance , which is at least partially accomplished with an accumulation of trehalose and the induction of heat shock proteins . Furthermore , the cell arrests its cell cycle in G0/G1 , and this arrest lasts for approximately one hour , during which time there is no growth . Once thermotolerance is achieved , the cell culture starts growing again in the second phase of the response , even if the temperature is still elevated . The first response phase is directly associated with two distinct features of sphingolipids . First , the structural characteristics of complex sphingolipids , together with sterols , contribute to the physical organization of specific membrane microdomains within membranes , called lipid-rafts . These rafts are known to be associated with membrane fluidity , protein compartmentalization , and protein sorting and trafficking through membranes ( e . g . , [18]–[20] ) . As core components of rafts , sphingolipids are thus directly involved in organizational structures with potential signaling functions , and alterations in these functions are effective at a short time scale [21] . The second role of sphingolipids in the early heat stress response is their capacity to serve as bioactive signaling molecules . This signaling function influences the regulation of the cell cycle response , nutrient uptake , and the synthesis of proteins , which can have important secondary effects , especially if heat shock proteins are not available to serve as protectors of other proteins [22] , [23] . Indeed , the groups of Ferguson-Yankey and Meier demonstrated that sphingolipid synthesis is required for an efficient initiation of translation , especially during heat stress [24] , [25] . Specifically , the translation rate is increased if sphingoid bases are synthesized and accumulate . Jenkins and collaborators [26] and Dickson and co-workers [13] showed that ceramides and other simple sphingolipids , such as dihydrosphingosine and phytosphingosine ( DHS and PHS ) , accumulate during heat stress in yeast . It appears that the short-term signaling role of sphingolipids is biphasic . In the first phase , sphingoid bases are required to regulate translation of heat shock mRNAs , a process that depends strongly on Pkh kinase , but not on Ypk kinases , which act downstream of Pkh . The second phase consists of a general increase in translation , which is dependent on the function of heat shock proteins . Without these heat shock proteins , the cell would run a severely elevated risk of protein aggregation or misfolding [25] . Sphingolipids also play roles over a longer time horizon . It has been known for a while that DHS induces the expression of a STRE-LacZ reporter gene , suggesting that the global stress element STRE can be activated by sphingolipid signals [13] . In particular , genes associated with the important trehalose stress response contain multiple copies of STRE . Knock-outs or overexpression of genes coding for the synthesis of dihydrosphingosine-1-phosphate ( DHS-1P ) show changes that resemble thermotolerant and heat sensitive yeast phenotypes , indicating that DHS-1P is an important regulator of heat stress [27] . Phytosphingosine-1-phosphate is involved with the regulation of genes required for mitochondrial respiration [28] . More generally , modulations in any of the sphingolipid enzymes cause ripple effects that change the concentrations of many sphingolipids and , possibly , the expression of a variety of genes . Futerman and Hannun [29] summarized the long-term signaling mechanisms of simple sphingolipids including sphingosine-1-phosphate , sphingosine , ceramide and ceramide-1-phosphate in yeast . Taken together it is evident that sphingolipids exert important roles within the coordinated heat-stress responses of a cell , and that these roles are pertinent over short and long time horizons . However , it is so far unclear how the cell is able to establish an appropriate sphingolipid profile very quickly in response to heat stress . To answer this question , we propose a computational analysis based on observed heat stress time courses and a dynamic model of sphingolipid biosynthesis and degradation that allows us to investigate the dynamic profiles of critical enzymes involved in the sphingolipid pathway .
Without any computational analysis , the measured data directly show which sphingolipids are apparently needed under heat stress at different points in time . Measured as absolute quantities , PHS increases by far the most in concentration , whereas PHS-P increases most relative to its baseline value . Interestingly , both adjustments are much stronger than in the corresponding dihydro-forms . For instance , the concentration of DHS-P remains very low throughout the observation period of 30 minutes ( Figure 10 ) . DHS reaches its modest peak earlier than PHS and PHS-P , whereas PHC reaches its peak later . It is difficult to discern the rationale for this timing and the differences in peak heights . What the computational analysis shown here suggests is how these observed adjustments are implemented by the cell . Initially , de novo biosynthesis increases quickly , but only for the first three or four minutes . The model actually allows us to quantify and compare the total amount of biosynthesis under optimal and heat stress conditions . Namely , we can record in the dynamic simulation the total production of 3-KDHS , while computationally omitting its degradation ( Figure 11 ) . Under optimal conditions , and with a constant influx of palmitate and serine , this accumulation is linear ( blue line ) , and considering consumption as well , the concentration of 3KDHS is constant ( results not shown ) . By contrast , under heat stress , the accumulation is faster for the first few minutes ( red line ) , but it is increasingly reduced subsequently . Considering consumption as well , the concentration of 3KDHS decreases ( results not shown ) . In the next five to ten minutes , the patterns diverge strikingly . Probably most intriguing , both the input to , and the exit from , central sphingolipid metabolism are almost completely shut down . During this time period , the cell not only counteracts the unavoidable Q10 effect in SPT , but down-regulates this enzyme to a mere residual amount , as shown in top left panel of Figure 3 . Similarly , the exit routes through the lyase and remodelase steps lose activity about 5 minutes into the heat stress ( Figure 3 ) . The second step of de novo biosynthesis , KDHS reductase , is less dramatically affected ( right panel in Figure 3 ) , but deprived of substrate . This substrate deprivation appears to be safer than enzyme down-regulation , as 3KDHS is toxic [34] and any accumulation could be dangerous . The computational deductions imply that de novo sphingolipid biosynthesis appears to be up-regulated only for the first few minutes [12] . To establish the needed changes in sphingolipid profile under heat stress , the cell appears to absorb and process residual substrate as vigorously as possible , but subsequently seems to count on the much more reliable use of existing complex sphingolipids for the generation of signaling molecules such as PHS , PHS-P and , to a lesser degree , DHS and DHC , and on a subsequent redistribution among the simple sphingolipid pools . This conclusion is based on the inferred reduction in biosynthesis after about five minutes , the shutting off of the lyase and remodelase steps , as well as three additional observations . First , IPCase ( Figure 6 ) is strongly upregulated in a sustained manner for about 15 minutes . Second , the hydroxylase , which converts DHC into PHC and DHS into PHS , loses almost all activity throughout the measured time period ( Figure 4 ) . Third , processes leading to the synthesis of complex sphingolipids , including IPC synthase and the synthesis of PI and DAG , are down-regulated after about 15 minutes ( Figure 6 ) , thereby slowing down the genesis of new complex sphingolipids from simple sphingolipids . Several of the enzymes associated with complex sphingolipids begin to become active again about 28 minutes into the heat stress , which may be a consequence of changes in gene expression . After 30 minutes , the six measured sphingolipid concentrations essentially return to their baseline levels . In stark contrast , the enzyme system has not returned to its original state , and several enzymes still exhibit an activity that is quite distinct from the profile under optimal temperature conditions . Thus , the cell , which is still under heat stress , is regaining a close resemblance of normalcy with respect to its metabolites , but this state is achieved with a significantly different flux and enzyme profile .
In this work , we have proposed a computational approach to analyze heat stress response strategies in yeast . Specifically , we have inferred how cells adjust their enzyme activities within sphingolipid metabolism , which has been demonstrated in numerous earlier reports as a heat sensitive signaling system . Using experimental measurements of metabolite concentrations following a shift in temperature , combined with a detailed dynamical model , we computationally inferred adjustments in enzyme activities that appear to be both sufficient and necessary for mounting the observed metabolic response . Rather than computing a single solution to the inverse task , we computed a comprehensive ensemble of over 4400 independent solutions and selected from among them the best 2004 solutions , based on SSE and AICc metrics . These 2004 solutions led to very similar trends in the activities of key enzymes , although not of enzymes at the periphery of the pathway system . The computed results suggest , first , that the response to heat is not achieved by drastic changes in a few “key” enzymes , but that numerous enzymes are involved . Second , the dynamic alterations in activities differ substantially in both , magnitude and timing , as well as in the general shape of the enzyme activity trends throughout the observed 30-minute time window following the initiation of heat stress . The main surprise in our results is the deduction that the changes in sphingolipid profile are apparently not achieved by sustained increases in de novo biosynthesis but through a brief initial spike , followed by the retrieval of simple sphingolipids from membrane-associated complex sphingolipids , as well as a complicated redistribution scheme among the different ceramide and sphingosine forms . While this strategy was not expected , its seems to have merit , because the cell cannot be sure that new resources are quickly available for de novo synthesis of sphingolipids , while complex sphingolipids such as IPC , MIPC and M ( IP ) 2C are integral components of membranes and therefore always available , with the possible exception of the most deprived situations . Thus , it seems that the cell sacrifices some of its membrane structures and recreates them once the stress situation is under control . This sacrifice , however , is not very substantial , as the concentrations of complex sphingolipids change very little during the heat stress response ( Figure 2 ) . These results are consistent with experimental finding of Jenkins et al . [12] , who studied different roles of sphingolipids during the heat stress response . Using isotope labeling , they showed that sphingoid bases and ceramides increase early on via de novo synthesis , but that IPC , MIPC and M ( IP ) 2C remain essentially constant over a period of more than one hour . Wells et al . [17] also studied the formation of ceramide in response to heat stress and , using labeled phosphosphingolipids , and concluded that ceramide formation from IPC , MIPC , and M ( IP ) 2C through the IPCase reaction was unlikely . However , the concentration profiles these authors observed were very different from those obtained by Cowart et al . [28] , which we used here . In particular , under Wells' 39°C treatment , ceramide remained elevated at a level five times its baseline throughout the two-hour measurement period . Outside the fact that these authors studied a temperature shift from 24°C to 39°C , the differences in concentration profiles to those used here ( Figure 1; Cowart et al . [28] ) remain unexplained . Although the computational results were obtained without any particular assumptions , some uncertainties are associated with the fact that many of the intermediate sphingolipids had not been measured and that the mathematical approach may not have revealed the one truly optimal solution . For instance , all results are obtained from large-scale simulations with a dynamical model that has been validated to some degree but could certainly be improved . Given the present data , it is unlikely that further simulations of the same type as shown here would lead to different results . However , if other metabolite concentrations could be measured , or if it were possible to determine some internal metabolic fluxes independently of the metabolite concentrations , the degree of reliability of our results would greatly increase . The study presented here elucidates a systemic strategy with which the cell establishes the observed sphingolipid profile , but it does not address the specific roles of the various sphingolipids in the heat stress response . Interestingly , some of the simple sphingolipids that are known to have signaling roles do not change all that much , while others do . In particular , DHS , which activates the stress element STRE in the expression of stress related genes , maximally rises to only about twice its normal level , about 5 minutes into the heat stress . Apparently , this increase is sufficient . By contrast , PHS-P , which was recently identified as an important gene regulator , rises to a level that corresponds to almost 10 times its baseline level and exhibits a sustained response that lasts over 20 minutes . PHS rises to a four-fold level . No direct signaling role is known , and it may just be that this compound is needed as a precursor of PHS-P . The experiments generating the data used here exposed the cells to persistent heat stress . At the end of the 30-minute observation period , all six key sphingolipids have essentially returned to their normal levels , except for DHC , which still seems to be very slightly elevated . By contrast , many of the enzyme activities are not “back to normal . ” Expressed differently , the cell manages to mount a strong transient response , which is known to lead to longer-term genomic responses . Subsequently , within a total of just 30 minutes , it is able to adjust its catalytic machinery to the persistent heat conditions in such a manner that the fluxes exhibit a distinctly different activity pattern which , nevertheless , re-establishes a favorable metabolic state that is remarkably close to that under optimal conditions . Our focus on sphingolipids sheds light on just one aspect of the well-coordinated , complex responses with which yeast adjusts to a new environmental condition . Nonetheless , this particular aspect is of special interest , as the roles of sphingolipids and their biosynthetic pathways have been preserved throughout evolution , from yeast to humans , where they are involved in numerous differentiation and disease processes ( e . g . , [35]–[38] ) .
The data , previously obtained in one of our labs , were described in the literature ( see Supplements of [28] ) . They consist of duplicate 30-minute time courses of six key sphingolipids , collected following a step increase in temperature from 30°C to 39°C . Specifically , changes in metabolite concentrations were measured at baseline ( t = 0; normal temperature ) and at 5 , 10 , 15 , 20 , 25 , and 30 minutes of heat stress . We used these measurements , averaged the duplicates , and then applied a smoothing spline technique to interpolate the trend of each time course so that concentration values at 31 time points ( 0 , 1 , … , 30 minutes ) became available for each sphingolipid . The smoothed transients are shown as absolute concentrations in Figure 11 ( see also Figure 1 for fold changes , which shows the smoothed data as symbols , along with a model fit based on averaged enzyme activities ) . For our computational analysis we used relative changes in each sphingolipid with respect to the baseline steady state before heat stress , which we directly obtained from the time series measurements , and scaled these with steady-state values , which were described in earlier work [8] , to obtain actual concentrations . The biosynthesis , metabolic conversions , and degradation of sphingolipids constitute a complex , highly regulated pathway system ( Figure 9 ) that exceeds intuitive capabilities and suggests computational modeling for quantitative systemic analyses . Over the past decade , we have developed a series of such models using a General Mass Action ( GMA ) formulation within the modeling framework of Biochemical Systems Theory ( BST ) [6]–[8] , [39] . Because these models have been described in detail elsewhere , we can keep their description here to a minimum . The simple and complex sphingolipids , as well as other pertinent metabolites , are represented in the model as dependent variables , each of which satisfies an ordinary differential equation ( ODE ) . Each ODE contains representations of the processes that produce or degrade this metabolite . According to the tenets of BST , each process is represented as a product of power-law functions , which consists of a rate constant and of every variable directly affecting this process , raised to an exponent , called a kinetic order . Variable names and equations are presented in Text S1 and an SBML implementation can be found in the file Model S1 . As an example for how to design a system equation , consider the dependent variable , which represents dihydrosphingosine ( DHS ) . This metabolite is generated from three possible sources . First , KDHS reductase ( ) catalyzes the reduction of 3-keto-dihydrosphingosine ( KDHS; ) . The formulation of this process consists of a rate constant , which is multiplied by , raised to the kinetic order , and by , raised to the kinetic order . Thus , the reduction process is modeled as . Second , DHS can be produced from dihydrosphingosine-1-phosphate ( DHS-P; ) , a process catalyzed by sphingoid 1-phosphate phosphatase ( ) . In analogy to the first process , this step is represented with its own rate constant , as well as the substrate and enzyme , which are both raised to appropriate kinetic orders . Third , dihydroceramide alkaline ceramidase ( ) converts dihydroceramide ( DHC; ) into DHS , and this process is formulated in an analogous manner . DHS is subject to three possible metabolic fates , namely through the ceramide synthase reaction toward DHC , through the 4-hydroxylase reaction toward phytosphingosine ( PHS ) , and through the sphingoid base kinase reaction toward DHS-P . Taken together , the ODE equation describing the dynamics of DHS contains three influx terms and three efflux terms as shown in Eq . ( 1 ) . ( 1 ) All differential equations for dependent variables are formulated in this manner . Values for all parameters were determined from the literature [8] , [40] . The complete model consists of 25 ordinary differential equations , including those representing the six key sphingolipids of interest here , namely dihydrosphingosine , dihydroceramide , dihydrosphingosine 1-phosphate , phytosphingosine , phytosphingosine 1-phosphate and phytoceramide . The model furthermore contains 41 independent variables , which represent enzyme activities and metabolites such as ATP , palmitate , acetate and phosphoserine , which were assumed to be constant or considered unaffected by the dynamics of the pathway system . The model was rigorously tested and validated against data not used for model construction [7] . It was also recently combined with a model of the sterol pathway , which has relevance for the composition of membrane rafts [39] . An SBML version of the model can be found in zip file Model S1 . As stated at the beginning of the Results section , it is our task to infer from the measured metabolite time courses which enzymes have to be altered dynamically , and by how much , in order for the model to generate the observed time-dependent metabolic profile ? Mathematically , this inverse problem is underdetermined and furthermore complicated by the fact that the pathway is described by a system of nonlinear differential equations , as discussed before . If we were only concerned with a baseline steady state and the move of the system to a new steady state appropriate for heat stress conditions , we could use methods of linear algebra and pseudo-inverses , as we have demonstrated elsewhere [32] . However , here we are interested in the entire trajectories between stimulus ( i . e . , the beginning of heat stress ) and the cell's metabolic adjustments over 30 minutes . We solved this dynamic inverse problem with an iterative , piecewise optimization approach . Specifically , we estimated optimal enzymatic profiles by minimizing the distance between the smoothed sphingolipid data and the simulation results at each time point , with 1-minute time intervals , from 0 to 30 minutes . At each time point , the optimization engine searched for the best set of enzyme activities , which were modeled as independent variables . To satisfy the specified objective function , we algorithmically minimized the distances between the six observed sphingolipid concentrations and the solutions produced by each trial set of independent variables . We executed this strategy 4144 times , using different random values for initial settings . We then selected the 2004 best models based on residual errors ( SSEs ) . In order to test the performance of this metric , we also selected models based on the Akaike criterion ( AICc ) , and both criteria produced very similar results . Please see Text S1 for a detailed comparison of results using these two criteria . Subsequently , scanning all solutions throughout the 30-minute time period yielded dynamic alteration profiles in all enzymes as well as corresponding metabolite profiles that were consistent with the observed profiles throughout the experimental time period . Further details of this procedure are presented in the Text S1 . Each optimization run produced a dynamic enzymatic profile throughout the time period from 0 to 30 minutes . Due to the randomization of initial values and to the fact that the system is underdetermined , the solutions from different runs were different . Thus , instead of searching for a single unique solution , we studied an entire large ensemble of solutions and asked whether the solutions would reveal consistent trends of enzymatic profiles with in the potentially large solution space . Indeed , the overall result of this strategy was a set of surprisingly tight ranges for the key enzymes of sphingolipid biosynthesis .
|
Sphingolipids play dual roles by serving as components of membrane rafts and by regulating numerous key cell functions . Although sphingolipids have been studied extensively , the details of their functioning are difficult to understand , because their synthesis , pathways of inter-conversion , and utilization constitute a complex , dynamically changing system . We analyze the role of yeast sphingolipids in the response to heat stress . Data show that the profile of these lipids changes almost immediately with the initiation of the stress , but it is a priori unclear how this response is organized . Using experimental data , a sophisticated dynamic model , and a novel optimization strategy , we show how changes in enzyme activities are temporally organized . Intriguingly , the results show that the cells take up as much material as possible in the first few minutes of heat stress and then shut down entry and exit routes of the biosynthetic pathway system . After about 30 minutes , when heat stress is still in effect , the enzyme activity profile is still significantly changed , but the metabolites have regained concentrations that are essentially the same as those under optimal conditions . The results demonstrate how novel insights are achievable with an effective combination of experimental and theoretical research .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"complex",
"systems",
"systems",
"biology",
"biochemistry",
"cellular",
"stress",
"responses",
"mathematics",
"theoretical",
"biology",
"biochemistry",
"simulations",
"metabolic",
"pathways",
"applied",
"mathematics",
"biology",
"molecular",
"cell",
"biology",
"metabolism",
"lipid",
"metabolism"
] |
2013
|
Coordination of Rapid Sphingolipid Responses to Heat Stress in Yeast
|
Sporotrichosis is a subcutaneous mycosis caused by pathogenic species of the Sporothrix genus . A new emerging species , Sporothrix brasiliensis , is related to cat-transmitted sporotrichosis and has severe clinical manifestations . The cell wall of pathogenic fungi is a unique structure and impacts directly on the host immune response . We reveal and compare the cell wall structures of Sporothrix schenckii and S . brasiliensis using high-pressure freezing electron microscopy to study the cell wall organization of both species . To analyze the components of the cell wall , we also used infrared and 13C and 1H NMR spectroscopy and the sugar composition was determined by quantitative high-performance anion-exchange chromatography . Our ultrastructural data revealed a bi-layered cell wall structure for both species , including an external microfibrillar layer and an inner electron-dense layer . The inner and outer layers of the S . brasiliensis cell wall were thicker than those of S . schenckii , correlating with an increase in the chitin and rhamnose contents . Moreover , the outer microfibrillar layer of the S . brasiliensis cell wall had longer microfibrils interconnecting yeast cells . Distinct from those of other dimorphic fungi , the cell wall of Sporothrix spp . lacked α-glucan component . Interestingly , glycogen α-particles were identified in the cytoplasm close to the cell wall and the plasma membrane . The cell wall structure as well as the presence of glycogen α-particles varied over time during cell culture . The structural differences observed in the cell wall of these Sporothrix species seemed to impact its uptake by monocyte-derived human macrophages . The data presented here show a unique cell wall structure of S . brasiliensis and S . schenckii during the yeast parasitic phase . A new cell wall model for Sporothrix spp . is therefore proposed that suggests that these fungi molt sheets of intact cell wall layers . This observation may have significant effects on localized and disseminated immunopathology .
Sporotrichosis , a subcutaneous mycosis that can also be present in disseminated and extracutaneous forms , was attributed for a century to a single etiological agent—Sporothrix schenckii . The disease was described in 1898 , and the parasitic yeast phase of this thermodimorphic fungus was isolated in Brazil , in 1907 [1] . Molecular studies have since demonstrated that a complex of numerous phylogenetic species , that contains new cryptic pathogenic species [2] . This classification was later refined , and S . mexicana was placed in a separate environmental clade ( the Sporothrix mexicana complex ) [3] . S . schenckii , S . brasiliensis . S . globosa and S . lurei were recently classified in the pathogenic clade of the Sporothrix genus [4] . Two emerging pathogenic species , S . brasiliensis and S . globosa , therefore have significant epidemiological importance in distinct geographical regions [5–7] . Interestingly , S . brasiliensis and S . schenckii , but not S . globosa , are associated with zoonotic sporotrichosis [7 , 8] . Moreover , S . brasiliensis is the causative species of the major zoonotic outbreak of sporotrichosis reported in the literature [5] . The number of cases due to cat-transmitted sporotrichosis in Brazil , only registered at the Oswaldo Cruz Foundation , is over 4000 in cats ( Felis catus ) and in humans [9] , but the total number of cases is unknown and underestimated . The real numbers are even higher because sporotrichosis is not compulsorily reported in Brazil or worldwide . Additionally , cutaneous sporotrichosis can be misdiagnosed ( sporotrichoid lesion/pattern ) as cutaneous leishmaniasis , cutaneous tuberculosis and other cutaneous infections ( http://medical-photographs . com/231-sporotrichoid-lesions . html ) . In addition to its epidemiological importance , S . brasiliensis is less susceptible to the azole class of antifungals [10–12] and exhibits a higher virulence profile in a mouse model than S . schenckii clinical isolates [13 , 14] . Accordingly , severe clinical cases in recent literature were attributed to S . brasiliensis infection , including fatal cases [15–17] . Very little is known about the biology of Sporothrix spp . , and few virulence factors have been reported [18] . The genomes of S . schenckii and S . brasiliensis have 97 . 5% similarity [19] , but evidence suggests that differences in protein expression in these fungal pathogens is significant [20] . Proteomic studies have shown that the major cell wall antigen of S . schenckii , Gp70 , has a 60 kDa isoform in S . brasiliensis and is absent from non-pathogenic environmental species [20] . This evidence reinforces our hypothesis that important biological differences can exist between the pathogenic species of the genus Sporothrix . However , the degree of similarity of the cell surfaces of these species remains to be determined . The cell wall is the outermost structure of fungal cells and is the first point of contact with the host upon infection and colonization . Knowledge of cell wall structure and organization that are unique to fungal pathogens is of key importance in understanding fungal pathogenesis [21] . Furthermore , understanding cell wall compositions can aid in unveiling specific mechanisms triggered by pathogen-associated molecular patterns ( PAMPs ) and the corresponding pathogen recognition receptors ( PRRs ) [22] . The cell wall of S . schenckii is composed mainly of β-glucans ( 1–3 , 1–6 , and 1–4 linkages ) , chitin [23] and a peptido-rhamnomannan [24] , but the cell surface structure and sugar composition of other Sporothrix pathogenic species remain unknown . In the present work , the S . brasiliensis cell wall was studied at the biochemical and structural level and compared with that of S . schenckii . As a result , a novel cell wall model for the dimorphic fungi Sporothrix spp . is proposed . Cell wall structure and organization were investigated during the different growth phases of both species , as was the impact of the differences in cell wall organization on host recognition .
Two S . schenckii strains , ATCC MYA4820 and ATCC MYA4822 , and two S . brasiliensis strains , ATCC MYA4823 and ATCC MYA4824 , were used in this study . Two strains , MYA 4820 and MYA4823 , are clinical isolates from the same endemic area of Rio de Janeiro State , Brazil [13] . To obtain the yeast parasitic phase , conidia of each strain were inoculated into YPD medium at pH 7 . 8 and grown for 7 days at 37°C under orbital agitation . A 1 mL sample of this pre-inoculum was inoculated into fresh YPD medium and cultivated for 3 to 10 days under the same growth conditions . Yeast cells were grown for either 4 or 10 days and diluted to 1 x 107 cells/mL in DMEM for the macrophage interaction assays . Samples were prepared by high-pressure freezing with an EMPACT2 high-pressure freezer and rapid transport system ( Leica Microsystems Ltd . , Milton Keynes , United Kingdom ) . After being frozen , cells were freeze-substituted in substitution reagent ( 1% [wt/vol] OsO4 in acetone ) with a Leica EMAFS2 . Samples were then embedded in Spurr’s resin , and additional infiltration was provided under a vacuum at 60°C before samples were embedded in Leica FSP specimen containers and polymerized at 60°C for 48 h . Semithin survey sections , 0 . 5 μm thick , were stained with 1% Toluidine Blue to identify areas containing cells . Ultrathin sections ( 60 nm ) were prepared with a Diatome diamond knife on a Leica UC6 ultramicrotome and stained with uranyl acetate and lead citrate for examination with a JEOL 1400 plus transmission microscope ( JEOL UK Ltd . , Hertfordshire , United Kingdom ) and imaging with an AMT UltraVUE camera ( Deben , Suffolk , United Kingdom ) . The thicknesses of the inner and outer layers of the cell wall were measured using ImageJ and by averaging 100 measurements for each species ( n = 10 cells ) . Yeast cells of S . schenckii and S . brasiliensis were grown for 3–4 or 7 days in YPD broth at pH 7 . 8 as described above . The yeast cells were collected by centrifugation , washed with phosphate-buffered saline ( PBS ) and fixed with 3% formaldehyde for 15 min . The yeast cells were then washed three times with PBS and incubated in the dark for 30 min with 50 μg/mL Wheat Germ Agglutinin ( WGA ) FITC conjugate , 25 μg/mL Concanavalin A- ( ConA ) Texas Red conjugate or 25 μg/mL Calcofluor White ( CFW ) . After being washed again with PBS , the final cell pellet was resuspended in 100 μL of PBS , and a 5 μL sample was spotted onto a poly-L-lysine microscope slide . After the sample was dried at room temperature , a drop of Vectashield ( mounting medium for fluorescence ) was placed on top of the sample and was covered with a coverslip to fix the sample in place . The slides were visualized in a Zeiss confocal microscope with a 63X objective . Images were processed with ZEN SP2 imaging software . S . schenckii ( MYA 4820 ) and S . brasiliensis ( MYA 4823 ) were cultivated for 10 days as described above . Candida albicans was used as a control and was cultivated for one week in YPD . A culture sample of each strain was then submitted to a fractionation step with a discontinuous sucrose gradient ( 40 and 80% ( w/v ) ) by centrifugation at 8 , 000 x g for 1 h at 10°C . The gradient fractions were collected by aspiration , dialyzed to remove the sucrose and fixed with 3% formaldehyde . Each fraction was labeled with 25 μg/mL of a Con A-Alexa Fluor 594 conjugate . After permeabilization of samples with 0 . 1% Triton X100 for 10 min and their subsequent washing , the nuclei of each fraction were labeled with 20 mM propidium iodide in DMSO . After the fractions were washed , each was resuspended in 1% formaldehyde and analyzed by flow cytometry . Flow cytometry was performed in a MoFlo XDP apparatus ( Beckman Coulter ) , collecting 50 , 000 singlet events . Fluorescence of ConA-positive events was recovered from the compensated FL3 ( orange ) channel using unlabeled yeast cells . PI staining was recorded in the compensated FL4 ( red ) channel . Total population densities were gated and analyzed using FlowJo ( version 10 . 0 . 7 ) software . Yeast cells from cultures that were 4 days old were collected by centrifugation as described above . Briefly , the fungal pellets were suspended in distilled water with an equal volume of glass beads ( 0 . 45–0 . 50 mm in diameter ) and shaken five times in a Braun homogenizer ( Braun , Melsungen , Germany ) for 1 min with cooling for 1 min on ice between shakings . Breakage of cells was monitored by light microscopy to confirm that the cells were disrupted . The homogenates were washed from the glass beads with distilled water and centrifuged at 480 x g for 5 min at 4°C . The pellet that contained the cell wall was freeze-dried and weighed . Alkali-soluble and alkali-insoluble cell wall fractions were obtained as described previously [23 , 25] . Briefly , the freeze-dried material was resuspended in 1 M NaOH for 16 h , after which the suspension was centrifuged to separate the alkali-insoluble material ( fraction 1 ) from the supernatant . The supernatant was neutralized with HCl and centrifuged , and the pellet ( fraction 2 ) separated from the supernatant ( alkali- and acid-soluble , fraction 3 ) , which was further analyzed as previously [26–28] . To obtain rhamnomannan , we treated fraction 3 by employing Fehling´s reagent at 4°C as previously reported [23] . The insoluble copper complexes that were generated were centrifuged , washed three times with 3% KOH and twice with ethanol and collected . The resulting residue was suspended in distilled water , and cations were removed with a Dowex 50W-X4 ( H+ form ) for 1 h at room temperature; the supernatant was precipitated by the addition of 4 volumes of ethanol . The residue was collected by centrifugation at 8000 RPM for 10 min ( Fraction 4 , Rhamnomannan ) . The mother liquor of the copper complexes was neutralized with acetic acid and centrifuged . The supernatant was dialyzed for 72 h against distilled water and deionized with a mixture of Dowex 1 ( HCO3- form ) and Dowex 50W-X4 ( H+ form ) , the filtrate was concentrated , and polysaccharide was precipitated by the addition of 3 volumes of ethanol ( Fraction 5 ) . All obtained fractions were freeze-dried and weighed . To determine the sugar and total amino acid content , we analyzed all cell wall fractions as follows: for hexose content , 10 mg of each fraction sample was suspended in 1 mL of 1 M HCl , sealed in a 2 mL Wheaton 176776 ampoule and heated for 3 h at 100°C . Hydrolyzed samples were diluted 1/10 or 1/100 . Quantification of sugar was conducted employing the hexose content quantification method that uses anthrone in concentrated H2SO4 . To determine amino acid and amino sugar contents , we suspended 10 mg of each sample in 1 mL 6 M HCl , sealed it in a 2 mL Wheaton 176776 ampoule and heated it for 16 h at 100°C . We then used published methods as described in [26] and [27] employing alanine and glucosamine solutions as standards , respectively . For rhamnose quantification , 10 mg of fraction 3 was suspended in 1 mL of 1 M HCl , sealed in a 2 mL Wheaton 176776 ampoule , and heated for 3 h at 100°C . Hydrolyzed samples were diluted 1/10 or 1/100 . Quantification of methylpentoses were conducted according to [28] using 85 . 7% H2SO4 and 3% cysteine in the reaction mixtures and rhamnose to construct a standard curve . Samples were prepared as KBr pellets . IR spectra were recorded from 3500 to 500 cm-1 using a Nicolet iS10 IR spectrometer ( Thermo Fisher Scientific , Waltham , MA , USA ) coupled to OMNIC 8 . 0 software and following the instructions of the Infrared Spectroscopy Service , Center of Chemistry , Instituto Venezolano de Investigaciones Científicas , Caracas , Venezuela . Further structural data were acquired via 13C- and 1H-NMR . The polysaccharide fraction to be analyzed ( ca . 20 mg ) and standards were dissolved in 1 mL of D2O and centrifuged ( 1000 g , 10 min ) , and the 13C spectra were obtained at 75 MHz and 70°C with a collection time of 16 h using a Bruker 300 Ultrashield spectrometer according to the instructions of the Nuclear Magnetic Resonance Service , Center of Chemistry , Instituto Venezolano de Investigaciones Científicas , Caracas , Venezuela . Yeast cells were harvested , washed once with deionized water and disrupted in a Braun homogenizer for 5 min , alternating 1 min periods of shaking and 2 min of cooling on ice . The cell homogenate was centrifuged , and the pellet was saved and washed with 1 mM NaCl , 2% SDS and 0 . 3 M β-mercaptoethanol as previously described [29] . Cell wall preparations were freeze-dried before hydrolysis , and aliquots containing 5 mg were suspended in 2 M trifluoroacetic acid and incubated at 100°C for 12 h . The acid-hydrolyzed samples were analyzed by high-performance anion-exchange chromatography with pulsed amperometric detection ( HPAEC-PAD ) in a carbohydrate analyzer system ( Dionex ) equipped with an ED50 electrochemical detector with a gold electrode , a GS50 pump gradient , and a CarboPac PA10 analytical column ( 3 x 250 mm ) with a CarboPac PA10 guard column ( 3 x 50 mm ) . Samples were eluted with a gradient of 8–20 mM NaOH with a flow rate of 0 . 3 mL/min for 30 min . Human monocyte-derived macrophages ( hMacs ) were obtained according to previous reports [30 , 31] . The hMac interaction assays were performed with yeast cells of S . schenckii ( ATCC MYA 4820 ) and S . brasiliensis ( ATCC MYA 4823 ) grown for either 4 or 10 days , as described above . Briefly , 106 peripheral blood mononuclear cells were seeded in 2 mL of supplemented DMEM containing 10% autologous human serum onto circular glass coverslips in a 24-well culture plate . The adhered monocytes were cultivated for 7 days at 37°C in an atmosphere containing 5% CO2 . The hMacs interacted with the yeast cells of S . schenckii or S . brasiliensis at an effector: target ratio of 3:1 . After 1 h of interaction , the coverslips were gently washed with PBS , pH 7 . 4 , and stained with an Instant Prov kit ( Newprov ) . The stained coverslips were mounted on glass slides and observed under light microscopy . The number of yeast cells inside macrophages ( supplementary material ) were counted . At least 50 macrophages were counted in a minimum of ten high-power fields . The number of yeasts endocytosed per macrophage was then determined . Venous blood was collected from the cubital veins of healthy adult volunteers , and all volunteers were informed of the study before providing written consent . The number of the Certificate of Presentation for Ethical Consideration related to this study is 62785716 . 2 . 0000 . 5259 , approved by the Ethical Committee of Hospital Universitário Pedro Ernesto , Universidade do Estado do Rio de Janeiro , Rio de Janeiro , Brazil . The sugar analysis and the uptake of S . schenckii and S . brasiliensis by hMac data were analyzed with a two-tailed t-test with a significance level set at p < 0 . 05 . The program used for both statistical analyses was GraphPad Prism 5 .
The structural organization of the cell wall of S . schenckii and , for the first time , that of S . brasiliensis , was studied using HPF-TEM . The cell wall of S . schenckii was observed initially as a single layer decorated with thin fibrils in young growing cells ( 3–4 days in culture ) . However in older 7 to 10 day cultures the wall was revealed as a double-layered cell wall ( Fig 1A ) . The cell wall thicknesses of two S . schenckii isolates were approximately 100 nm ( Table 1 ) . Interestingly , this secondary cell wall layer that appeared in older cultures of S . schenckii ( stationary phase ) was observed to detach entirely from the underlying cell wall via a precise fracture in the middle of the two layers ( Figs 1A and 2 ) . The same phenomenon was observed in three human clinical isolates of S . schenckii from distinct geographical regions ( Figs 1A and 2 and S1 Fig ) . The second outer cell wall layer was shed as an intact laminate sheet ( Figs 1 and 2 ) , and intact sheets of detached cell wall were imaged in the culture fluid suspending the older cell cultures ( e . g . Fig 2B ) . This process was observed only after 7 days in liquid culture ( Fig 1A and S1 Fig ) and was not correlated with any significant change in the content of structural polysaccharides ( chitin and β-glucan ) , rhamnomannan and amino acids ( Tables 2 and 3 ) . To confirm that complete cell wall detachment occurred , S . schenckii yeast cells were isolated from cultures that were 10 days old and fractionated in a discontinuous sucrose gradient ( 40–80% sucrose ) ( Fig 3A ) Three fractions were collected and labeled with ConA-Alexa Fluor and propidium iodide ( PI ) to detect PRM and nucleic acids , respectively . For DNA detection , all fractions were permeabilized before PI staining , as detailed in Methods . Fractions from the gradient flow cytometry analysis ( Fig 3 ) revealed the presence of two cell populations ( cells P1 and cells P2 ) by their size and granularity ( complexity ) features and the presence of isolated cell walls ( gradient top ) These observations were corroborated by ConA-Alexa Fluor 594 staining ( Fig 3C ) and PI nuclei staining of the different fractions ( Fig 3D ) . Cell walls exhibited a distinct granularity and traces of PI staining compared to both P1 and P2 fractions suggesting that the observed pattern was due to cell wall compaction and contamination with nucleic acid debris after its detachment from the cell membrane . Two clear S . schenckii cell populations were identified and denoted as cells with a double cell wall ( Cells-DCW ) ; single cell wall ( CW ) fractions . S . brasiliensis had a thicker inner cell wall layer than S . schenckii ( Table 1 ) when grown under the same culture conditions , and cells had an impressive 400 nm outer fibrillar layer that could be clearly observed in yeasts at later stages of growth ( Fig 1B and Fig 4 ) . Differences in the inner cell wall layer appeared to be species-specific as indicated by the analysis of two strains of each species ( Table 1 ) . On average , S . brasiliensis cell walls were 20 to 40% thicker than S schenckii cell walls but this difference was dependent on the isolate ( Table 1 ) . Chitin , β-glucan , rhamnomannan and amino acid content and cell wall sugar composition did not change markedly in either S . brasiliensis or S . schenckii between 4 and 10 days in culture ( Tables 2 and 3 ) . However , the cell wall of S . brasiliensis has 30% more chitin content and 100% more rhamnose content than that of S . schenckii ( Tables 2 and 3 ) . HPF-TEM showed that the cell wall fibrils of S . brasiliensis sometimes formed bridges between yeast cells ( Fig 1B and Fig 4 ) . Confocal microscopy of 4 days yeast cells labeled with ConA and CFW showed a positive PRM and chitin for both Sporothrix species ( Fig 5 ) . The ConA fluorescence pattern observed for S . schenckii , in contrast with that of S . brasiliensis ( Figs 5A and 6F ) had had a more uniform labeling pattern on the yeast cell surface ( Fig 5C ) . Fc-dectin-labeling of exposed β-1 , 3 glucan was negative for both species . Sporothrix spp . is a highly pleomorphic fungus and some differences in the yeast cell shape and size were expected for each batch of yeast cells [13] . Confocal images of 7 days yeast cell suggested that these cell wall bridges of S . brasiliensis could potentially form a network of yeast cells that adhered to each another ( Fig 6 ) . Similar networks of cells were not formed by S . schenckii under the same experimental conditions . Additionally , S . brasiliensis yeast cells were ConA and CFW positive . A non-uniform punctate distribution of ConA red fluorescence was observed on the surface of the yeast cells of S . brasiliensis . Few yeast cells were positive for WGA , indicating no chitin exposure in the native cell wall ( Fig 6C and 6D ) . For the structural analysis of polysaccharides , the cell walls of S . schenckii and S . brasiliensis cells in the yeast phase were purified and fractioned by the alkali solubility method for fungal cell wall analyses , as described in the Methods section . Structures were characterized by infrared ( IR ) spectroscopy as well as by 1H-NMR and 13C-NMR . The 1H-NMR and 13C-NMR spectra were compared with spectra that were previously reported for S . schenckii [32–36] . IR spectra of the alkali insoluble cell wall fraction exhibited absorption signals characteristic of chitin and β-glucans ( Fig 7 ) , including a strong , wide band at approximately 3400 cm-1 and additional bands at approximately 2920 and 1412 cm-1 [37] . Absorption bands at approximately 1560 and 1640 cm-1 are evidence of the presence of chitin , while the presence of glucan with a β-configuration was evident because of an absorption band at approximately 890 cm-1 ( Fig 7A and 7B ) . Additionally , the presence of absorption peaks belonging to β- ( 1 , 3 ) - ( 1 , 6 ) -glucan ( at 1156 , 1076 and 1041 cm-1 ) are present in the IR spectra of both species [37] . α-Glucans , distinctly from other dimorphic fungi , were not found in the Sporothrix spp . cell wall [36] . The rhamnomannan fraction isolated from the alkali-soluble fraction , as described In Methods , was further analyzed by nuclear magnetic resonance . The general pattern of 13C-NMR spectra for both species displayed signals corresponding to a rhamnomannan structurally similar to that described for the yeast phase of S . schenckii , and whose main chain or backbone is composed of mannose units linked by α-1 , 6 glycosidic bonds with single units of rhamnose as side chains ( Table 4 and S2 Fig ) . 1H-NMR spectra for both species are shown in Fig 8A and 8B . We focused on the H1 region ( Fig 8C and 8D ) , which has been previously used to identify rhamnomannan from S . schenckii [33] . Analysis of the complete 1H-NMR spectra for both species ( Fig 8A and 8B ) indicated the presence of a methyl group because of three protons linked to the C of the methyl group that are represented at chemical shifts between 1 . 17 and 1 . 20 ppm; this methyl group belongs to rhamnose , as it is the only monosaccharide present in rhamnomannan that possesses this chemical structure . Additionally , signals are also found that correspond to a proton linked to carbon 5 ( 3 . 5–3 . 39 ppm ) near the methylene group from carbon 6 in the mannose ring ( Fig 8A and 8B ) . The presence of proton signals at 5 . 21–5 . 25 , 5 . 08–5 . 12 and 4 . 84–4 . 87 ppm are characteristic of S . schenckii rhamnomannan , as previously reported [33 , 35] . However , an extra signal ( 4 . 97 ppm ) was only present in S . brasiliensis rhamnomannan ( Fig 8D ) and was not related to any previous reported result . The supramolecular organization of glycogen was recently determined in mammals [38] . Liver glycogen is composed of beta-particles ( ~20 nm ) , which form larger clusters linked by hydrogen bonds called glycogen alpha-particles . These alpha-particles have a characteristic rosette structure evidenced by TEM [37–39] . The glycogen alpha-particles measure at least 50 nm and can reach 300 nm [39 , 40] . We report the presence of the characteristic rosette-like structures in dimorphic fungus Sporothrix spp . , indicating glycogen alpha-particles ( Fig 8 and S3 Fig ) . The glycogen alpha-particles in S . schenckii and S . brasiliensis yeast cells were consistently distributed around the plasma membrane and close to the cell wall ( Fig 1 and Fig 9 , S3 Fig ) . Most glycogen alpha-particles were localized at the budding poles of yeast cells ( Fig 9C ) . The glycogen alpha-particles disappeared in both species after 7 and/or 10 days in culture ( Fig 1A and 1B ) . Differences in the wall architecture and composition of pathogenic fungi impact host recognition by innate immune cells [22] . To correlate the differences observed on the cell walls of S . schenckii and S . brasiliensis with host recognition ( innate immune system ) , we determined the uptake of yeast cells by human monocyte-derived macrophages . Fig 10 and Fig 11 shows that phagocytosis of S . brasiliensis yeast cells was significantly greater than that of S . schenckii yeast cells . The impact of cell wall modifications was observed in culture over time ( Fig 1 ) , and hMac phagocytosis of yeast cells of both species was much less when cells were grown for 10 days ( Fig 10 and Fig 11 ) than when grown for 4 days . These results suggest that differences in the composition ( Tables 2 and 3 ) and/or architecture of the cell surface ( Fig 1 ) during yeast growth might lead to differences in recognition of the fungus by professional phagocytes . The cell wall is a dynamic structure that can vary according to environmental conditions , which leads us to conclude that the cell wall plasticity observed for Sporothrix spp . can also occur inside the host .
Cell wall glycoconjugates of pathogenic fungi are involved in virulence and pathogenicity and can modulate the innate immune response [21 , 22 , 40] . The present study shows that Sporothrix spp . , especially S . schenckii , can slough off cell wall layers , which have the potential to cause antigenemia or inflammation at a distance from the site at which the mother pathogen cell is located . The most common lymphocutaneous form of sporotrichosis is characterized by the migration of the pathogen by the lymphatic vessels , but sporotrichosis can also spread in the manner observed in disseminated cutaneous and extracutaneous forms [1] . Some structural cell wall glycoconjugates , chitin and β-1-3 and β-1-6-glucans , are found in pathogenic and non-pathogenic species and are involved in the innate immune response [41–42] . These glycoconjugates are known as PAMPs . The exposure of chitin and β-glucans on the fungus surface favors their binding to the corresponding PRRs exposed by the host cells , allowing the uptake of the microorganism and/or triggering the secretion of specific cytokines [41 , 42] . In addition to these common structural polysaccharides , other homo- and heteropolymers and/or mannoproteins are present on the surface of fungal pathogens and can be very specific for certain genera or species [22 , 43–45] . A typical example is the expression of α-glucans that was described to occur on the surface of dimorphic fungi . These polymers play an important biological role by protecting the fungal cell of the dimorphic fungi Paracoccidioides brasiliensis , Histoplasma capsulatum and Blastomyces dermatitidis from the host immune response by preventing the exposure of β-glucans that are recognized by the dectin-1 receptor [22] . Furthermore , α-glucans were also found in the pathogenic species of the Scedosporium/Pseudallescheria complex [45] . We show here that α-glucans are not present on the cell surfaces of the dimorphic fungi S . schenckii and S . brasiliensis . Furthermore , S . brasiliensis , which is a highly virulent emerging species in the Sporothrix pathogenic clade [13 , 14] and is related to cat-transmitted sporotrichosis [5 , 9] , had a higher chitin and rhamnose contents and , thicker cell wall than S . schenckii . Cell wall peptido-rhamnomannans were first described on the surface of S . schenckii and Ceratocystis species [24 , 43 , 44] . The PRMs of S . schenckii consist of a peptide core adorned with O- and N-linked chains [34 , 43 , 44] . PRMs were later found in other pathogenic species of the Scedosporium/Pseudallescheria complex that bear epitopes similar to those of S . schenckii [44–49] . The main epitope described in the N-linked chains of S . schenckii yeast cell PRMs was the α-L-Rhap 1→3 α-D-Manp side chain [43 , 47] and is also present in the PRM O-linked chains [34] . The most important epitopes described in all human pathogenic species presenting PRMs on their surface were subsequently related to the O-linked chains , which exhibit important biological functions [44 , 45 , 49] . The O-linked chains of S . schenckii PRMs were characterized as bearing the cell wall ConA-binding sites of α-Manp 1→2 α-Manp linked to Ser/Thr and presenting the antigenic epitopes of α-L-Rhap 1→3 α-D-Manp , α-L-Rhap 1→4 α-D-GlcAp and α-L-Rha1→4 [α-L-Rhap 1→2] α-D-GlcAp [34 , 44] . These three epitopes were present in the O-glycosidically linked tri- , tetra- and pentasaccharides , respectively [49] . The ConA reactivity reported for the PRM is absent in the rhamnonannan fraction isolated by alkali extraction ( N-linked chains ) [50] , In the present work , we used confocal fluorescence microscopy to demonstrate surface mannan evidenced by positive ConA cell surface reactivity of not only S . schenckii yeast cells but also S . brasiliensis . We also identified a rhamnomannan moiety on the S . brasiliensis yeast cell surface that has a repeating α-D-Manp 1→6 α-D-Manp 1→ backbone with α-L-Rhap 1→3 α-D-Manp side chains . Our 13C-NMR results suggest that the structure of this cell wall polysaccharide is conserved in the yeast phase of both S . schenckii and S . brasiliensis . Although we could not find major differences in the structure of rhamnomannans ( free polysaccharides or PRM N-linked chains ) at the biochemical level , an extra resonance signal was present in the H1 region ( at 4 . 97 ppm ) of the 1H-NMR spectra in S . brasiliensis samples but was absent in S . schenckii spectra . This signal was previously observed in the 1970s [33 , 35] in the so-called type II rhamnomannan , and this polymer was only found in a few isolates of S . schenckii [43] . The rhamnose content of S . brasiliensis is 100% higher than that of S . schenckii . The former species also has 30% more chitin and significantly more rhamnomannan than that latter ( Tables 2 and 3 ) . The higher content of these cell wall components correlates with the differences observed between these two species at the ultrastructural level . HPF-TEM data show that two S . brasiliensis strains ( MYA 4823 , a feline clinical isolate , and MYA 4824 , a human clinical isolate ) have a thicker cell wall than two S . schenckii strains ( MYA 4820 and MYA 4822 , human clinical isolates ) . S . schenckii and S . brasiliensis isolates exhibited low and high virulence profiles , respectively [13 , 14] . A fibrillar material was present on the outer layers of both species , a length of up to 400 nm . The fibrils were more clearly observed in the older cells of both species and , although not statistically significant , the rhamnomannan content was slightly higher in yeast cells kept in culture for 7 and 10 days ( Table 2 ) . The ConA reactivity observed by confocal microscopy corroborates this model , indicating that PRM is present in the outermost cell wall layer since the ConA epitopes have been described to occur in only PRMs [50] . Our data suggest that the longer fibrils were associated with significantly higher rhamnose and rhamnomannan content of S . brasiliensis yeast cells . The cell wall architecture in both species differed greatly , and the cell wall differences became clearer as yeast cells aged . The conditions associated with longer term culture ( starvation , cell density dependent changes in physiology , etc . ) induced the formation of a second cell wall layer in S . schenckii . This layer had been observed by electron microscopy previously but was attributed to a cell wall fracture [51] and , later , to an unknown cell wall-shedding mechanism in S . schenckii [52] . We show here for the first time that S . schenckii produces a second cell wall layer in culture , using cryo-fixation methods that rapidly immobilizes the cells and preserves their fine structure of cells with much greater fidelity than conventional fixation protocols . Furthermore , we clearly show that by an unknown process , this second layer can become detached from the outer cell surface and delivered into the extracellular milieu ( S1 Fig ) . In contrast , older S . brasiliensis yeast cells were shown by confocal microscopy to have an extensive outer cell wall fibrillar layer that was more electron-dense than those of S . schenckii . and that formed interconnecting bridges between yeast cells ( Fig 4 ) to form organized cell clusters ( Fig 6 ) . These differences in cell wall architecture in the parasitic phase of S . schenckii and S . brasiliensis can suggest explanations for differences observed in the uptake of each species by human macrophages . The uptake rate was significantly different between yeast cells cultivated for 4 and 10 days and correlated with the cell wall modifications resulting from longer times spent in culture ( Fig 1 ) . Our group demonstrated in a murine model of sporotrichosis that S . brasiliensis is more virulent than S . schenckii [13] . We also demonstrated that S . brasiliensis strains frequently exhibit drug resistance to itraconazol [11] and to miltefosine [12] . The drug resistance of microorganisms is currently related to , among other factors , their capacity to form biofilms [53 , 54] . Interestingly , we observed that S . brasiliensis fibrils interconnected yeast cells forming cell clusters . The capacity of this species to form a mature biofilm that could alter its drug resistance profile needs further investigation . Glycogen , an intracellular storage polysaccharide , is a complex branched polymer of glucose units that can assemble into a number of morphologies that include supramolecular clusters of beta-particles that form larger clusters of alpha-particles . These alpha-particles have a characteristic rosette-like structure [37–39] , and are related to a distinct physiological function in mammalian cells [39 , 55] [38] . Glycogen rosettes that have a similar appearance to mammalian glycogen alpha-particles have also been observed in plants [56 , 57] . In the present work , we used HFP-TEM to show the well-organized presence of glycogen alpha-particles around the plasma membrane of Sporothrix spp . and close to the cell wall . Our hypothesis , that is being investigated in an ongoing project , is that these particles serve as a source of glucose for cell wall enzymes . We found a great number of glycogen alpha-particles at the budding pole of yeast cells , where intense de novo synthesis of cell wall components occurs ( Fig 9 ) . A previous report showed that a polysaccharide-rich particulate fraction could be isolated from cytoplasmic extracts of C . albicans and that those polysaccharide particles were similar to those of rabbit liver glycogen [58] . The polysaccharide particles identified in C . albicans have the macromolecular structure that is characteristic of glycogen , and they appear as rosette-like structures in electron microscopy images [39 , 58] . The presence of rosette-like structures around the plasma membrane of C . albicans was also observed in HFP-TEM preserved preparations [Neil Gow , personal communication] . We propose a new cell wall model for S . schenckii and S . brasiliensis based on the data presented here ( Fig 12 ) . Our model includes the presence of PRMs on the outermost cell wall layer as observed in the confocal images with ConA . The ConA reaction is related to PRM sites , where ConA binds [34 , 44 , 50] . Additionally , the higher chitin content of S . brasiliensis is illustrated in addition to its longer outer fibrillar layer . Chitin is proposed to be localized in the inner layer of the cell wall , as WGA-negative labeling indicates that this structural polysaccharide is not normally exposed on the yeast cell surface of Sporothrix . Rosette-like structures characteristic of glycogen alpha-particles organized around the plasma membrane is also illustrated in both species .
|
Sporotrichosis was for many years attributed to a single etiological agent , Sporothrix schenckii . However , more recently , a new emerging pathogenic species , Sporothrix brasiliensis , has been shown to cause severe cases in humans and is associated with cat-transmitted sporotrichosis . In contrast , S . schenckii is related to a sapronosis and a benign human subcutaneous mycosis . The fungal cell wall is the first point of contact between the host and the pathogen and , therefore , plays an important role in pathogenesis . The data presented here demonstrate a novel cell wall structural organization for S . brasiliensis and S . schenckii . Both Sporothrix species exhibited a bilayered cell wall structure . S . brasiliensis has a thicker cell wall , which correlates with a 30% higher chitin and 100% higher rhamnose content . Variations in the cell wall architecture in culture over time are described for both species , but these changes did not correlate with significant changes in the protein or polysaccharide content . Chitin and β-glucans , but not α-glucan , were identified in the cell walls of S . brasiliensis and S . schenckii , and a conserved rhamnomannan structure was identified . This report is the first describing the cell wall architecture of Sporothrix species and sheds new light on the high virulence and immunopathology of S . brasiliensis .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"cell",
"walls",
"chitin",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"microbiology",
"fungal",
"structure",
"fungi",
"materials",
"science",
"sporothrix",
"schenckii",
"cellular",
"structures",
"and",
"organelles",
"fungal",
"pathogens",
"macromolecules",
"materials",
"by",
"structure",
"polymers",
"polymer",
"chemistry",
"mycology",
"medical",
"microbiology",
"microbial",
"pathogens",
"chemistry",
"yeast",
"biochemistry",
"eukaryota",
"cell",
"biology",
"glycogens",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"glycobiology",
"organisms"
] |
2018
|
Cell walls of the dimorphic fungal pathogens Sporothrix schenckii and Sporothrix brasiliensis exhibit bilaminate structures and sloughing of extensive and intact layers
|
Immunity to malaria is widely believed to wane in the absence of reinfection , but direct evidence for the presence or absence of durable immunological memory to malaria is limited . Here , we analysed malaria-specific CD4+ T cell responses of individuals living in an area of low malaria transmission in northern Thailand , who had had a documented clinical attack of P . falciparum and/or P . vivax in the past 6 years . CD4+ T cell effector memory ( CD45RO+ ) IFN-γ ( 24 hours ex vivo restimulation ) and cultured IL-10 ( 6 day secretion into culture supernatant ) responses to malaria schizont antigens were detected only in malaria-exposed subjects and were more prominent in subjects with long-lived antibodies or memory B cells specific to malaria antigens . The number of IFN-γ-producing effector memory T cells declined significantly over the 12 months of the study , and with time since last documented malaria infection , with an estimated half life of the response of 3 . 3 ( 95% CI 1 . 9–10 . 3 ) years . In sharp contrast , IL-10 responses were sustained for many years after last known malaria infection with no significant decline over at least 6 years . The observations have clear implications for understanding the immunoepidemiology of naturally acquired malaria infections and for malaria vaccine development .
It is well established that immunity to severe clinical symptoms of malaria is acquired rapidly , but immunity to malaria infection is slow to develop and incomplete [1] , [2] . Naturally acquired protective immunity against blood stage malaria involves both antibodies and CD4+ T cells ( reviewed in [2] ) . Antibodies provide protection by blocking invasion of merozoites into new red blood cells ( RBCs ) , blocking cytoadherence of infected RBCs ( iRBCs ) to endothelial cells , and enhancing phagocytic activity of monocytes and macrophages . CD4+ T cells play crucial roles by providing help to B cells for the production of antibodies and by producing immune mediators essential for regulating cellular immune effector mechanisms . Although the contribution of CD4+ T cells to blood-stage malaria immunity has been extensively studied , the development and maintenance of malaria-specific memory CD4+ T cells is not well understood . It has been proposed that antigenic diversity [3] , inhibition of maturation of dendritic cells [4] , [5] , and apoptotic deletion of malaria-specific T cells [6] , [7] impair the development of memory responses after malaria infection , in particular impeding the development and/or longevity of memory CD4+ T cells . However , studies in animal models of malaria infection indicate that memory CD4+ T cells do develop and are maintained normally after malaria infection [8] , [9] . Whether the results from these experimental infections are representative of responses in humans remains to be elucidated . Memory CD4+ T cells typically respond to lower doses of antigen , require less costimulation , and rapidly differentiate into cytokine-producing effector cells after encounter with specific antigen [10] . They are characterized by expression of surface markers such as CD62L ( L-selectin ) , CD45RO and lack of CCR7 [11] but it is becoming clear that the pool of CD4+ memory T cells against any particular pathogen is phenotypically and functionally heterogeneous [12] . Understanding the development and maintenance of memory CD4+ T cells is fundamental to vaccine development . However , the presence of substantial numbers of malaria-reactive memory T cells in malaria naïve individuals [13] , [14] , [15] makes it difficult to interpret and understand the longevity of malaria-specific memory T cells . In the present study , we have identified malaria-specific cellular immune parameters among malaria-exposed individuals living in an area of very low malaria endemicity in Northern Thailand and determined the duration of the memory CD4+ T cell response to P . falciparum under conditions of infrequent re-exposure/boosting of the immune response .
Fully informed , written consent was obtained from each participant prior to enrolment in the study . Ethical approvals were obtained from the research ethics committees of the Research Institute for Health Sciences at Chiang Mai University , of the Ministry of Public Health , Thailand and of the London School of Hygiene and Tropical Medicine , UK . Study subjects were recruited from among long-term adult residents of Muang Na , a village in the malaria endemic Chiang Dao region of northern Thailand , near the border with Myanmar , or were permanent adult residents of the city of Chiang Mai where malaria transmission does not occur [16] . Venous blood was collected in acid citrate dextrose on the day of recruitment and again 3 , 6 and 12 months after recruitment . Giemsa-stained blood films were examined for the presence of malaria parasites . As HIV infection may affect immunological parameters , all subjects were tested for HIV infection ( presence of anti-HIV antibodies by gel particle agglutination assay ) at the time of recruitment and at the end of the study; subjects received pre- and post-test counseling from trained HIV counselors . Data from HIV-infected subjects were excluded from the analysis . None of the subjects were infected with P . falciparum or P . vivax - as determined by blood film examination and PCR - at any visit . The laboratory 3D7 strain of P . falciparum was maintained in continuous culture under standard conditions , as described previously [17] . Cultures were periodically tested for mycoplasma contamination by polymerase chain reaction ( PCR ) ( Venor GeM , Minerva Biolabs ) and found to be mycoplasma free . Mature schizonts were obtained by gradient centrifugation over 60% Percoll ( Amersham Biosciences ) , adjusted to a concentration 1×108 schizont-infected red cells/ml and exposed to three freeze/thaw cycles to obtain P . falciparum schizont extract ( PfSE ) . A single batch of PfSE was used throughout the study . The initial batch was aliquoted and kept frozen at −80oC until required . Blood samples from each subject were checked for subpatent malaria parasitaemia by PCR . DNA was isolated using FlexiGene DNA extraction kits ( Qiagen ) according to the manufacturer's protocol and subjected to nested PCR for P . falciparum and P . vivax , as described previously [18] . PBMC were separated from citrated blood by gradient centrifugation over Ficoll-Hypaque ( Amersham Biosciences ) . Contaminating erythrocytes were removed by incubation with lysis buffer ( 0 . 15 M NH4Cl , 10 mM KHCO3 , 0 . 1 mM Na2EDTA ) at RT for 5 minutes . The cells were washed twice with RPMI , resuspended in 10% human AB serum/RPMI ( R10 ) culture medium , stained with trypan blue to identify viable cells , counted and adjusted to the required concentration . Freshly isolated PBMC were phenotyped by staining 5×105 PBMCs with fluorochrome-conjugated anti-human CD3 , anti-human CD4 and anti-human CD45RO ( all from Caltag ) for 30 min at 4°C . Cells were washed twice with PBS containing 1% foetal bovine serum ( FBS , Gibco ) and 0 . 05% sodium azide ( FACS buffer ) , fixed in 1% paraformaldehyde ( PFA ) in PBS , and analysed by flow cytometry ( FACSCalibur , Becton Dickinson ) using Cell Quest software . Cell integrity was ascertained from their FSC/SSC distribution . PBMC were washed twice with sterile PBS and adjusted to 1×107 cells per ml in 0 . 1% FBS in PBS . The cells were then stained with CFSE ( Molecular Probes ) at a final concentration of 10 µM at RT for 8 min . Labeled cells were washed three times with 5% FBS in PBS , adjusted to 1×106 cells/ml in R10 and stimulated with 10 µg/ml PPD , 5 µg/ml PHA , 5×105/ml schizont equivalents of PfSE or medium alone for 6 days at 37°C , 5% CO2 . Labeled cells were then stained with mAbs as described above . Two million PBMC were stimulated with 10 µg/ml PPD , 5 µg/ml PHA , 5×105/ml schizont equivalents of PfSE , or medium alone at 37°C , 5% CO2 for 20 hours . Brefeldin A ( Sigma ) at a final concentration of 10 µg/ml was added to cultures for the last 4 hrs of incubation . Each culture was divided into two aliquots , washed with FACS buffer and stained with anti-human CD3-Tri-colour , anti-human CD4-APC and anti-human CD45RO-FITC for 30 min at 4°C . The cells were fixed with 4% PFA , permeabilized with 0 . 2% saponin ( Fluka ) in FACS buffer for 10 min at RT and incubated with either anti-human IFN-γ-PE or anti-human IL-10-PE for 30 min at 4°C , washed and then fixed in 1% PFA . Appropriate isotype controls were used . The cells were analysed by flow cytometry , in which 30 , 000 lymphocytes were collected . Two million PBMC were stimulated with 10 µg/ml PPD , 5 µg/ml PHA , 5×105/ml schizont equivalents of PfSE , or medium alone at 37°C , 5% CO2 for 6 days . Culture supernatants were analysed for IL-10 by ELISA . Maxisorb immunoplates ( Nalge Nunc International ) were coated with 2 µg/ml of purified anti-human IL-10 ( BioLegend ) in coating buffer overnight at 4°C . Plates were washed , blocked with 1% BSA/PBS for 2 hr at RT and washed . One hundred microlitres of supernatants or serially diluted standard cytokines in sample diluent ( 0 . 05% Tween-20 in blocking buffer ) were added to each well , plates were incubated at RT for 2 h , washed and then incubated for 1 hr at RT with 100 µl of 1 µg/ml biotinylated anti-IL-10 ( BioLegend ) . Plates were then developed with Streptavidin-HRP ( R & D Systems ) and OPD/H2O2 . The colorimetric reaction was stopped with 2 N sulphuric acid and absorbance read at 492 nm on a Spectra MR plate reader ( Dynex Technology ) . Sample values were calculated by interpolation from a standard curve of recombinant IL-10 which was included on every plate . Standard curves were compared ( between and within batches ) and plates with substantial deviations were repeated . Samples were processed in date order with most plates containing samples from different groups and/or from different follow-up visits . Statistical analysis was performed as described previously [16] . Briefly , Mann Whitney U test was used to analyse differences in the T cell responses among groups ( GraphPad Prism Software ) . Decay rates of memory CD4 T cell responses were calculated using logarithmically transformed data . The effect of time since last malaria infection was analysed using a log-linear mixed-effects regression model incorporating random intercepts assuming a Gaussian distribution across individuals . This resulted in an estimate of the decay rate of memory T cell responses , assuming a single-exponential decay model . Half-lives were calculated from the estimated decay rate and the boundaries at 95% confidence interval obtained from the mixed-effects model . Where the decay rate is a positive value , the calculated half-life is reported as infinity . All analyses were undertaken using Stata ( version 10 , Statacorp LP ) .
Three groups of study subjects were recruited based on their place of residence and their prior malaria history . Of the 93 subjects originally recruited [16] , sufficient PBMC from 87 subjects were available for this part of the study . The characteristics of these 87 subjects are summarized in Table 1 . Subjects from Chiang Mai were designated “City Naïve” ( n = 17 ) . Subjects from Muang Na ( Chiang Dao ) were designated “Rural with no clinical malaria episode” ( Rural 1; n = 29 ) if they reported no prior episodes of malaria infection and/or if no record of malaria infection was found in the past 6 years . Muang Na residents who had one or more fully documented episodes of infection with P . falciparum , P . vivax or both parasite species were designated as “previously malaria infected” ( Rural 2; n = 41 ) . Among these 41 subjects , 21 ( 51 . 2% ) individuals were known to have had at least one episode of P . falciparum infection , 14 ( 31 . 1% ) were known to have had at least one episode of P . vivax infection and 6 ( 14 . 6% ) individuals were known to have been infected with both P . falciparum and P . vivax in the past 6 years . None of the subjects were positive for P . falciparum or P . vivax as determined by blood film examination and PCR at any study visit . The three groups did not differ significantly by age or sex . It has previously been shown that T cells from naive ( malaria non-exposed ) volunteers proliferate in response to Pf schizont extract ( PfSE ) and that these responses are likely due to recognition of cross-reactive antigens [13] , [15] , [19] . The initial objective of this study was thus to determine whether there were differences in the proliferative capacity and the phenotype of proliferating cells amongst city ( naïve ) and rural ( definitely and putatively exposed ) individuals . Proliferation was assessed by CFSE dilution in CFSE-labeled PBMCs cultured with PfSE and subsequent phenotyping by flow cytometry . Since Thai people are routinely vaccinated with BCG ( Mycobacterium bovis Bacille Calmette-Guérin ) , PPD ( purified protein derivative of Mycobacterium spp . ) was used as a positive control for a recall antigen response . PHA was used a control for cell viability . The flow cytometric gating strategy is shown in Figure 1 for a subject with known prior exposure to P . falciparum . CD3+CD4+ lymphocytes were gated ( Figure 1A ) and then analysed for CFSE levels ( Figure 1B ) . CFSE dilution was also analysed separately for CD45RO+ and CD45RO- cells ( Figure 1C ) . Significant cell division ( i . e . accumulation of CFSElow cells ) was seen among CD4+ cells cultured with PfSE , PPD and PHA . In all cases , the majority of dividing cells were CD45RO+ suggesting that the responding cells are of the effector memory phenotype and confirming previous observations [20] . Consistent with published data , T cells from ‘City Naïve’ subjects proliferated in response to PfSE and their proliferative capacity did not differ from that of T cells from ‘Rural 1’ and from ‘Rural 2’ donors ( Figure 1D; Supplementa1 Table S1 ) . Similarly , there was no difference in the percentages of dividing cells ( Figure 1E ) between subjects who had previously been infected with P . falciparum only or with P . vivax only . There were no consistent differences in PHA-induced or PPD-specific proliferative responses between City and Rural subjects although PHA responses were somewhat higher in Rural 1 subjects ( Rural 1 vs . City naïve , p = 0 . 03 ) and PPD responses were higher in Rural 2 subjects ( Rural 2 vs . City naïve , p = 0 . 004 ) ( Supplemental Table S1 ) . Since immunological memory depends on being able to mount a fast and effective response to infection , the number and/or function of effector memory cells is likely to be a more relevant indicator of an anamnestic response than simply the number or proportion of proliferating cells . We therefore examined the capacity of memory CD4+ T cells to produce the effector cytokine IFN-γ and the regulatory cytokine IL-10 in response to 24 hrs stimulation with PfSE or PPD . Representative flow cytometry plots showing intracellular IFN-γ and IL-10 among CD4+ CD45RO+ T cells are shown in Figure 2A . The number of IFN-γ+ or IL-10+ T cells in PfSE-stimulated cultures is shown as the fold increase compared with the number in unstimulated cultures . Very few CD45RO+ CD4+ T cells from City Naïve or Rural 1 subjects produced IFN-γ when stimulated with PfSE ( averaging 1 . 2 and 1 . 8 fold increase over background , respectively ) ( Figure 2B ) . By contrast , IFN-γ production in response to PfSE among memory CD4+ T cells of Rural 2 individuals ( averaging 2 . 2 fold increase over background ) was significantly higher than for City Naïve subjects ( p = 0 . 004; Mann Whitney U test ) . Data from subjects who had been infected with P . falciparum only and P . vivax only were pooled since there was no difference in IFN-γ production between the two groups ( data not shown ) , however exclusion of P . vivax only subjects did increase the level of significance of the difference between naïve and exposed groups ( p = 0 . 002 ) . Median levels of PfSE-induced IFN-γ production among CD45RO+ CD4+ T cells did not change over the 12 months of the study in the Rural 1 group ( Figure 2C ) whilst in the Rural 2 group , median levels of PfSE-induced IFN-γ declined slightly but significantly ( p = 0 . 037; paired t-test ) over a period of 12 months ( Figure 2D ) . Among the 41 rural subjects ( Rural 1 + Rural 2 ) who had PBMCs tested at the beginning and at the end of the study , 23 ( 56 . 1% ) responded to PfSE by making IFN-γ at the time of recruitment and all of these remained IFN-γ positive 12 months later . Six rural subjects ( 14 . 6% ) who did not make IFN-γ at the beginning of the study did however make IFN-γ in response to PfSE at the end of the study , 12 months later . There was no indication that these 6 individuals were infected with malaria during the 12 months of the study and thus boosting of their cellular immune response is unlikely , although this cannot be ruled out . Alternatively , it is possible that in these individuals the frequency of circulating PfSE-responsive Th1 cells fluctuates close to the level of detection of the assay leading to stochastic variation in whether they are detected or not at any particular point in time . PBMC from very few individuals produced IL-10 after short-term stimulation with PfSE and numbers of IL-10 positive cells were not significantly above control values ( median fold increase 1 . 0 , 0 . 9 and 1 . 2 for naïve , rural 1 and rural 2 individuals respectively ) . Thus , there were no significant differences in immediate IL-10 responses among the groups and no significant change in immediate IL-10 production in response to PfSE during the 12 months of study was observed ( data not shown ) . There were no differences between the groups in immediate IFN-γ responses to PPD and the magnitude of PPD-specific IFN-γ responses did not change over the 12 months of the study ( Supplemental Figure S1 ) . To determine the longevity of the immediate memory T cell response to PfSE , we analysed the frequency of IFN-γ-producing memory cells versus time since last known infection , in Rural 2 individuals . Mixed-effects regression models indicate a steady decline in the IFN-γ response to PfSE in the P . falciparum-experienced individuals ( Figure 2E ) . The best estimate for the half-life of PfSE-specific IFN-γ-producing CD4+ T cells in these individuals is 3 . 27 years ( 95% CI: 1 . 94–10 . 26 years ) ( Table 2 ) . Although the decay of IFN-γ-producing T cells was much less obvious among P . vivax-experienced individuals ( estimated half-life 12 . 6 years , although the 95% CI included infinity ) ( Figure 2F ) it did not differ significantly from the half-life in P . falciparum-experienced individuals ( p = 0 . 227 ) . We found little evidence of malaria-specific IL-10 responses by flow cytometry after 24 hrs co-culture with PfSE , PHA or PPD . For PfSE , the fold increase in IL-10+ CD4+ T cells between PfSE-stimulated and unstimulated cells averaged 1 . 08 with no difference between groups and with very low median fluorescence intensity values ( Figure 1A and data not shown ) . We therefore examined the accumulation of IL-10 in Day 6 cell culture supernatants as an indication of central memory responses [21] . Strikingly , although PfSE-stimulated PBMC cultures from Rural 1 and Rural 2 subjects contained similar concentrations of IL-10 , in both cases IL-10 concentrations were significantly higher than in cultures of PBMC from City Naïve subjects ( Figure 3A ) . Data from subjects who had been infected with P . falciparum only and P . vivax only were pooled since there were no differences in IL-10 production between the two groups and exclusion of P . vivax-exposed subjects did not affect our conclusions ( data not shown ) . Interestingly , median concentrations of IL-10 were stably maintained in both Rural 1 and Rural 2 subjects over the 12 months of the study ( Figure 3B , 3C ) and mixed-effects regression models showed that PfSE-induced IL-10 concentrations were stably maintained for up to 6 years after last documented malaria infection in both P . falciparum- ( Figure 3D ) and P . vivax- ( Figure 3E ) exposed Rural 2 subjects . The best estimate of the half life of these IL-10 responses is infinity ( Table 2 ) . Of note , there was no significant association between individuals making an IFN-γ response and those making an IL-10 response , i . e . individuals making IFN-γ were no more or less likely to make IL-10 than were individuals who did not make IFN-γ . There were no significant differences between City and Rural subjects in Day 6 IL-10 responses to PPD ( although responses were higher among Rural 2 than Rural 1 subjects ) and the magnitude of PPD-specific IL-10 responses did not change over the 12 months of the study ( Supplemental Figure S1 ) . We have previously described long-lived antibody and memory B cell responses to malarial antigens in this same group of individuals [16]; only rural residents had detectable humoral immune responses to malaria antigens . To determine whether humoral and cellular immune memory to malaria are linked , we compared cellular immune parameters between seropositive and seronegative rural individuals and between those who did or did not have detectable memory B cell responses to malaria antigens . Immediate ( 24 hr ) IFN-γ responses to PfSE were significantly higher in individuals with serum antibodies to PfSE ( Figure 4A ) or with serum antibodies to at least one recombinant P . falciparum antigen ( apical membrane antigen-1 , merozoite surface protein ( MSP ) -1 , MSP-2 and circumsporozoite protein ) ( Figure 4B ) than in individuals who were seronegative for all of these antigens; individuals responding to three or more antigens had the highest IFN-γ responses . Furthermore , immediate ( 24 hr ) IFN-γ responses to PfSE were also significantly higher among individuals who had memory B cells against at least one P . falciparum antigen ( as determined by ELISPOT assay ) than among those who had no detectable memory B cells ( Figure 4C ) . Similarly , PfSE-specific 6 day IL-10 responses were also higher in seropositive than in seronegative individuals ( Figure 4D , 4E ) although IL-10 responses were not different between subjects with memory B cells and those without ( Figure 4F ) .
The induction and maintenance of immunological memory to malaria has been a topic of debate for many years ( reviewed in 2 , 22 , 23 ) but there are remarkably few studies that have attempted to examine memory responses over time . We have recently observed that B cell memory responses are stably maintained for at least 6 years in a rural Thai population living in an area where P . falciparum and P . vivax are endemic but where transmission is kept at extremely low levels [16] . In an extension of this study , we have now examined both the short-term ( 12 months of the study ) and long-term ( history of infection in the past 6 years ) stability of the CD4+ T cell memory responses to malaria antigens . We find that although immediate Th-1 effector memory responses ( 24 hr IFN-γ secretion from CD45RO+ CD4+ T cells ) decay with a half-life of approx . 3 years , central memory regulatory responses ( 6 day accumulation of IL-10 in culture supernatants ) are stably maintained for at least 6 years after last-documented malaria infection . In line with previous studies [13] , [15] , [19] , we observed that CD4+ T cells from both malaria-naïve and malaria-exposed individuals proliferated extensively when co-cultured with PfSE . Our observation that , even in malaria naïve individuals , the vast majority of these proliferating cells exhibited a memory ( CD45RO+ ) phenotype is consistent with previous studies indicating that these cells have been primed by exposure to commensal micro-organisms , pathogens and/or vaccine antigens carrying minimal T cell epitopes that cross-react with those of malaria proteins . Presumably the cells differentiate into memory cells as a result of microbial priming and subsequently proliferate when exposed to cross-reacting malaria antigens . Moreover , lymphoproliferative , IFN-γ and IL-10 responses to PfSE were so similar among cells from individuals known to have been previously infected with P . vivax and those known to have been infected with P . falciparum that we were able to pool the data for P . vivax and P . falciparum-infected subjects for all analyses . Pooling of these data was appropriate since there is extensive antigen cross-reactivity between crude P . falciparum and P . vivax extracts [24] and thus individuals who were seropositive for P . falciparum antigens may well have been exposed to P . vivax , and vice versa . Despite the lack of specificity of the T cell proliferative responses , malaria antigen-induced effector functions – immediate ( 24 h ) IFN-γ-producing effector memory cells and cultured ( day 6 ) IL-10 responses - were seen only in malaria-exposed subjects . These observations suggest that – at least among malaria-naïve donors – many of the cross-reactive proliferating cells are relatively undifferentiated and may best be described as Th0 rather than Th1 or regulatory cells . Conversely , among subjects with known malaria infection in the past 6 years , a substantial proportion of these cells fit the phenotype of effector memory Th1 cells , suggesting either that malaria infection is required to drive Th1 differentiation of cross-reactive memory cells or that malaria infection induces Th1 differentiation of additional T cell populations ( possibly of differing antigen specificity ) . Interestingly , we found no evidence of immediate effector cells producing IL-10; IL-10 secretion seemed to occur significantly later ( accumulating over 6 days ) suggesting that regulatory responses may reside within the central memory population and that their reactivation may be secondary to activation of Th1 cells . Whilst the cellular source of IL-10 in 6 day culture supernatants is not known , it was secreted in an antigen-specific manner . B cells secreting IL-10 in an antigen-specific manner have been described in mice but the antigen specificity of IL-10-secreting B cells is poorly documented in humans and they are found at much lower frequencies than IL-10 producing T cells [25] . Although NK cells and monocyte/macrophages are potential sources of IL-10 , their responses are not expected to be antigen specific or to differ between exposed and unexposed individuals . By analogy to studies in children recovering from acute malaria infections , one likely source of IL-10 is a population of CD4+ CD25− Foxp3− memory cells [26] but further studies are required to test this supposition . Both IFN-γ and IL-10 responses were more prevalent in subjects who had malaria-specific antibodies and/or memory B cells suggesting – unsurprisingly , perhaps - that the three arms of the adaptive immune response ( cellular , humoral and regulatory ) are induced in a coordinated fashion . The significant decline in the magnitude of the IFN-γ effector memory response with time since last known malaria infection was in stark contrast to the highly stable nature of the humoral immune response and of the IL-10 regulatory response and suggests that IFN-γ-producing cell lineages are less stable than IL-10-producing lineages . In support of our observations , secreted IL-10 responses to malaria peptides were detected in adults in a malaria epidemic-prone area of Kenya after several years of low transmission [27] , and were stably detected over an interval of 9 months in an area of high transmission [28] , but IFN-γ ELISPOT responses were much less stable in both settings . The half life of serum antibodies , memory B cells [16] and 6 day IL-10 responses did not differ significantly from infinity in the very same individuals in whom IFN-γ effector responses declined with a half-life of ∼3 years . These differing decay rates indicate that even though different aspects of the immune response may be induced in a coordinated fashion , they have different requirements for long term maintenance . On the other hand , given its well-described role as a B cell growth and differentiation factor [29] , [30] and its role in controlling production of malaria-specific IgG [31] , [32] , long term maintenance of memory B cells may well benefit from the sustained antigen-specific IL-10 responses . The relatively short half-life of Th1 effector cells compared to the long term maintenance of IL-10-secreting cells might be expected to lead to a shift , with increasing time since last malaria infection , of the balance of the anti-malarial immune response from a pro-inflammatory response towards a more anti-inflammatory response . This expectation is not entirely consistent with clinical observations in migrants , in whom re-exposure to malaria after many years typically results in low parasite densities ( implying retention of anti-parasitic effector mechanisms ) accompanied by quite significant clinical discomfort ( reviewed in 23 ) . On the other hand , long term retention of anti-malarial antibodies [16] may account for the ability to contain parasitaemia and the very low levels of severe malarial disease in these patients may well be due to long-lived regulatory responses [23] . Effector IFN-γ responses to malaria antigens appear to be rather short-lived , at least as compared to those induced by vaccination against viral diseases . Data on the longevity of immediate IFN-γ-producing CD4+ T cell responses are , surprisingly , rather few since most studies have tended to use IFN-γ ELISPOTS to enumerate these cells and thus the precise identity of the IFN-γ-producing cells ( CD4 T cell , CD8 T cell or even NK cell [33] ) is frequently unknown . However , Combadiere et al . [34] found that 20% of smallpox vaccinees retained circulating immediate IFN-γ-producing cells more than 13 years after vaccination with no obvious decline in the response between 13 and 25 yrs post-vaccination; although these responses were assessed by ELISPOT the cytokine-producing cells were shown to be CD4+ T cells in a follow-on study [35] . Similarly , Hanna-Wakim et al [36] detected mumps virus-specific immediate IFN-γ+ CD4+ T cells in all 10 individuals who had received mumps vaccination at least 10 years previously . Conversely , we observed that none of the individuals exposed to malaria more than 4 years ago had detectable malaria-specific immediate IFN-γ+ CD4+ T cells . It is not clear whether this reflects a true difference in duration of immune memory to viruses and malaria or whether it simply reflects differences between natural infection and vaccination . IFN-γ secreting effector T cells recognizing the P . falciparum thrombospondin-related adhesive protein ( TRAP ) have been reported to be short-lived , on the basis that IFN-γ ELISPOT responses to individual TRAP peptides were unstable from one year to the next [37] . However , it is possible that the precursor frequency of Th1 effector memory cells to individual malaria antigens hovers around the detection threshold for the assay and is thus subject to random fluctuation , as previously suggested [27] , [38] . The apparent acquisition of IFN-γ responses in some members of this study cohort over the 12 month period of the study is consistent with this interpretation and similar observations were made for IFN-γ responses to fragments of malarial merozoite proteins [38] . On the other hand , 56% of the rural subjects who made an immediate T cell IFN-γ response to PfSE at the first cross-sectional bleed also responded at the survey 12 months later suggesting that effector cell numbers remain substantially above the limit of detection in these individuals . Of interest however , the half-life of IFN-γ responses to malaria antigens appeared substantially ( although , with this sample size , not significantly ) longer in individuals previously infected with P . vivax than in those previously infected only with P . falciparum . Using a similar study design to the one we have used here , Zevering et al also observed a discernable decline in IFN-γ responses to the P . falciparum circumsporozoite protein within two years of the most recent malaria infection [39] and again , responses were much more stable over time among individuals previously infected with P . vivax [40] . Taken together these data suggest that persistence of dormant hypnozoite stages in the livers of P . vivax-infected subjects may provide a source of persisting antigen for efficient renewal of effector memory cells . Although there was a clear decline in the frequency of malaria antigen-specific IFN-γ secreting effector T cells with time since last known malaria infection , the half-life of these cells is still orders of magnitude greater than the lifespan of an individual effector memory cell . It has been estimated that effector memory and central memory T cells disappear from the peripheral circulation with half-lives of 6 days and 17 days respectively [41] . Thus , to persist for many months or years , memory T cells must either reside in tissues for extended periods of time or undergo relatively frequent self-renewal . Whilst we cannot formally exclude the possibility that effector cells are persisting – undetected - in tissues , we consider it unlikely . Only activated lymphocytes expressing adhesion molecules and chemokine receptors are able to migrate across endothelial barriers and gain access to non-lymphoid compartments and the endothelial barrier itself also needs to be activated ( e . g . by inflammation ) for this to occur . Whilst trafficking into the liver might occur in subjects with P . vivax hypnozoites , this would not be expected in subjects previously infected with P . falciparum . Cells remaining within the lymphoid compartment would be expected to recirculate and thus should appear , transiently but regularly , in the blood where they would be available for sampling . Conversely , intermitotic intervals of 15-50 days for memory T cells [41] are consistent with the self-renewal hypothesis and periodic exposure to persisting or cross-reacting antigens offers an obvious means by which this might happen . In summary , our data demonstrate the acquisition of specific , anti-malarial effector memory IFN-γ and central memory IL-10 responses in individuals undergoing very infrequent exposure to malaria infection . Although larger studies are required to more accurately estimate the longevity of IFN-γ and IL-10 responses , the apparently very different half-lives of these two responses raise interesting questions regarding the requirements for survival and self-renewal of these two populations of cells . Although it remains possible that persistent and repeated malaria infections in areas of very high endemicity may eventually lead to T cell anergy or clonal exhaustion , the fact that individuals in these areas eventually become resistant to high density malaria infections and clinical symptoms argues against this as a major impediment to development of effective immune responses . Finally , if the ∼3 year half life for effector memory responses to PfSE is typical for malaria antigens , and if responses of similar durability can be induced by vaccination , our results could be seen as encouraging for vaccine developers since they imply that – once induced – anti-malarial immune responses are likely to persist for long enough to confer a reasonable degree of protection even in the absence of frequent boosting .
|
Despite some recent successes in reducing the burden of malaria in several African countries , malaria still causes up to 500 million cases of acute illness every year , killing over a million people . The widespread availability of a safe and effective vaccine would greatly increase our chances of controlling this disease and possibly , even , eliminating it as a major health concern . Attempts to develop a vaccine have had limited success . The fact that people can be repeatedly infected with malaria over many years has raised the concern that immunity to malaria may be short-lived , complicating the induction of long term protection by vaccination . In this study we have calculated the half-life of cellular immune responses to malaria in previously infected individuals from Thailand . We have found that , in the absence of boosting of immunity by reinfection , malaria-specific inflammatory responses are relatively short-lived , with a half life of approximately 3 years . However , malaria-specific anti-inflammatory responses ( which have been linked to resistance to severe malarial disease ) seem to be very long-lived ( the half life being indistinguishable from infinity ) . Our observations have important implications for understanding the immunoepidemiology of naturally acquired malaria infections and for malaria vaccine development .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"infectious",
"diseases/protozoal",
"infections",
"immunology/immunity",
"to",
"infections"
] |
2011
|
Short-Lived IFN-γ Effector Responses, but Long-Lived IL-10 Memory Responses, to Malaria in an Area of Low Malaria Endemicity
|
The brain can learn and detect mixed input signals masked by various types of noise , and spike-timing-dependent plasticity ( STDP ) is the candidate synaptic level mechanism . Because sensory inputs typically have spike correlation , and local circuits have dense feedback connections , input spikes cause the propagation of spike correlation in lateral circuits; however , it is largely unknown how this secondary correlation generated by lateral circuits influences learning processes through STDP , or whether it is beneficial to achieve efficient spike-based learning from uncertain stimuli . To explore the answers to these questions , we construct models of feedforward networks with lateral inhibitory circuits and study how propagated correlation influences STDP learning , and what kind of learning algorithm such circuits achieve . We derive analytical conditions at which neurons detect minor signals with STDP , and show that depending on the origin of the noise , different correlation timescales are useful for learning . In particular , we show that non-precise spike correlation is beneficial for learning in the presence of cross-talk noise . We also show that by considering excitatory and inhibitory STDP at lateral connections , the circuit can acquire a lateral structure optimal for signal detection . In addition , we demonstrate that the model performs blind source separation in a manner similar to the sequential sampling approximation of the Bayesian independent component analysis algorithm . Our results provide a basic understanding of STDP learning in feedback circuits by integrating analyses from both dynamical systems and information theory .
Neurons receive inputs from a large number of other neurons encoding a variety of information about various signals . Despite the diversity and variability of input spike trains , neurons can learn and represent specific information during developmental processes and according to specific task requirements . Spike-timing-dependent plasticity ( STDP ) [1 , 2] is a candidate mechanism of neural learning . Extensive studies have revealed the type of information that a single neuron can learn through STDP [3–7]; however , the type of information that a population of neurons interacting with each other learns through STDP has not yet been determined . Understanding this extension from a single neuron to a population of neurons is crucial because a single neuron learns and represents only a limited amount of information that may be transmitted to it from thousands of inputs . Among neural interactions , lateral inhibition is a basic interaction widely observed in various regions , such as the olfactory bulb [8] , visual cortex [9] , somatosensory cortex [10] , and entorhinal cortex [11] . Previous theoretical results showed that neural circuits with lateral inhibition enhance signal detection [12 , 13] and improve learning performance [14–16] . Several simulation studies further revealed that neurons acquire receptive field [17–19] or spike patterns [20] through STDP by introducing lateral inhibition; yet , those studies were limited to simplified cases for which a large population of independent neurons was suggested to be sufficient [5 , 21 , 22] . Therefore , it remains unclear whether lateral inhibition plays a crucial role in STDP learning; in particular , the spike level effects of lateral inhibition remain elusive . Moreover , recent experimental results suggest that animals learn and discriminate mixed olfactory signals [23–25] or auditory signals masked by noise [26 , 27] , but it is still unknown how feedback interactions contribute to such learning . Here , by considering a simple feedback network model of spiking neurons , we investigated the algorithm inherent to STDP in neural circuits containing feedback . We analyzed the propagation of spike correlations through inhibitory circuits , and revealed how such secondary correlations influence STDP learning at both feedforward and feedback connections . We discovered that the timescale of spike correlation preferable for learning depends on whether the noise is independent from any signal ( random noise ) or generated from the mixing of signals ( cross-talk noise ) . We also found that excitatory and inhibitory STDP cooperatively shapes lateral circuit structure , making it suitable for signal detection . We further found a possible link between stochastic membrane dynamics and sampling process , which is necessary for neural approximation of learning algorithm of Bayesian independent component analysis ( ICA ) . We applied our findings by demonstrating that STDP implements a spike-based solution in neural circuits for the cocktail party problem [26 , 28 , 29] .
We constructed a network model with three feedforward layers as shown in Fig 1A ( see Neural dynamics in Methods for details ) . The external source layer represents the external environment or neural activity at sensory systems . The external layer also provides common inputs to the input layer and induces correlations in the neurons in the input layer . The input layer shows rate-modulated Poisson firing based on events at the external layer and external noise , which is approximated with the constant firing rate {rio} . Subsequently , spike activity at the input layer projects to the output layer , which also receives inhibitory feedback from the lateral layer . Neurons in the lateral layers are excited by inputs from the output layer . We assumed that all neurons in the input layer and the output layer are excitatory , whereas lateral-layer neurons are assumed to be inhibitory . Although excitatory lateral interactions also exist in the sensory cortex , they are typically sparse [30] and weak [10] compared with inhibitory interactions; thus we concentrated on the latter . For the analytical treatment , the neurons in the output and lateral layers were modeled with a linear Poisson model . We first studied synaptic plasticity at the feedforward connections ( connections from the input layer to the output layer ) , while fixing lateral connections ( i . e . , connections from the output layer to the lateral layer and connections from the lateral layer to the output layer ) . For STDP , we used pairwise log-STDP ( Fig 1B ) [31] , which replicates the experimentally observed long-tailed synaptic weight distribution [32 , 33] . We considered the case for information encoded in the correlated activity of input neurons [34 , 35] , and fixed the average firing rate of all input neurons at the constant value υoX ( See Table 1 and 2 for the list of variables and parameters ) . If the firing rate of input neuron i is given as rio+∑μ=1pqiμ∫0∞ϕ ( t′ ) sμ ( t−t′ ) dt′ , for external event sμ ( t ) and the response probability of the neuron qiμ , then common inputs from the external layer induce a temporal correlation proportional to h ( τ;θt ) ≡∫max ( τ , 0 ) ∞dt′ϕ ( t′ ) ϕ ( t′−τ ) . ( 1 ) where φ ( t ) is a response kernel ( see Eqs ( 14 ) and ( 24 ) in Methods for details ) . If we use ϕ ( t ) =t2e−t/θt/2θt3 , where θt is the parameter that controls the timescale of spike correlations , then h ( τ;θt ) =116θt3 ( τ2+3θt|τ|+3θt2 ) e−|τ|/θt ( gray line in Fig 1C ) . For the kernel function , we used the gamma distribution with shape parameter kg = 3 in order to reproduce broad spike correlations typically observed in cortical neurons [36 , 37] . Synaptic weight dynamics by STDP is written as dwjiXdt=xi ( t−djiXa ) ∫0∞Fd ( wjiX , s ) yj ( t−s−djiXd ) ds+yj ( t−djiXd ) ∫0∞Fp ( wjiX , s ) xi ( t−s−djiXa ) ds for Fd ( wijX , s ) =fd ( wijX ) e−s/τd , Fp ( wijX , s ) =fp ( wijX ) e−s/τp , where fd ( w ) and fp ( w ) are synaptic weight dependence of LTD/LTP ( long-term depression/potentiation ) , respectively . By taking the average of above equation over time and ensemble ( see Average synaptic weight velocity in Methods for details ) , the weight change of the feedforward connection WX can be approximated as WX•≈WX ( g1XE−g2XWZWY ) Ct , ( 2 ) where g1X and g2X are scalar coefficients , C is the correlation matrix , and E is the identity matrix ( see Eqs ( 25 ) – ( 30 ) for derivation ) . The first term describes the synaptic weight change directly caused by an input spike correlation and can be rewritten into the convolution of the temporal correlation and correlation kernel function χX1 as g1X≡G1X ( woX ) , G1X ( w ) ≡∫−∞∞χ1X ( τ;w ) h ( τ ) dτ , χ1X ( τ;w ) =∫−τ+2dXd∞dsF ( w , s ) εX ( τ+s−2dXd ) , ( 3 ) where F ( w , s ) = Fd ( w , -s ) if s<0 , else F ( w , s ) = Fp ( w , s ) , and εX is the EPSP curve of input neurons ( see Eqs ( 15 ) and ( 31 ) in the Methods ) . By the deconvolution of G1X ( w ) , we can separate the effect of the intrinsic network property χX1 and that of the input correlation h ( τ ) for STDP-based learning . Due to causality , LTP/LTD balance , and dendritic delay , χ1X ( τ;w ) typically becomes LTP-dominant around τ = 0 ( blue line in Fig 1C; we set w = woX ) , so that g1X takes positive values , which enables coincidence-based learning [4 , 5 , 38] . The second term of Eq ( 2 ) , which is of particular interest in this model , describes how the input correlation influences STDP learning at feedforward connections through lateral inhibition: g2X≡G2X ( woX ) , G2X ( w ) ≡∫−∞∞χ2X ( τ;w ) h ( τ ) dτ , χ2X ( τ;w ) =∫−τ+D∞dsF ( w , s ) ∫0τ+s−DdrεZ ( r ) ∫0τ+s−r−DdqεY ( q ) εX ( τ+s−r−q−D ) , ( 4 ) where D = 2dXd+dY+dZ , and εY and εZ are EPSP/IPSP curves of output/inhibitory neurons , respectively . This term primarily causes LTD as the sign flips through lateral inhibition ( −χ2X ( τ;w ) ; shown as the green line in Fig 1C ) . Previous simulation studies showed lateral inhibition has critical effects on excitatory STDP learning [17–19]; however , it has not yet been well studied how a secondary correlation generated through the lateral circuits influences STDP at feedforward connections , and it is still largely unknown how lateral inhibition functions with various stimuli in different neural circuits . For example , the correlation kernel of the feedback term exhibits a delay as the signal propagates through the inhibitory circuit; yet , we do not know how much delay is permitted for effective learning or if realistic synaptic delays satisfy such a condition . Furthermore , it is also unknown what information a circuit can learn if there are several mixed signals with different amplitudes for which symmetry-breaking learning [5 , 39] is not valid . Therefore , using theoretical analysis and simulation , we first investigated the properties of the inhibitory kernel −χ2X ( τ;w ) in STDP learning . In Eq ( 2 ) , if lateral inhibition is negligible ( i . e . , g2X/g1X = 0 ) , all output neurons acquire the principal component of the response probability matrix Q , and the other information is neglected [7 , 40 , 41] . On the other hand , if lateral inhibition is effective , different output neurons may acquire various components of the external structure . We first examined that point in a simple network model with two independent external sources ( Fig 2A ) . In the model , each external source drives an independent subgroup of input neurons ( we defined those input neurons as A-neurons and B-neurons ) , which project excitatory inputs to all of the output neurons . Here , we assume that source A drives input neurons with a higher probability than source B ( qA = 0 . 6 , qB = 0 . 5 ) , so that input neurons projected by source A show higher correlations ( cA = 0 . 36 ) than those receiving the output of source B ( cB = 0 . 25 ) . In the matrix form , Q= ( qA00qB00 ) , C= ( cA000cB0000 ) . The third row in Q represents response probabilities of background neurons in the input layer ( gray triangles in Fig 2A; note that C = QQt ) . We refer to this as the minor source detection task below . Here , for lateral connections , we assumed that both excitatory-to-inhibitory ( E-to-I ) and inhibitory-to-excitatory ( I-to-E ) connections are well organized such that inhibition only works mutually between two output neuron groups ( Fig 2A; blue lines are E-to-I and red lines are I-to-E connections . See also Eq ( 30 ) in Methods ) . The origin of these structured lateral connections is discussed later . When the network is excited by inputs from external sources , excitatory postsynaptic potential ( EPSP ) sizes of feedforward connections WX change according to STDP rules . Initially , in all output neurons , synaptic weights from A-neurons ( blue triangles in Fig 2A ) become larger because A-neurons are more strongly correlated with one another than B-neurons are . However , as learning proceeds , one of the output neuron groups becomes selective for the minor source B ( Fig 2B ) . After 30 min , the network successfully learns both sources . If we focus on the peristimulus time histogram ( PSTH ) for the average membrane potential of output neurons aligned to external events , both neuron groups initially show weak responses to both correlation events , and yet the depolarization is relatively higher for source A than for source B ( Fig 2C left ) . After 10 min of learning , both neuron groups show relatively stronger initial responses for source A , but group 1 shows a hyperpolarization soon after the initial response ( Fig 2C middle ) . As a result , synaptic weights from A-neurons to group 1 become weaker , and group 1 neurons eventually become selective for the minor source B ( Fig 2C right ) . The mean cross-correlation ( see cross-correlation in Methods for details ) between the external sources and the population activity of output neurons is maximized when the delay is approximately 10–15 ms ( Fig 2E ) . If we fix the delay at 14 ms , then the cross-correlation gradually increases as the network learns both sources ( Fig 2D ) . The same argument holds if mutual information is used for performance evaluation ( green lines in Fig 2D and 2E ) . Interestingly , the network better detects the minor source when it is learned with a highly correlated source compared with when it is learned with another minor source ( Fig 2F ) , because a highly correlated opponent source causes strong lateral inhibition on the output neurons , which enhances minor source learning . Similar results are also obtained for conductance-based leaky integrate-and-fire ( LIF ) neurons ( S1 Fig ) . To investigate how and when the network can acquire multiple sources represented by correlated inputs , we further analyzed the model above ( see Mean-field approximation of a two-source model in Methods for details ) . Because both output excitatory neurons and lateral inhibitory neurons are bundled into groups , in the mean-field approximation , we can approximate M excitatory populations and N inhibitory populations into two representative output neurons and two inhibitory neurons . Similarly , input neurons can be bundled into three groups ( A-neurons , B-neurons , and Background-neurons ) . In addition , we assumed that the synaptic connections from Background-neurons to output neurons are fixed because they showed little weight change in the simulation ( orange lines in Fig 2B ) . In this approximation , by inserting Eq ( 32 ) into Eq ( 29 ) , the mean synaptic weight changes of feedforward connections follow dwμνXdt≅∑ν′=1L/LaLawμν′XνoSG1X ( wμνX ) ∑ρqνρqν′ρ−NawZMawY∑ν′=1L/LaLawμ¯ν′XνoSG2X ( wμνX ) ∑ρqνρqν′ρ +F¯ ( wμνX ) [ ( νoX ) 2∑ν′=1L/LaLawμν′X− ( νoX ) 2NawZMawY∑ν′=1L/LaLawμ¯ν′X+ ( NawZ ) 2MawYνoXνμZ ] , ( 5 ) where μ = 1 , 2 and μ¯=2 , 1 ( μ≠μ¯ ) , and ν = A , B . The first two terms are correlation-based learning , and the last term is the homeostatic effect intrinsic to STDP [5] . G1X and G2X are coefficients determined by synaptic delays , EPSP/IPSP ( Inhibitory postsynaptic potential ) shapes , and correlation structure , as shown in Eqs ( 3 ) and ( 4 ) . By solving the self-consistency condition ( Eq ( 34 ) in Methods ) , the firing rates of inhibitory neurons are approximated as We estimated the nullclines by calculating the lines that satisfy w˙1μ ( w1A , w1B , w2A* ( w1A , w1B ) , w2B* ( w1A , w1B ) ) =0 for μ = A or B . As a result , we found that when the mutual inhibition is weak ( wI = 10 ) , the system has only one stable point at which w1A is larger than w1B ( Fig 3A left ) . At this point , w2A is also larger than w2B ( w2A = 9 . 64 , w2B = 3 . 60; not shown in the figure ) , which means that both output neuron groups are specialized for the major source A ( we call this state a winner-take-all state or T-state ) ; however , if the inhibition is moderately strong ( wI = 21 . 5 ) , two new stable fixed points and two unstable fixed points appear in the system ( Fig 3A middle ) . In the stable point on the left , neuron group 1 picks up source B while neuron group 2 picks up source A ( w2A = 12 . 52 , w2B = 2 . 87 ) . On the right-hand side , neuron group 1 selects source A while neuron group 2 selects source B ( we denote those two states as winners-share-all states or S-states below ) . At the stable point in the middle , both groups detect source A ( w1A = w2A = 9 . 47 , w1B = w2B = 3 . 61 ) . Note that because of the mutual inhibition , the synaptic weight from A-neuron is smaller when both groups learn A than it is when only group 1 learns A . For strong inhibition ( wI = 40 . 0 ) , the stable point in the middle disappears , and the system is stable only when two neuron groups detect different sources ( Fig 3A right ) . Simulation results confirm this analysis because strong inhibition indeed causes a winner-share-all state in which multiple neuron groups survive in competition [15] , whereas the network tends to show a winner-take-all learning when the inhibition is weak ( Fig 3B ) . We measured the degree of winner-share-all/winner-take-all states by defining the specialization index wSI as w′SI= ( w1A−w1B ) ( w2B−w2A ) , wSI=w′SI/|w′SI| . ( 7 ) If w’SI = 0 , we set wSI = 0 . If two output groups are specialized for different sources , wSI becomes positive , whereas if two groups are specialized for the same source , wSI becomes negative . When the synaptic delay in the lateral connections is small , only S-states are stable , whereas at longer delays , both S-states and T-states are stable . In the simulation , the network typically grows toward the latter state in the bistable strategy ( Fig 3C ) . Moreover , if we change the shape of the IPSP curve while keeping τZB = 5 τZA , for steep IPSP curves ( i . e . , both τZA and τZB are small ) , only the S-states are stable , whereas T-states also become stable for slower IPSPs ( Fig 3D ) . Therefore , both analytical and simulation studies indicate that lateral inhibition should be strong , fast and sharp to detect higher correlation structure . Moreover , lateral inhibition does not need to be pathologically strong because the I/E balance of NawZ/LwoX≅20% is sufficient to cause multistability . In the previous section , we revealed the effects of network properties for a fixed input correlation structure; however , actual neurons show various timescales for correlations depending on the brain region [37 , 42] and characteristics of the stimuli [43 , 44] , and it is largely unknown how different timescales influence correlation-driven learning . Therefore , we next considered the effect of correlation timescales , especially on noise tolerance . In our current model , input neurons respond to external sources with input kernel ϕ ( t ) =t2e−t/θt/2θt3 ( Fig 4A left ) , and so the correlation between input neuron i and l is given as Cil ( s ) =νoS∑μ=1pqiμqlμh ( s ) . By changing the parameter θt , we studied the effect of the correlation timescale on learning . The correlation is precise when θt is small , whereas it becomes broad at large values of θt ( Fig 4A right , Fig 4B ) . Because STDP causes homeostatic plasticity that does not depend on a correlation , as shown in the third term of Eq ( 5 ) , in a more precise approximation , Eq ( 2 ) should be written as WX•≈WX ( g1XE−g2XWZWY ) Ct+〈homeostatic term〉 . ( 8 ) We first calculated g1X and g2X at various θt . Both g1X and g2X become smaller for a larger θt , but decreases in g2X are slower than those in g1X , and , as a result , κ = g2X/g1X becomes larger for a longer correlation timescale ( Fig 4C ) . This means that a longer temporal correlation is more suitable for the detection of multi-components . This is indeed confirmed in the simulation ( Fig 4D ) . When θt = 0 . 5 and the minor component is slightly weaker than the major one ( cA = 0 . 36 , cB = 0 . 25 ) , the minor component is no longer detectable . On the other hand , at θt = 2 . 0 , the minor component is detectable even if the strength of the induced correlation is less than half ( cA = 0 . 36 , cB = 0 . 16 ) . At θt = 4 . 0 , g1X becomes smaller so that even the major signal is not fully detectable . Similar results hold for crosstalk noise . In the model above , the noise is provided through the spontaneous Poisson firing of input neurons as random noise ( Fig 4E top , black dots are spikes caused by random noise ) . In reality , however , there would be crosstalk noise among input spike trains caused by the interference of external sources . We implemented this crosstalk noise by introducing non-diagonal components in the response probability matrix as Q= ( qSqNqNqS00 ) , where qS is the response probability to the preferred signal and qN is that to the non-preferred signal ( Fig 4E bottom ) . We refer to this as the noisy source detection task below . To make a clear comparison , in the simulation of random noise , we kept qN = 0 and changed the spontaneous firing rate of the input neurons ( rio ) to modify the noise intensity , whereas in simulation of crosstalk noise we removed random noise ( i . e . , rio = 0 ) and changed qN . For random noise , a smaller θt enables better learning because a large g1X competes with the homeostatic force ( Fig 4F ) . By contrast , for crosstalk noise , the performance is better at θt = 2 . 0 than at θt = 0 . 5 because strong lateral inhibition suppresses crosstalk noise ( Fig 4G ) . Although for small noise regimens , the network performs better at θt = 0 . 5 than at θt = 2 . 0 , but the difference is almost negligible . Therefore , to cope with crosstalk noise , the spike correlation needs to be broad , whereas a narrow spike correlation is better for random noise . We note that qualitatively the same arguments as above also hold for the exponential kernel ϕe ( t ) =e−t/θt/θt ( S3D and S3E Fig ) . However , the ratio of two coefficients ( i . e . , κe = ge2X/ge1X ) is typically smaller for this kernel than for the kernel we used throughout this study ( S3B and S3C Fig vs . Fig 4D ) because lateral inhibition is less effective due to highly peaked spike correlation ( S3A Fig ) . To this point , we have considered a network already clustered into two assemblies that inhibit one another ( as in Fig 5A left ) . This means that the network somehow knows a priori that the number of external sources is two; however , in reality , a randomly connected network should also learn such information . To test this idea , we introduced STDP-type synaptic plasticity in lateral excitatory connections and feedback inhibitory connections and investigated how different STDP rules cause different structures in the circuit . We first checked whether structured lateral connections were helpful for learning . For comparison , we also considered a model with random lateral connections in which all output neurons and inhibitory neurons are randomly connected with probability 0 . 5 ( Fig 5A middle ) . When lateral connections are random , mean-field equations are modified as We separated lateral connections into two groups as in the previous case , but this approximation is legitimate only when two input sources are symmetrical ( i . e . , qA = qB ) . In other cases , neurons are often organized into two groups with different population sizes . In such cases , for evaluating performance , we measured average weights from source A on the output neurons receiving stronger inputs from A-neurons than from B-neurons or Background-neurons . For randomly connected lateral inhibition , learning performance dropped significantly in noisy source detection ( Fig 5B ) and in minor source detection ( Fig 5C ) ; thus clustered connectivity is indeed advantageous for learning . We next investigated whether such structure can be learned using STDP rules . We first introduced Hebbian STDP for both E-to-I and I-to-E connections . With these learning rules , the lateral connections successfully learn a mutual inhibition structure ( Fig 5D ) ; however , this learning is achievable only when the learning of a hidden external structure is possible from the random lateral connections ( magenta lines in Fig 5B and 5C; note that orange points are hidden by magenta points because they show similar behaviors in noisy cases ) , which means either when crosstalk noise is low or two sources have similar amplitudes . Nevertheless , once a structure is obtained in easy settings ( qN = 0 or qA = qB ) , that network outperforms the network with random lateral connections in both noisy source detection ( Fig 5E ) and minor source detection ( Fig 5F ) . In Fig 5E , we evaluated the performance of noisy source detection by first conducting STDP learning at qN = 0 , and then we terminated STDP and performed simulations at the various noise levels qN . Similarly , in the minor source detection task depicted in Fig 5F , we first performed STDP learning with qA = qB = 0 . 6 , and then evaluated the performance for a smaller qB . STDP can also generate similar lateral connection structures when the total number of input sources is larger than two ( S2A and S2B Fig ) . Therefore , STDP at lateral connections helps signal detection by efficiently organizing the connection structure . We next studied the analytical conditions for learning of the clustered structure ( see Analytic approach for STDP in lateral and inhibitory connections in Methods for details ) . The synaptic weight dynamics of lateral excitatory and inhibitory connections are approximately given as WY•≈g1YWYWXCtWXt , g1Y≡∫−∞∞dsFY ( s ) ∫DrX∫DuY∫Dr′Xh ( u+r′−s−r ) WZ•≈g1ZWXCWXtWYt , g1Z≡∫−∞∞dsFZ ( s ) ∫DrX∫DuY∫Dr′Xh ( r−s−u−r′−dZ−dY ) . ( 10 ) Both equations represent indirect effects of the input correlation propagated into the lateral circuit . From a linear analysis , we can expect that when gY1 is positive , E-to-I connections tend to be feature selective ( see Eq ( 35 ) in Methods ) . Each inhibitory neuron receives stronger inputs from one of the output neuron groups and , as a result , shows a higher firing rate for the corresponding external signal . On the other hand , if gZ1 is positive , I-to-E connections are organized in reciprocal form , where one of the reciprocal connections is enhanced and the other is suppressed ( see Eq ( 36 ) in Methods ) . We can evaluate feature selectivity of inhibitory neurons by φY=1N∑k=1N ( 1| ΩAY |∑j∈ΩAYwkjY−1| ΩBY |∑j∈ΩBYwkjY ) / ( 1M∑j=1MwkjY ) , ( 11 ) where ΩYA and ΩYB are the sets of excitatory neurons responding preferentially to sources A and B , respectively . Indeed , when the LTD time window is narrow , analytically calculated gY1 tends to take negative values ( the green line in Fig 6A ) , and E-to-I connections organized in the simulation are not feature selective ( the blue points in Fig 6A ) . By contrast , for a long LTD time window ( i . e . , when LTD is weakly spike-timing dependent ) , gY1 tends to take positive values , and E-to-I connections become clustered . In the simulation , WZ is also plastic , but as shown in Eq ( 10 ) , the effect of WZ on the plasticity of WY is negligible in first-order approximations . Similarly , for I-to-E connections , we measure the degree of mutual inhibition ( non-reciprocity ) with When LTD is strongly spike-timing dependent , gZ1 is negative and ϕZ calculated from the simulation data tends to be large ( Fig 6B ) , which means that inhibitory connections are organized such that the inhibition functions as mutual inhibition between excitatory neuron groups . Note that the organized neuronal wiring patterns are not a pure product of the pre-post causality of STDP but the effect of spike correlations propagating through lateral inhibitory circuits . If the structural plasticity is merely caused by the pre-post causality , both ϕY and ϕZ can decrease with increases in the inhibitory population while maintaining the total synaptic weights because the causal effect becomes weaker as each synaptic weight becomes smaller [45]; however , in our simulations , the values of both quantities generally increased for larger inhibitory populations ( S2C Fig ) . Hebbian inhibitory STDP at lateral connections is not always beneficial for learning . For example , in minor source detection , if we use Hebbian inhibitory STDP , a slightly minor source is not detectable , whereas for anti-Hebbian STDP , a small number of neurons still detect the minor source because reciprocal connections from strong-source responsive inhibitory neurons to strong-source responsive output neurons inhibit synaptic weight development for the stronger source ( Fig 6C ) . Our results to this point have revealed that correlation-based STDP learning combined with lateral inhibition can successfully detect signals from mixed inputs masked by noises . To confirm this mechanism is indeed effective in realistic tasks , we applied the above method to blind source separation . We first examined the condition in which the network could capture external sources . We extended the previous network to include four independent sources mixed at the input layer ( Fig 7A ) . In the present application , we used structured lateral connections because learning for clustered structures is difficult with noisy stimuli , as shown in the preceding section . The response probability matrix Q and correlation matrix C are given as Q= ( qSqN0qNqNqSqN00qNqSqNqN0qNqS ) , C= ( qS2+2qN22qSqN2qN22qSqN2qSqNqS2+2qN22qSqN2qN22qN22qSqNqS2+2qN22qSqN2qSqN2qN22qSqNqS2+2qN2 ) . Therefore , the principal components of matrix Q ( i . e . , eigenvectors of C ) are {1 , 1 , 1 , 1 , } , {-1 , 0 , 1 , 0} , {0 , -1 , 0 , 1} , {-1 , 1 , -1 , 1} . Because the first-order approximation of synaptic weight dynamics follows WX•≈g1XWXCt , we may expect that synaptic weight vectors converge to the eigenvectors of the principal components; however , this was not the case in our simulations , even if we took into account the non-negativity of synaptic weights ( see Fig 7B , where we renormalized the principal vectors to the region between 0 and 1 ) . Instead , each weight vector converged to a column of the response probability matrix Q ( Fig 7B , the left panel is the projection to the first two dimensions , and the right panel is the projection to the other two dimensions ) . This result implies that the network can extract independent sources , rather than principal components , from multiple intermixed inputs . We next evaluated the performance of hidden external source detection , especially its tolerance against crosstalk noise . To this end , we compared the performance of the model with that of the Bayesian ICA algorithm , in which independence of external sources is treated as a prior [46 , 47] . In the algorithm , the learned mixing matrix may converge to its Bayesian optimal value estimated from a stream of inputs . Although we cannot directly argue the optimality of cross-correlations , if the mixing matrix is accurately estimated , external activity is also well inferred , and thus we can use the mean cross-correlation as a measure for the optimality of learning . In terms of discretized input activity X , the external source activity S and prior information I , we can express the conditional probability of the estimated response probability matrix Q˜ as P[Q˜|X , I]=P[Q˜|I]P[X|I]∫P[X|S , Q˜ , I]P[S|I]dS ( see Bayesian ICA in Methods for details ) . This means that even if no prior information is given for Q˜ itself ( i . e . P[Q˜|I]=const . ) , posterior P[Q˜|X , I] still depends on a prior given for S . If we introduce a prior that each external source follows an independent Bernoulli Process ( i . e . P[S|I]=∏k=1T/Δt∏i=1L ( rsΔt ) sμk ( 1−rsΔt ) 1−sμk ) , then the stochastic gradient descendent of posterior function is given as , ∂∂q˜iμlogP[Q˜|X , I]=1Zp∑k=1T/Δt∫P[S , X|Q˜ , I]2xik−1xikpik/ ( 1−pik ) + ( 1−xik ) ∑k′=0∞ϕk′sμk−k′1−q˜iμ∑k′=0∞ϕk′sμk−k′dS , where We approximated this Bayesian ICA algorithm by a sequential sampling source activity instead of calculating the integral over all possible combinations in the estimation of the log-posterior of the response probability matrix Q . In this approximation , the learning rule of the estimated response probability matrix Q˜ obeys Δq˜iμk∝2xik−1xikpik ( Y1:k−1 ) / ( 1−pik ( Y1:k−1 ) ) + ( 1−xik ) ×∑k′=0∞ϕk′yμk−k′1−q˜iμk∑k′=0∞ϕk′yμk−k′pik ( Y1:k−1 ) =1− ( 1−rioΔt ) ∏μ=1p[1−q˜iμk∑k′=0∞ϕk′yμk−k′] , ( 13 ) where Y is the sampled sequence , and pik ( Y1:k-1 ) is the sample based approximation of pik in the previous equation . This rule has spike-timing and weight dependence similar to those seen in STDP ( Fig 7D ) . Although the performance of STDP is much worse than the ideal case ( when the true Q is given ) , this performance is similar to that for the sample-based learning algorithm discussed above ( Fig 7C ) . Therefore , the network detects independent sources if crosstalk noise is not large . We further studied the response of the models for the same inputs and found that the logarithm of the average membrane potential uμE=1|Ωμ|∑j∈ΩμujE well approximates the log-posterior estimated in Bayesian ICA , even in the absence of a stimulus ( Fig 7E ) . This result suggests that in the STDP model , expected external states are naturally sampled through membrane dynamics that are generated through the interplay of feedforward and feedback inputs . We finally performed the blind separation task using the same network as shown in Fig 7A . We created “sensory” inputs by mixing four artificially created auditory sequences ( Fig 8A and S1 Auditory File ) . In the auditory cortex , various frequency components of a sound , particularly high-frequency components , are represented by specific neurons typically organized in a tonotopic map structure [48] , whereas low-frequency components are expected to be perceived as a change in sound pressure . Furthermore , populations of neurons in the primary auditory cortex are known to synchronize the relative timing of their spikes during auditory stimuli and provide correlated spike inputs for higher cortical areas in which the auditory scene is fully analyzed and perceived [49 , 50] . We modeled these features by assuming that input neurons have a preferred frequency {fi} defined as fi=exp[iL ( logfmax−logfmin ) +logfmin] , and auditory inputs are provided as time-dependent response probabilities , which follow qi ( t ) =qo∑qalq ( t ) ahq ( fi ) , where aqh ( f ) is the spectrum of auditory source q ( left panel of Fig 8C ) , and aql ( t ) is the temporal change of the sound pressure ( black lines in Fig 8B ) . In this representation , each sound source is represented by correlated spikes of neural populations ( right panel of Fig 8C ) . Even if signals have overlapping frequency components {aqh ( f ) }q , blind separation is possible as long as {aql ( t ) }q are independent and have sharp rising profiles sufficient to cause spike correlations . After learning , four output neuron groups successfully detected changes in the sound pressure of the four original auditory signals ( colored lines in Fig 8B ) by correctly identifying the input neurons that encoded the signals . Therefore , STDP rules implemented in a feedforward neural network with lateral inhibition serve as a spike-based solution to the blind source separation or cocktail party effect problem .
Simultaneously recorded neurons in close proximity often show correlated spiking , yet the precision of these correlations varies across brain regions . Neurons in the lateral geniculate nucleus show strong spike correlations [42 , 51] , while correlations in V1 [36 , 52] or higher visual areas [37] are less precise . Our results indicate the interesting possibility that these differences may reflect the different characteristics of the noise with which the various cortical areas need to contend . At an early stage of sensory processing , the major noise component may be environmentally produced background noise from various sources; thus precise spike correlation is beneficial at this stage for noise reduction during signal detection and learning ( Fig 4G ) . By contrast , in higher sensory cortices , crosstalk noise accumulated through signal propagation in circuits may form the primary noise source , so less precise spike correlation is preferable ( Fig 4H ) . It would be intriguing to examine whether lower and higher cortical areas similarly change the strength of spike correlations for other sensory modalities . It is known that both glutaminergic synapses on inhibitory neurons [53 , 54] and GABAergic synapses on excitatory neurons [55 , 56] show STDP , and it is also known that STDP at E-to-I connections plays an important role in developmental plasticity [57]; however , detailed properties of these plasticities are still largely disputable [58 , 59] and , reportedly , highly dependent on inhibitory cell type [60] , neuromodulator [61] , and region [58] . We showed that in a feedback circuit , Hebbian inhibitory STDP preferred winner-take-all while anti-Hebbian inhibitory STDP tended to cause winner-share-all ( see Fukai and Tanaka 1997 for winner-share-all ) at excitatory neurons ( Fig 6D ) . This result indicates that different inhibitory STDP imposes different functions for excitatory STDP , which suggests that a neural circuit may select optimal inhibitory STDP for a specific purpose or strategy of learning , and this may differ across regions and be modified by neuromodulators . A recent study showed that inhibitory plasticity even directly influences the plasticity at excitatory synapses of the postsynaptic neuron [62] . In such cases , algorithm selection would play a more important role than it did for the standard STDP implemented in our model . Recently , inhibitory neurons in the rodent hippocampus CA1 were shown to display context-dependent activity rate changes during a spatial learning task , in association with the activity rate changes in excitatory cells [63] . In addition , the authors suggested the candidate mechanism for this change in activity is STDP at E-to-I synapses . Our results examining E-to-I STDP confirmed this configuration of inhibitory cells modulated by plasticity at feedforward excitatory connections ( Fig 5D , S2A and S2B Fig ) . In our model , although inhibitory neurons are not directly projected from input sources , as excitatory neurons learn a specific input source ( Fig 5D , left panel ) , inhibitory neurons acquire feature selectivity through Hebbian STDP at synaptic connections from those excitatory neurons ( Fig 5D , middle panel ) . Furthermore , our results indicate an important function of these feature-selective inhibitory neurons . Once an adequate circuit structure is learned and inhibitory connections are organized into a feature-selective pattern , even if the input to the network becomes noisy or faint , the network can still robustly detect signals ( Fig 5E and 5F ) . This robustness would be useful for spatial learning , as contextual information is often uncertain . Our results indicated that STDP in a lateral inhibition circuit mimicked Bayesian ICA [46 , 47] . First , output neurons were able to detect hidden external sources , without capturing principal components ( Fig 7B ) . Previous results suggest that for a single output neuron , an additional homeostatic competition mechanism is necessary to detect an independent component [7 , 22] . In addition , when information is coded by firing rate , homeostatic plasticity is critically important , because STDP itself does not mimic Bienenstock-Cooper-Munro learning [18] . However in our model , information was encoded by correlation , and mutual inhibition naturally induced intercellular competition so that intracellular competition through homeostatic plasticity was unnecessary . Moreover , our analytical results suggested the reason that independent sources are detected . To perform a principal components analysis using neural units , the synaptic weight change needs to follow WX•=WXC−LT[WXCWXt]WX , where LT[] means lower triangle matrix [64 , 65] . This LT transformation protects principal components caused by the lateral modification from higher order components; however in our model , because all output neurons receive the same number of inhibitory inputs Eq ( 2 ) , all neurons are decorrelated with one another and develop into independent components . Recently , it was shown that STDP can perform Bayesian optimal learning [66 , 67] . In the model used by those authors , the synaptic weight matrix is treated as a hyper parameter and estimated by considering the maximum likelihood estimation of input spike trains . By contrast , in the Bayesian ICA framework , the mixing matrix ( corresponding to synaptic weight matrix ) is treated as a probabilistic variable . Using this framework , we needed to calculate an integral over all possible source activities in the past to derive stochastic gradient descendent; however , as shown in Fig 7C , the stochastic learning was well performed by employing an approximation with sequential sampling . Moreover , we naturally derived an adequate LTP time window from the response kernel of input neurons to external events ( Fig 7D ) . We also found that STDP self-organized a lateral circuit structure that performed better than a random global inhibition ( Fig 5E and 5F ) . Mathematically , to perform sampling from a probabilistic distribution , we first needed to calculate the occurrence probability of each state; however , in a neural model , membrane potentials of output neurons approximately represent the occurrence probability through membrane dynamics . In machine learning methods , integration over possible source activities is often approximated using Markov chain Monte Carlo ( MCMC ) sampling [68] . Interestingly , a recent study showed that a recurrent network performed MCMC sampling [69 , 70] , suggesting that our network may perform a more accurate sampling in the presence of recurrent excitatory connections . Previous theoretical results suggest that STDP can modulate synaptic weights in a way that optimizes information transmission between pre- and postsynaptic neurons [71 , 72] . In the Bayesian ICA framework , blind source separation can be formulized as an optimization problem , but , in this case , the problem itself is ill-defined because optimality does not guarantee the true solution . In addition , local minima are often unavoidable for online learning rules . Nevertheless , the problems faced by the brain are often ill-defined , and suboptimality is inevitable [73] . Because we performed both nonlinear dynamics-based and machine learning-based analyses , we can offer some insights regarding the origins of local minima in stochastic gradient descendent learning . In the initial state , synaptic weights are typically homogeneously distributed , and this state is often locally stable . As a result , the homogeneous stable point is more likely to be selected in learning ( Fig 2C and 2D ) than the non-homogenous , more desirable , points; however , introducing additional noise may change this situation . Indeed , in Fig 4B and Fig 7C , the performance of the model was improved by adding a small amount of noise to input activities , although the improvement was not significant; however , because a large amount of noise is harmful for computations and stable learning , the benefit of noise addition is highly limited , and the brain may recruit other mechanisms for near optimal learning . Humans and nonhuman animals can detect a specific auditory sequence from a mixed , noisy auditory stimulus , a phenomenon often called the cocktail party effect . The mechanism underlying the cocktail party effect remains elusive [26 , 28 , 29] , although several solutions have been proposed [74 , 75] . An effective solution for this problem is ICA [76–78] , and the neural implementation of the algorithm has been studied by several authors [14 , 18 , 79 , 80] . Our study extended these results through a rigorous analytical treatment on biologically plausible STDP learning of spiking neurons , and our analyses enabled us to discover interesting functions of correlation coding . Moreover , by explicitly modeling inhibitory neurons , we found that STDP at E-to-I and I-to-E connections cooperatively organized a lateral structure suitable for blind source separation . In addition , we successfully extended a previous model for the formation of static visual receptive fields [18 , 19] to a more dynamic model in an auditory blind source separation task . In realistic auditory scene analysis , the frequency spectrum of acoustic signals is first analyzed in the cochlea , where each frequency component is the mixture of sound components from independent sources . Components belonging to the same source may be separated and integrated by downstream auditory neurons for the perception of the original signal . These frequency components can be considered a mixed signal in the ICA problem [81]; thus even if signals are mixed in frequency space , if the amplitudes of the signals are temporally independent , blind separation is still achievable . In the neural implementation of the problem , if two frequencies are commonly activated in the same signal , neurons representing those frequencies show spike correlation under the presence of the signal; thus the learning process is naturally achieved by STDP learning . These results indicate an active role of spike correlation and STDP in efficient biological learning .
|
In natural environments , although sensory inputs are often highly mixed with one another and obscured by noise , animals can detect and learn discrete signals from this mixture . For example , humans easily detect the mention of their names from across a noisy room , a phenomenon known as the cocktail party effect . Spike-timing-dependent plasticity ( STDP ) is a learning mechanism ubiquitously observed in the brain across various species and is considered to be the neural basis of such learning; however , it is still unclear how STDP enables efficient learning from uncertain stimuli and whether spike-based learning offers benefits beyond those provided by standard machine learning methods for signal decomposition . To begin to answer these questions , we conducted analytical and simulation studies examining the propagation of spike correlation in feedback neural circuits . We show that non-precise spike correlation is useful for handling noise during the learning process . Our results also suggest that neural circuits make use of stochastic membrane dynamics to approximate computationally complex Bayesian learning algorithms , progressing our understanding of the principles of stochastic computation by the brain .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion"
] |
[] |
2015
|
Mixed Signal Learning by Spike Correlation Propagation in Feedback Inhibitory Circuits
|
Pythium guiyangense , an oomycete from a genus of mostly plant pathogens , is an effective biological control agent that has wide potential to manage diverse mosquitoes . However , its mosquito-killing mechanisms are almost unknown . In this study , we observed that P . guiyangense could utilize cuticle penetration and ingestion of mycelia into the digestive system to infect mosquito larvae . To explore pathogenic mechanisms , a high-quality genome sequence with 239 contigs and an N50 contig length of 1 , 009 kb was generated . The genome assembly is approximately 110 Mb , which is almost twice the size of other sequenced Pythium genomes . Further genome analysis suggests that P . guiyangense may arise from a hybridization of two related but distinct parental species . Phylogenetic analysis demonstrated that P . guiyangense likely evolved from common ancestors shared with plant pathogens . Comparative genome analysis coupled with transcriptome sequencing data suggested that P . guiyangense may employ multiple virulence mechanisms to infect mosquitoes , including secreted proteases and kazal-type protease inhibitors . It also shares intracellular Crinkler ( CRN ) effectors used by plant pathogenic oomycetes to facilitate the colonization of plant hosts . Our experimental evidence demonstrates that CRN effectors of P . guiyangense can be toxic to insect cells . The infection mechanisms and putative virulence effectors of P . guiyangense uncovered by this study provide the basis to develop improved mosquito control strategies . These data also provide useful knowledge on host adaptation and evolution of the entomopathogenic lifestyle within the oomycete lineage . A deeper understanding of the biology of P . guiyangense effectors might also be useful for management of other important agricultural pests .
Mosquitoes are a major threat to global health since they are vectors of numerous devastating diseases , including malaria , dengue fever , Zika virus and other arboviruses , which together result in hundreds of millions of cases and several million deaths annually [1] . Existing commonly used control methods for reducing disease rely on the application of residual synthetic pesticides . However , intensive and repeated use of pesticides leads to ongoing development of resistance , environmental pollution and toxicity to human and non-target organisms [2] . Control strategies utilizing naturally occurring microbial pathogens have therefore emerged as a promising alternative . A particular focus has been on biological control agents [2] . Among them , the oomycete Lagenidium giganteum and the fungal pathogens , Beauveria bassiana and Metarhizum anisopliae , are well characterized , promising agents for mosquito larvae control , and have been produced commercially for field tests [2–4] . However , so far , available agents for mosquito control are rather limited . Recently , a new mosquito-pathogenic oomycete , Pythium guiyangense X . Q . Su was isolated from infected larvae of Aedes albopictus from Guizhou , China [5] . It is a virulent pathogen of a wide range of mosquito larvae and is safe to non-target organisms [6 , 7] . The pathogen also shows robust adaptability to a variety of natural environments and can be easily mass-produced [7] . All these properties of P . guiyangense make it of interest for practical applications as a potential mosquito control agent . However , little is known about the molecular mechanisms underlying the pathological processes on its mosquito hosts . It belongs to the genus Pythium ( kingdom Stramenopila; phylum Oomycota , class Peronosporomycetes ) [8] . Within the oomycetes , the genus Pythium is a genetically diverse group with a broad host range . For example , many Pythium species are important plant pathogens causing a variety of diseases [9 , 10] . On the other hand , P . undulatum can infect fish and P . insidiosum is a well-known pathogen that is capable of infecting human and other mammals [11] . To date , only P . guiyangense has been proposed for mosquito control [5] . The availability of genome sequences of a variety of pathogenic oomycetes , including Pythium species , provides a unique opportunity for comparative analysis between P . guiyangense and other oomycetes with respect to the evolution of pathogenicity . A number of genome sequences of plant pathogenic oomycetes are now available [12–14] . In addition , the genomes of mycoparasitic Pythium species that infect fungi , the human pathogen P . insidiosum , and the fish pathogens S . parasitica and S . diclina have also been sequenced [15 , 16] . Oomycetes secrete an arsenal of effectors into the host to manipulate the host immune system and enable parasitic infection [17] . These effectors have been a central question in the study of plant-oomycete interactions , including extracellular proteins such as toxins and hydrolases , and cell-entering proteins such as the RxLR ( Arg-X-Leu-Arg ) and Crinkler ( CRN ) effectors [18] . P . guiyangense may share conserved virulence effectors with other pathogenic oomycetes because of their close evolutionary relationships . Here , we determined the infection cycle of P . guiyangense and demonstrated an unusual process , in which mycelia were devoured by the mosquito larvae and the mycelia inside digestive system could effectively initialize infection . Then we produced a high-quality genome sequence assembly , which represents the first draft genome from an insect pathogenic oomycete . Based on the transcriptome , effector prediction , and comparative genome analyses with other Pythium species , we investigated its insect-killing mechanisms and finally identified several cytoplasmic effectors having virulence functions in insect cells , thus expanding the roles of oomycete effectors .
The P . guiyangense isolate was reported to be highly virulent on mosquito larvae , but infection was not quantitated [5 , 19] . We created a quantitative virulence assay by inoculating early second-instar larvae of Aedes albopictus or Culex pipiens pallens with P . guiyangense mycelia . The cumulative survival curves revealed that daily survival of A . albopictus and Cx . pipiens pallens larvae quickly declined from 3~4 days post-inoculation ( dpi ) onwards , reaching 76% mortality for A . albopictus and 69% for Cx . pipiens pallens by 10 dpi ( S1A Fig ) . A . albopictus larvae died faster than Cx . pipiens pallens larvae , reaching 50% survival by day 6 compared to day 8 ( S1A Fig ) . Furthermore , P . guiyangense could infect all the tested stages of Cx . pipiens pallens , including eggs , larvae , pupae and adults , resulting a visible accumulation of mycelia by 3–4 dpi ( S1 Video ) and all tissues were fully covered by mycelia ( S1B Fig ) . Together , the results confirmed that P . guiyangense is highly efficacious in killing all life stages of Aedes albopictus and Cx . pipiens pallens . To investigate the infection process of P . guiyangense , early second-instar larvae of Cx . pipiens pallens were incubated with mycelia or swimming zoospores . Our results showed that zoospores attached to almost any part of the mosquito larval cuticle ( Fig 1A showing zoospore attachments on the thorax and abdomen ) . Then , germination of the cysts occurred and appressorium-like swellings appeared at the tip of the germ tubes as visualized using scanning electron microscopy ( SEM ) ( Fig 1B ) . Penetration hyphae emerging from the appressorium traversed the insect integument ( Fig 1C ) and invaded the hemocoel of larvae . Eventually the mycelia filled the whole body , then emerged through the inner cuticle and formed sporangia ( Fig 1D ) . We also observed that Cx . pipiens pallens larvae readily ingested P . guiyangense mycelia even in an adequate food environment ( Fig 1E , S1C Fig , S2 Video ) . A thick section of a moribund larva visualized with SEM showed that , following feeding , the midgut was completely packed with mycelia that could initialize infection ( Fig 1F ) . After fixation , embedding in paraffin , and sectioning , microscopic observations showed that the midgut epithelium , muscles , and connective tissues appeared disrupted in the P . guiyangense-infected larvae ( Fig 1G and 1H ) . After 48 hpi , infected larvae appeared almost devoid of internal organs or tissues and the whole body was permeated with mycelia ( Fig 1I ) . Thus invasion through the digestive tract is an effective route of infection by P . guiyangense . Taken together , these observations define two routes of invasion , namely infection through the exterior cuticle and through the digestive tract . A high quality genome sequence of P . guiyangense was generated using a hybrid strategy that combined sequences from Pacific Biosciences long reads and Illumina short reads . The genome assembly indicated in an estimated P . guiyangense genome size of 110 Mb and annotation predicted 30 , 943 protein-coding genes ( Table 1 ) . The assembled genome consisted of 239 contigs with an N50 contig length of 1 , 009 kb . To assess the completeness of the genome assembly , CEGMA analysis , which identifies orthologs of 248 ultra-conserved core eukaryotic genes ( CEGs ) , was used to identify core genes in the P . guiyangense genome . The results revealed complete matches to 97 . 6% of CEGs and at least partial matches to 98 . 4% of CEGs within the P . guiyangense assembly; these results compared to only 78 . 2–94 . 4% of complete CEGs and 91 . 5–95 . 6% of partial CEGs in other Pythium genomes ( S1 Table ) . Taken together , the comparison with other assembled Pythium genomes including P . insidiosum , P . ultimum , P . aphanidermatum , P . arrhenomanes , P . irregulare and P . iwayamai , revealed that the P . guiyangense assembly represented the best quality among the sequenced Pythium genomes so far . Whole transcriptome sequencing ( RNA-seq ) was performed using Illumina sequencing of RNAs from P . guiyangense mycelia and from early second-instar larvae 24 hr after inoculation with P . guiyangense mycelia . Transcripts from each individual library matched approximately 69% of the genes , and together matched approximately 74% of the genes ( S2 Table ) . A total of 3 , 354 genes ( 10 . 8% ) were differentially expressed ( >4-fold expression difference and statistical GFOLD value > 1 or < -1 ) between the two samples; 1 , 654 genes were up-regulated in the infection stage while 1 , 700 genes were down-regulated . Functional enrichment analysis revealed that genes encoding tyrosine kinase-like ( TKL ) kinases , subtilisin proteases , kazal-type protease inhibitors , and elicitin proteins , were over-represented among the differentially expressed genes ( S3 Table ) . To validate the differentially expressed genes , we selected 18 genes that belonged to the above mentioned over-represented gene families and that were up- or down-regulated based on RNA-Seq data , and then measured their transcript levels by the qRT-PCR assay . The qRT-PCR results showed that the transcriptional patterns of 17 among the 18 genes were consistent with RNA-Seq results ( S2 Fig ) , which further supported the general reliability of the RNA-Seq data . Our comparative genome analysis revealed that the genome size ( 110 Mb ) and predicted gene number of P . guiyangense ( 30 , 943 genes ) were approximately twice those of the other sequenced Pythium species ( Table 1 ) . To explore the potential mechanisms underlying such a large genome size , we initially analyzed the repetitive DNA content within the P . guiyangense genome and found that the repeat sequence content was 6% , similar to that of P . ultimum ( 7% ) which has a genome size of only 43 Mb , thus excluding the possibility that high repeat content was responsible for the large genome size , which was observed in Ph . infestans [14] . Previously , hybridization between two parental species has been reported in yeast and Phytophthora evolution [20 , 21] . Most ( 84% ) of the Pythium core genes ( present in all the 7 Pythium genomes ) were present in two copies in the P . guiyangense genome , consistent with a hybrid origin , which is similar to the yeast [20] . Typically , the copied core genes shared the highest sequence similarity with genes derived from the P . irregulare genome , however , the two copied genes were about 8% different in nucleotide sequence . To further confirm the hypothesis of hybridization , internal synteny is selected as a useful indicator , which could be used to evaluate hybridization between two parental genomes [22 , 23] . Therefore , we characterized internal synteny across the P . guiyangense genome using the MCScanX program [24] . In the whole genome , 468 conserved synteny blocks with an average size of 192 kb were identified ( Fig 2A , S4 Table ) . These synteny blocks covered 74% of all the contigs , and together spanned 84% of the genome , suggesting that the P . guiyangense genome could be classified into two subgenomes . To further compare the genetic relatedness of the two parental subgenomes , the average nucleotide identity ( ANI ) was calculated , and the ANI between the two subgenomes revealed approximately 91% identity . Notably , a total of 11 , 068 pairs of homologous genes were identified in these synteny blocks ( Fig 2B , S4 Table ) , consistent with a hybrid origin . We estimated the rates of synonymous substitutions per synonymous site ( Ks ) of 11 , 068 pairs of homologous genes . This analysis showed a synonymous site divergence peak of Ks = 0 . 35 ( Fig 2C ) , indicating that the two subgenomes were relatively diverse . Taken together , we inferred that P . guiyangense was a hybrid genome derived from two distinct parental species . To further investigate the parental species of P . guiyangense , we systematically analyzed CEGs in the 7 sequenced Pythium genomes . The majority of CEGs were present as two copies in the P . guiyangense genome but only one copy in each of the other Pythium genomes ( Fig 2D and 2E ) . A total of 167 CEGs that contained 2 copies in P . guiyangense and also had orthologs in other Pythium species were utilized for phylogenetic analysis . For each phylogenetic tree , the two copies of the P . guiyangense CEG always clustered together most closely , and then clustered with the orthologs from the other Pythium species ( one tree based on the KOG1439 protein is shown in Fig 2F as an example ) , indicating that the parental species of P . guiyangense were not represented in the data set . In addition , cytochrome oxidase II ( cox II ) and β-tubulin genes , which have been widely used as phylogenetic maker genes , were available in 35 Pythium species and contained 2 copies in P . guiyangense . Phylogenetic analyses of the two genes showed that the two P . guiyangense orthologs were more similar to one another than the nearest known species ( P . orthopogon ) ( S3A and S3B Fig ) , indicating that the parental species of P . guiyangense were not represented based on current information . We speculate that the parental species of P . guiyangense are more closely related to each other than to the known Pythium species . In parallel with the genome analysis , we noticed that an unusual high percentage of P . guiyangense zoospores contained two nuclei rather than one ( S3C Fig ) . Among 500 observed zoospores , nearly 22% of them contained two nuclei in P . guiyangense while in P . aphanidermatum and Ph . capsici , all the spores had only one nucleus ( S3C Fig ) . We then found that the P . guiyangense zoospores containing only one nucleus could also breed similar percent of zoospores containing two nuclei , and PCR amplifications resulted in presence of both of the two copied genes . This observation suggested that P . guiyangense might be a dikaryon , and its relationship with the complex genome is still under investigation . To establish the phylogenetic relationship of P . guiyangense among oomycetes , a phylogenetic tree was constructed based on 248 CEGs from P . guiyangense and other 12 oomycetes , with the diatoms as outgroups ( Fig 3A ) . The tree clearly showed that P . guiyangense was clustered within the clade formed by the plant pathogenic Pythium species , and was distantly related to other genera , including Hyaloperonospora and Phytophthora . This phylogeny was consistent with that in previous publications [15 , 25] . These results imply that P . guiyangense , along with the mammalian pathogen P . insidiosum share common ancestors with the plant pathogenic Pythium species ( Fig 3A ) . To identify the genes responsible for host adaptation in P . guiyangense , the OrthoMCL tool was used to cluster the seven Pythium proteomes on the basis of protein sequence similarity . A total of 25 , 602 ( 83% ) P . guiyangense genes had orthologs in other Pythium species . Among these , P . guiyangense shared 13 , 000 core genes with the other Pythium species . In addition , 5 , 341 genes were identified to be specific to P . guiyangense ( S4 Fig ) . To gain insights into the features of species-specific genes in P . guiyangense , we compared the frequency of occurrence of protein family domains and identified highly over-represented domains included kinase ( PF00433 ) , kazal inhibitor ( PF00050 ) , elicitin ( PF00964 ) , and protease ( PF02902 ) ( Fig 3B , S5 Table ) . These gene families were also enriched among genes differentially expressed during infection . By searching with the HMM profiles of kinase domains derived from KinBase , 471 unique protein kinases ( 943 kinases in total ) were identified in the P . guiyangense genome , greatly surpassing the numbers in plant pathogenic Pythium genomes , which range from 152 to 192 ( Table 2 ) . Intriguingly , other two animal pathogenic oomycetes , P . insidiosum and S . parasitica , also have expanded kinomes , coding for 286 and 538 kinases , respectively [15 , 16] . We further classified the kinases into 9 families defined by Hanks and Hunter [26] . Five families , including TKL ( tyrosine kinase-like ) , CAMK ( calcium/calmodulin-dependent kinase ) , CMGC [including cyclin-dependent kinases ( CDKs ) , mitogen-activated protein kinases ( MAP kinases ) , glycogen synthase kinases ( GSK ) and CDK-like kinases] , AGC ( cAMP-dependent , cGMP-dependent and protein kinase C ) and "other" were noticeably expanded in P . guiyangense ( Fig 4A ) . Among the 5 expanded families , a total of 220 unique TKL genes were identified in P . guiyangense kinome . A comparison of the locations of TKL genes in the P . guiyangense and P . ultimum genomes revealed extensive rearrangements , which resulted from species-specific expansions at the locations of these genes ( Fig 4B ) . Forty-six unique kinases belonging to AGC family and 50 unique members of the CAMK family were identified from P . guiyangense . Based on the RNA-Seq data , a total of 92 kinase genes were differentially expressed at the infection stage , including 52 TKL kinase genes ( S6 Table ) . These results suggest that many of the protein kinases may be involved in regulation of infection processes and adaptation to the mosquito hosts . Since major structural and physiological differences were observed between plant cell walls and insect cuticles , we compared the repertoire of plant cell wall and cuticle degrading enzymes encoded in the P . guiyangense genome to other oomycete genomes . Several groups of plant cell wall degrading enzymes , such as GH53 , GH78 , CE5 , GH10 and GH11 , and GH12 were completely absent in P . guiyangense ( S7 Table ) . Genes encoding 12 unique pectin/pectate lyases ( PL1 , PL3 and PL4 ) , two unique GH28 and 1 unique GH43 involved in pectin backbone degradation were identified in the P . guiyangense genome; however , RNA-Seq data showed that none of these genes exhibited up-regulation during mosquito infection processes . P . guiyangense had more genes encoding proteases potentially involved in insect cuticle degradation than plant pathogenic Pythium species ( Fig 4C ) . A total of 307 unique proteases ( 615 genes in total ) were encoded in the P . guiyangense genome , compared to an average of 260 proteases in the plant pathogenic Pythium species ( Table 2 ) . The two animal pathogen genomes also had large numbers of proteases ( Table 2 , Fig 4C ) . Among them , genes encoding cysteine- , metallo- and serine-proteases were particularly highly expanded in P . guiyangense ( Fig 4C ) . The subtilisin serine-protease family had the highest relative expansion with 32 unique genes in P . guiyangense ( Table 2 , Fig 4C ) . Phylogenetic analysis revealed that over half of the subtilisin proteases were recently expanded in P . guiyangense due to lineage-specific gene duplications ( S5 Fig ) . Based on the RNA-Seq data , 31% of the total subtilisins were significantly up-regulated during mosquito infection . The peptidase_C1 and carboxypeptidases also exhibited significant expansion in P . guiyangense ( Table 2 , Fig 4C ) . Protease inhibitors regulate various biological and physiological processes in all living systems as modulators of protease activity [27] . Among them , the kazal-type protease inhibitor ( KPI ) family is one of the best characterized [27] . A total of 19 unique kazal inhibitors were identified in the P . guiyangense genome , which exceeded those in plant pathogenic Pythium species ( Table 2 ) . The animal pathogen , P . insidiosum also encoded larger numbers of kazal inhibitors . A phylogenetic tree was constructed using the Pythium kazal inhibitors , and the majority of genes derived from P . guiyangense and P . insidiosum formed clusters that were species-specific ( S6A Fig ) , implying that these genes were retained and diversified independently in these two animal pathogens . During mosquito infection , 25% of the P . guiyangense kazal inhibitors were up-regulated , and four of these exhibited transcript levels over 40 times those in the mycelia sample ( S6B Fig ) . The transcriptional patterns of the 4 kazal inhibitors at three infection time points were analyzed by qRT-PCR , and results revealed that all the 4 genes were up-regulated during the infection process ( S6C Fig ) . Further analysis demonstrated that all of the up-regulated kazal inhibitors contained signal peptides , indicating that they could play important roles in the pathogenesis . A common feature of many plant pathogenic oomycetes is the secretion of a variety of apoplastic ( extracellular ) proteins to promote infection , some of which can be detected by the host immune system . These include elicitins ( ELIs; lipid-binding proteins ) , elicitin-like ( ELL ) proteins and Nep1-like proteins ( NLP ) [28] . Ten unique ELI genes were identified in the P . guiyangense genome . In contrast , only 2 ELI genes were found in the P . irregulare genome , and none were identified in the other Pythium and S . parasitica genomes ( Fig 5A , Table 2 ) . Based on phylogenetic analysis of the elicitin domains , P . guiyangense ELIs were distributed into two clades , and one clade included genes from diverse species while the second clade only contained P . guiyangense ELIs ( Fig 5B ) . Moreover , 19 of the 20 ELI genes were physically clustered in the P . guiyangense genome , suggesting that ELIs were expanded in a species-specific manner . In contrast to the ELI genes , ELL genes were widely distributed in all the detected Pythium genomes . Both P . guiyangense and P . insidiosum had more ELL genes ( 45 and 50 unique genes ) than the plant pathogenic Pythium species ( 23–40 genes ) ( Fig 5A , Table 2 ) . Further phylogenetic analysis showed that over half of the ELL genes were distributed in nine clades which were specific to P . guiyangense and contained at least four members; thus many ELL genes were specifically expanded in P . guiyangense . Based on the P . guiyangense transcriptome analysis , 45% of the ELI genes were differentially expressed , and all were down-regulated during infection ( Fig 5C ) . In contrast , 31% of the ELL genes were up-regulated while 15% were down-regulated during infection ( Fig 5C ) . This observation suggested that the diverse ELIs and ELLs had a variety of different functions relative to growth and infection . Another common apoplastic effector family is the necrosis and ethylene-inducing-like proteins ( NLP ) genes . Many NLPs , but not all , can trigger cell death and defense responses in plants [29] . Only 1 unique NLP gene was found in P . guiyangense and none were found in P . insidiosum ( Table 2 ) . This NLP protein belonged to type 1 NLP subfamily with two conserved cysteine residues . Transcriptional analysis revealed no significant change during infection . These observations suggest that NLP proteins may not participate in oomycete-animal interactions . Crinkler ( CRN ) , a large class of cytoplasmic effectors , was first identified in Ph . infestans as a family of proteins that could cause plant cell death and defense responses [30] . A total of 38 CRN candidates were predicted in P . guiyangense ( S8 Table ) , compared to 10–46 predicted CRN proteins in the other Pythium species using the same method ( Table 2 ) . Examination of protein alignments of P . guiyangense CRN effectors revealed considerable conservation of the characteristic LxLFLAR/K and HVLVxxP motifs , which were similar to those observed in plant pathogenic Pythium species [13 , 25] . Based on five secretion signal predictors , 74% of CRN candidates in P . guiyangense contained a potential signal peptide or non-classical secretion signal ( S8 Table ) , suggesting that the majority of P . guiyangense CRN proteins might be secreted into mosquito hosts . A homology network of the oomycete CRN proteins was generated to investigate the evolutionary relationships between P . guiyangense and other oomycetes . The network is composed of 633 nodes in which each node represents an individual CRN protein . The network contains 34 , 149 edges that link nodes if the node proteins are homologous based on an all-versus-all BlastP search with an E-value cutoff of 10−10 . As shown in Fig 6A , the network was comprised of a crowd of disconnected clusters and a small number of singletons . P . guiyangense CRN proteins were mainly distributed in 3 large and 3 small clusters ( cluster I-VI represented by red dotted circles ) . Cluster I and III were composed primarily of P . guiyangense CRN proteins , with some of these proteins having homology to Pythium proteins . Notably , cluster II , IV and V only contained CRN proteins derived from P . guiyangense , revealing that these CRN proteins did not share significant sequence similarity with other oomycete CRN proteins . Moreover , all of the P . guiyangense CRN proteins showed sequence divergence of at least 50% with the most similar CRN protein in any plant pathogenic Pythium species , indicating that the CRN proteins are highly divergent between insect pathogenic Pythium species and plant pathogenic Pythium species . To explore the possible functions of P . guiyangense CRN proteins in insect cells , twenty-six CRN genes were expressed in Spodoptera frugiperda cell ( Sf9 ) lines; successful expression of the proteins was confirmed with western blots or by detecting fluorescence signals under the fluorescence microscope ( S7A and S7B Fig ) . The cell counting Kit-8 assay was used to determine protein toxicity to cells , with the Bacillus thuringiensis Delta-Endotoxin Cry1C as a positive control [31] . The results showed that 7 CRN proteins ( CRN31 , 33 , 34 , 36 , 37 , 38 , and 28 ) significantly decreased the viability of Sf9 cells while the remaining CRN proteins produced responses similar to the negative control ( Fig 6B ) . Notably , CRN31 appeared to be the most toxic to Sf9 cells . To further validate the toxicity of these 7 CRN proteins , a prokaryotic expression system was used to obtain recombinant CRN proteins ( S7C Fig ) . E . coli crude extracts containing the expressed proteins were then incubated with Sf9 cells and with mosquito Aedes albopictus C6/36 cells , respectively , to determine the toxicity using the cell counting Kit-8 assay . The results showed that CRN31 and CRN28 significantly reduced the viability of Sf9 cells and C6/36 cells ( Fig 6C and 6D ) . Since the transcript levels of the CRN31 and CRN28 genes were not elevated during infection at 24 hpi as measured by RNA-seq ( S8 Table ) , we used qRT-PCR to test whether the two CRN genes were significantly up-regulated during earlier infection stages ( 1–4 hpi ) ( S7D and S7E Fig ) . CRN31 exhibited the highest transcript level change with a 65 fold change at 2 hpi while CRN28 showed 4–5 fold changes at 1–3 hpi ( S7D and S7E Fig ) . Together these results suggested that CRN31 and possibly CRN28 might act as cell-killing effectors during insect infection .
In this study , we have determined the mode of infection of P . guiyangense on mosquito larvae . In our experiments , P . guiyangense caused up to 76% mortality for A . albopictus and 69% for Cx . pipiens pallens larvae . Analogous to most of the entomopathogenic fungi , P . guiyangense hyphae emerging from germinating zoospore cysts entered their host directly through the exterior cuticle , propagated inside hosts , and produced sporangia to start a new cycle of infection . Another infection route of P . guiyangense was through the ingestion of mycelia by larvae . Mycelia in the digestive tract progressively destroyed internal tissues of the larval midgut , leading to host death . The most common invasion route for aquatic insect pathogens , including Metarhizium anisopliae , Aspergillus clavatus and Beauveria Bassiana , was through ingestion of spores to infect their host [4 , 32 , 33] . P . guiyangense has evolved a similar strategy to initiate infection in the digestive system . Overall , P . guiyangense utilizes cuticle penetration and ingestion of mycelia into the digestive system to infect mosquito larvae . We speculate that firm adhesion of zoospores to the mosquito larvae epicuticle is critical for the success of the P . guiyangense pathogen which involves a combination of passive hydrophobic and electrostatic forces as well as protein interaction . Hydrophobins found in the outer layer of the spore cell wall of Beauveria Bassiana , mediate adhesion to the arthropod cuticle [34 , 35] . Hydrolytic enzymes , Mad1 and Mad2 , also assisted in attachment of the fungi to insects [36] . To identify the factors that promote attachment and ingestion of P . guiyangense by the mosquitoes would be interesting to further explore in the future . To probe the molecular basis underlying the interactions of P . guiyangense with insects , a high-quality genome assembly and transcriptome sequences were generated for P . guiyangense . Our results reveal that P . guiyangense is probably a hybrid genome derived from two parental species . Natural interspecies hybridization events have been described in the genus Phytophthora such as Ph . andina , Ph . nicotianae and Ph . cactorum [37 , 38] . It is believed that interspecies hybridization has the potential to create new strains that have a new or expanded host range [21] . Considering the distinct hosts , we speculate that the hybrid feature of P . guiyangense contributes to its adaptation of the mosquito host . The two parental subgenomes of P . guiyangense are approximately 9% different in nucleotide sequence , suggesting that the two parents are relatively diverse , however , the potential parents are still mysterious based on limited Pythium data . We will pay close attention to the new information of Pythium and update the concerns in future study . The phylogenetic analysis of the currently sequenced oomycete pathogens together with two diatoms demonstrated P . guiyangense is closely related to three plant pathogenic Pythium species ( P . irregulare , P . iwayamai and P . ultimum ) but has a slightly more distant relationship with the mammalian pathogen , P . insidiosum . This finding suggested that as a facultative mosquito pathogen , P . guiyangense , may have evolved from a common ancestor with the plant pathogens . This result is highly concordant with recent analysis indicating the mosquito oomycete pathogen , L . giganteum has also evolved from a plant pathogen [3] . In conjunction with the transcriptome analysis , oomycete genome comparisons identified several gene families that might contribute to P . guiyangense virulence . In this study , 471 putative unique kinases ( 943 kinases in total ) were identified in the P . guiyangense genome . Comparison with other sequenced oomycete genomes revealed that the genomes of the animal pathogens , P . guiyangense , P . insidiosum and S . parasitica , also encoded significantly more kinases than the plant pathogenic Pythium genomes [15 , 16] . Transcriptome analysis revealed that a total of 92 kinases were differentially expressed during infection of P . guiyangense against mosquito larvae , implying that protein kinases may be involved in regulating virulence . We also found that genes involved in insect cuticle degradation were expanded in P . guiyangense while proteins for plant cell wall penetration were absent or lost functions . The P . guiyangense genome encoded a significantly larger number of proteases than plant pathogenic Pythium species , including cysteine- , metallo- , and serine-proteases . Transcript levels of 31% of the total subtilisin-like serine proteases were significantly elevated when P . guiyangense invaded mosquito larvae . Some of these proteases were reported as key virulence determinants in entomopathogenic fungi [39] , supporting a potential role of these proteases in P . guiyangense infection . A large number of Kazal proteinase inhibitors ( KPIs ) were characterized from P . guiyangense and 25% of these genes were up-regulated during infection of mosquito larvae , suggesting KPIs may be involved in pathogenicity . Our study demonstrated that one invasion route of P . guiyangense was through ingestion of mycelia in the digestive system . The mosquito midgut contains an abundant array of secreted serine proteases for digestion , providing nutrition for development [40 , 41] . To aid in colonization in its hosts , P . guiyangense may secrete protease inhibitors , such as KPIs for protection from these proteolytic enzymes . This is consistent with previous studies which show that the animal parasite , Toxoplasma gondii secretes TgPI-1 and TgPI-2 , and Hookworm , Ancylostoma ceylanicum , secretes a 8-kDa broad spectrum serine protease inhibitor of the Kunitz family into the host digestive tract to aid in infection [42–44] . Serine proteases are also key components of immune responses and KPIs may manipulate host immune defenses for pathogenicity [45] . A kazal-like serine protease inhibitor was characterized from the plant pathogenic oomycete , Ph . infestans , and it targeted protease P69B to counteract tomato defense responses [46] . Another oomycete pathogen , Ph . palmivora also produced a KPI , PpEP to suppress plant defense [47] . Therefore , we speculate that the large number of KPIs secreted by P . guiyangense may suppress mosquito immune defenses by targeting serine proteases . In addition to hydrolytic enzymes , plant pathogenic oomycetes deliver a diverse battery of other secreted proteins into host tissue to support infection and interfere with host immune responses , including lipid-binding proteins ( elicitins ) , toxins ( e . g . NLPs ) , and host-cell-entering RxLR and CRN effectors [48 , 49] . We found distinct sequence and evolutionary features of these proteins in P . guiyangense . Firstly , no statistically significant evidence for RxLR effectors encoded in the P . guiyangense genome was found , in agreement with previous reports [13 , 25] . Secondly , 10 unique ELI genes were present in the P . guiyangense genome whereas these genes were largely absent from the other Pythium species , including P . insidiosum . The P . guiyangense ELI genes appear to have expanded relatively recently to form two species-specific clades , and one clade appears to share a common origin with the Ph . sojae genes . In contrast to ELIs , ELLs have been widely found in all sequenced Pythium species . RNA-Seq data showed that ELI and ELL genes show differential expression patterns in P . guiyangense . ELI genes were typically highly expressed in the mycelia stage while a large number of ELL genes were up-regulated in the infection stage . These results suggested that the two subclasses of elicitins may be involved in different functions . Elicitins have also been reported in another mosquito pathogenic oomycete , L . giganteum [3] , suggesting that elicitins in the two aquatic insect oomycetes may be linked to pathogenicity towards the insect hosts . CRN effectors are considered more ancient cytoplasmic effectors than RxLRs , as they are distributed across a wide range of oomycetes [14 , 25] and have also been reported in the fungal animal pathogen , Batrachochytrium dendrobatidis [50] and in arbuscular mycorrhizal fungi [51] . Interestingly , CRN proteins were also detected in the mosquito pathogenic oomycete L . giganteum [3 , 52] . They are presumed to enter the host cytoplasm and manipulate cell death and defense responses [30 , 53] . It has been widely reported that only a handful of oomycete CRN proteins were predicted to contain canonical signal peptides [13 , 14] . In this study , four different signal peptide predictors and one non-classical secretion signal predictor were used , and the majority ( 74% ) of P . guiyangense CRN proteins were predicted to contain potential secretion signals , suggesting that these CRN proteins very likely were secreted into mosquito hosts . Once inside host tissue , the roles of these putative effectors in animal pathogenic oomycetes remains unclear . One investigation detected CRN effectors in an entomopathogenic oomycete , Lagenidium giganteum , but their roles in the mosquito pathogenicity process remained unclear [52] . In this study , twenty-six CRN candidates were characterized in P . guiyangense and insect cell line transformation experiments revealed that CRN31 and CRN28 were toxic to Spodoptera frugiperda ( Sf9 ) cells and to a lesser extent to Aedes albopictus ( C6/36 ) cells . Therefore , we speculate that P . guiyangense has evolved distinct lineages of CRN effectors that are secreted into mosquito cells as virulence factors to induce host cell death . Overall , we have demonstrated that two infection routes are available for infection of mosquitoes by P . guiyangense . The high-quality genome sequence of P . guiyangense provides new insights into study oomycete evolution and host adaptation because it is an oomycete pathogen that has adapted to mosquitoes . Genome comparisons suggest adaptations to a mosquito-pathogenic lifestyle include loss of plant cell wall degrading enzymes and NLP proteins , and expansions of kinases , proteases , and kazal-type protease inhibitors . Oomycete intracellular CRN effectors were identified and insect cell toxicity was identified in at least one of them , which could serve as a new resource to control agricultural important pests .
The P . guiyangense strain Su was kindly provided by Dr . Xiaoqing Su from Guiyang Medical University , Guiyang , China and was maintained on 10% vegetable juice ( V8 ) medium in the dark at 25 ± 1°C . The Nanjing laboratory strains of Aedes albopictus and Cx . pipiens pallens were obtained from Jiangsu Provincial Center for Disease Control and Prevention , Nanjing , China , and were kept at a temperature of 25 ± 1°C in a 16L: 8D photoperiod . For mycelia infection assays , tests were carried out in plastic cups ( capacity of 200 mL ) , each containing 25 early second-instar larvae and 4 agar plugs ( 10 mm × 10 mm in size ) of P . guiyangense mycelia in 100 mL of deionized distilled water . The numbers of dead larvae were recorded every 24 hours for 10 days and each treatment was replicated at least three times . For zoospore infection assays , zoospores were prepared according to the method previously described [54] , and then batches of 25 early second-instar larvae were exposed to a concentration of 107 zoospores ml-1 in individual cups to examine the cuticle penetration process . The progress of infection in the larvae was initially documented using light microscope every 2 h for 48 h . For scanning electron microscopy ( SEM ) , representative larvae were collected at 0 . 5 , 2 , 4 , and 48 hpi and fixed in 2 . 5% glutaraldehyde solution . The fixed larvae were then rinsed three times in 0 . 1 M PBS , dehydrated sequentially in 30% , 50% , 70% , 80% , 95% and 100% ethanol , subjected to critical point drying , mounted , and finally gold coated for viewing . To investigate the digestive system infection , larvae were inoculated with 4 agar plugs ( 10 mm × 10 mm in size ) of P . guiyangense mycelia . After different times post-inoculation , larvae were examined with SEM as described above or else embedded in paraffin , sectioned , and stained with haematoxylin and eosin for light microscope observation . High-quality genomic DNA of P . guiyangense was prepared and submitted for genome sequencing using the PacBio and Illumina NGS platforms by Berry Genomics Corporation . The 450-bp paired-end libraries were constructed and sequenced on the Illumina HiSeq 2500 platform . The resultant short reads were processed to remove adapter sequences and low-quality sequences , resulting in 10 . 32 Gb of clean data ( approximately 100-fold coverage ) . Two PacBio 20-kb SMRTcell libraries were constructed and sequenced on the Sequel platform . The generated raw reads were filtered by trimming the adapter sequences and low-quality sequences . This produced 7 . 55 Gb of cleaned sequences , with an average cleaned read length of 7 . 12 kb ( approximately 70-fold coverage ) . Both the Illumina and PacBio SMRT sequencing data were used for the genome assembly . The de novo assembly was produced using the PacBio Hierarchical Genome Assembly Process HGAP 3 . 0 program [55] . First , the PacBio SMRT sequence data were error-corrected using the long filtered read and sub-read cyclic consensus sequences using HGAP error correction . The error-corrected long reads were then assembled using HGAP with default parameters . The Illumina paired-end reads were aligned to the PacBio assembly with BWA [56] , and paired-end reads with concordant alignments were selected with SAMtools view for error correction . A final genome assembly error correction was conducted using the Pilon tool [57] . This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession QXDM00000000 . 1 . The CEMGA pipeline was used to evaluate the completeness and continuity of the genome on the basis of 248 core eukaryotic genes [58] . The P . guiyangense assembly was masked for low complexity , as well as known transposable elements using RepeatMasker ( www . repeatmasker . org ) . Genes in the repeat-masked genome were predicted using two predictors , AUGUSTUS [59] and SNAP [60] . The P . guiyangense core eukaryotic genes identified by CEGMA analysis were used to train the gene predictor SNAP . The AUGUSTUS predictor was trained using P . ultimum proteins . SNAP and AUGUSTUS were then used as a part of the MAKER software to conduct the gene prediction . Protein sequences from six sequenced Pythium species , plus Ph . sojae , Ph . infestans , H . arabidopsidis and S . parasitica were submitted to MAKER as extrinsic sources of gene structure evidence to improve sensitivity of gene predictions . The transcripts discovered based on the RNA-Seq data were also submitted to MAKER as EST evidence . MAKER was set to filter out short gene models that produce proteins with fewer than 30 amino acids . All the protein sequences from P . guiyangense were searched against themselves using the BlastP program with the E-value setting to 10−10 . Then , the BlastP result file and the GFF file of the P . guiyangense genome were inputted into the MCScanX program to analyze the synteny blocks and homologous genes located in synteny blocks [24] . Circular representations of these homologous genes were produced using Circos program [61] . The Ks values of each pair of homologous genes were calculated using KaKs_Calculator 2 . 0 [62] . Whole genome protein families were classified by Pfam analysis [63] . The proteomes were screened for CAZymes ( carbohydrate active enzymes ) using Hmmscan from the HMMER package and the dbCAN HMM profile database [64] . Putative proteases and protease inhibitors were identified by using batch BLAST at the MEROPS server [65] . Protein kinases were classified by Hmmsearch against KinBase ( www . kinase . com ) . A sample of 30 early second-instar larvae inoculated with P . guiyangense mycelia was collected at 24 hpi as the infection stage , and another sample of P . guiyangense mycelia was harvested as the control . There were no biological replicates . The total RNAs of the two samples were extracted according to the method previously described [54] , and then sequenced by Berry Genomics Corporation using the Illumina 2500 platform . The 150 bp paired-end reads were filtered for quality as described above and aligned to the P . guiyangense genome assembly using Tophat with a maximum of two mismatches [66] . The mapped reads were quantified using the Cufflinks program [67] , and the transcript level of each gene was quantified as RPKM ( reads per kilobase transcript length per million reads mapped ) . Differentially expressed genes were identified using the GFOLD algorithm [68] . GFOLD was developed for unreplicated RNA-Seq data and assigns statistics for expression changes based on the posterior distribution of log fold change . Genes with four-fold change and GFOLD > 1 or < -1 were considered differentially expressed between two samples . The reads were also assembled de novo using the Trinity package [69] with default settings to serve as additional evidence for gene prediction . To detect the transcript levels of particular P . guiyangense genes during infection stages , qRT-PCR assays were performed . The samples of early second-instar larvae inoculated with P . guiyangense mycelia at different infection time points were collected , and another sample of P . guiyangense mycelia was harvested as the control . Then the total RNAs of the above samples were extracted for qRT-PCR assay . qRT-PCR was performed using an ABI Prism 7500 Real-Time PCR system ( Applied Biosystems ) with SYBR Premix Ex Tag according to the manufacturer’s instructions . The comparative threshold cycle ( Ct ) method was used to determine relative transcript levels through ABI 7500 System Sequence Detection Software . The relative transcript levels of particular P . guiyangense genes were normalized to the mycelia data using the actin gene as internal standard . At least three biologically independent replicates of the qRT-PCR experiments were carried out . A phylogenetic analysis was conducted on the core eukaryotic genes identified using the CEGMA pipeline . Multiplex sequence alignments of proteins were made with ClustalW [70] and subsequently concatenated . A neighbor-joining tree was built using MEGA5 with 1000 fold bootstrapping for distance estimation [71] . Orthologous and paralogous groups among the seven Pythium genomes were determined using OrthoMCL with default parameters: BLASTp E-value cutoff of 10−5 and inflation index of 1 . 5 [72] . The output of OrthoMCL was parsed to separate core , conserved and specific clusters . Pfam domain enrichment analysis was undertaken on genes that were specific to P . guiyangense . The fold-enrichment corresponds to the frequency of a given PFAM domain within a specific gene set divided by the frequency in the rest of the P . guiyangense proteome; a chi-square test with p-value <0 . 05 was used for significance tests . The elicitin domain ( PF00964 ) was retrieved from the PFAM database [63] , and then used to search against P . guiyangense proteome . Hits with E-value less than 10−5 were considered to be elicitin candidates . To distinguish elicitin ( ELI ) and elicitin-like ( ELL ) proteins , the previously known sequence features of the elicitin domain were used [28] . ELIs contain a highly conserved 98-amino acid domain with six cysteine residues and a typical cysteine spacing pattern . ELLs possess shorter or longer elicitin domains and sequences are more diverse . For CRN effector prediction , well-characterized CRN proteins from Ph . sojae and Ph . infestans were obtained from a previous publication [14] , and then were used to construct HMM profiles based on the LFLAK and HVLVVVP motifs . The HMM profiles were used to search against all possible proteins derived from six open reading frames of the genome . The resulting CRN candidates were manually examined for the presence of LFLAK and HVLVVVP motifs . After that , the validated CRN proteins were used to update the HMM profile , which was then used to search the protein database again for new candidates . To determine whether CRN candidates were full length or pseudogenes , we aligned the CRN candidates with previously characterized CRN proteins . If the CRN candidates shared similar sequence with known CRNs at the DNA level , but had an obvious frameshift mutation or earlier stop codon , they were considered to be pseudogenes . To predict secretion signals for CRN proteins , four signal peptide predictors including SignalP 3 . 0 ( http://www . cbs . dtu . dk/services/SignalP-3 . 0/ ) , SignalP 4 . 1 ( http://www . cbs . dtu . dk/services/SignalP/ ) , iPSORT ( http://ipsort . hgc . jp/ ) , PrediSi ( http://www . predisi . de/ ) , and one non-classical secretion signal predictor named SecretomeP 2 . 0 ( http://www . cbs . dtu . dk/services/SecretomeP/ ) , were utilized . Spodoptera frugiperda sf9 cell lines were cultured with sf-900™ III SFM medium ( Gibco ) at 27°C and 140 rpm in suspension flasks until they reached 2×106 cells/mL . Sf9 cells were subcultured using fresh medium every 3 days . The mosquito C6/36 cell lines were cultured in Schneider’s Drosophila Medium ( Gibco ) with 10% Fetal Bovine Serum ( Gibco ) at 27°C and the culture medium was renewed every 3 days . Cell density was determined using a Countess Automated Cell Counter ( Invitrogen ) and cell viability was evaluated by staining with trypan blue exclusion dye . To evaluate CRN protein toxicity , Sf9 cells were transfected with DNAs encoding CRN proteins . The ORFs of CRNs excluding the signal peptide were amplified , and CRNs without the signal peptide predicted by SignalP were amplified by excluding the N-terminal twenty-five amino acids . The PCR products were cloned into pIB/V5-His vector ( Invitrogen ) using ClonExpress II One Step Cloning Kit ( Vazyme ) . Sf9 cell suspensions were seeded into 96-well plates ( 100 μL/well ) . After 24 h of incubation , cells were transfected with 0 . 2 ug plasmid DNA using the FuGENE HD Transfection Reagent ( Promega ) as described by the manufacturer , and six parallel wells were used in each group . Cell toxicity was detected by cell counting with the Cell Counting Kit-8 ( CCK-8 , Dojindo Laboratories Kumamoto , Japan ) according to the manufacturer’s instructions . Briefly , after transfection of the Sf9 cells for 60 hours , 10 μL CCK8 solution was added to the cells . After the cells were incubated for 24 hours at 27°C , the absorbance was analyzed at 450 nm with a reference wavelength of 600 nm using SpectraMax M5 microplate reader . The cells receiving empty vector DNA were considered as 100% viable . Then , the cell viability rate was calculated as follows: Cell viability ( % ) = [ ( As-Ab ) / ( Ac-Ab ) ]×100% , where As represents the test well reading , Ac represents the empty vector well reading , and Ab represents a blank well reading . The data are expressed as the means ± SE based on at least three independent experiments . CRN constructs were compared to empty vector DNA using Student’s t-test . A difference with P < 0 . 05 was considered to be statistically significant . To validate the toxicity of specific CRN proteins , the relevant CRN genes excluding the signal peptide were inserted in frame into the pET32a ( + ) vector ( Novagen ) by directional cloning between the BamHI and HindIII sites . The pET32a empty vector was used as a negative control . Cell growth and induction of expression were carried out as described in the pET system manual . Briefly , Escherichia coli BL21 ( DE3 ) strains were grown at 37°C to an OD600 of 0 . 6 . At that time , 1 mM IPTG was added in order to induce protein expression at 18°C for 18 h . Induced cells were harvested by centrifugation at 4°C and washed three times with PBS buffer . The cells were resuspended in PBS buffer and lysed using short ( 3 s ) ultrasonic bursts separated by 6 s intervals for 6 min . Crude protein extracts were centrifuged for 10 min at 12 , 000 rpm . Then the extracts were filtered with a 0 . 22 um filter . Adherent sf9 or C6/36 cell monolayers in 96-well plates were incubated with the protein extracts for 4 hours . Cell viability was assayed using the methods in the above paragraph , and at least three independent repeats were performed . Cells were lysed in ice-cold lysis buffer ( Solarbio ) for 10 min . Following this , samples were centrifuged at 12 , 000 g at 4°C for 5 min . The supernatants were collected and boiled with loading buffer at 100°C for 10 min . The samples were separated by 10% SDS-PAGE and transferred onto a polyvinylidene difluoride membrane ( Millipore , Billerica , MA , USA ) . Membranes were blocked with 5% non-fat milk then incubated with anti-His primary antibody for 2 h . The membranes were washed with 0 . 1% Tween 20 in PBS and probed with IRDye 800CW-conjugated goat ( polyclonal ) anti-mouse IgG secondary antibodies for 1 h at room temperature . PVDF membranes were visualized using a scanner ( LI-COR Odyssey ) with excitation at 700 and 800 nm .
|
Utilization of biocontrol agents has emerged as a promising mosquito control strategy , and Pythium guiyangense has wide potential to manage diverse mosquitoes with high efficiency . However , the molecular mechanisms underlying pathological processes remain almost unknown . We observed that P . guiyangense invades mosquito larvae through cuticle penetration and through ingestion of mycelia via the digestive system , jointly accelerating mosquito larvae mortality . We also present a high-quality genome assembly of P . guiyangense that contains two distinct genome complements , which likely resulted from a hybridization of two parental species . Our analyses revealed expansions of kinases , proteases , kazal-type protease inhibitors , and elicitins that may be important for adaptation of P . guiyangense to a mosquito-pathogenic lifestyle . Moreover , our experimental evidence demonstrated that some Crinkler effectors of P . guiyangense can be toxic to insect cells . Our findings suggest new insights into oomycete evolution and host adaptation by animal pathogenic oomycetes . Our new genome resource will enable better understanding of infection mechanisms , with the potential to improve the biological control of mosquitoes and other agriculturally important pests .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"taxonomy",
"oomycetes",
"enzymes",
"computational",
"biology",
"microbiology",
"enzymology",
"invertebrate",
"genomics",
"fungal",
"structure",
"developmental",
"biology",
"fungi",
"plant",
"science",
"mycelium",
"phylogenetics",
"data",
"management",
"phylogenetic",
"analysis",
"genome",
"analysis",
"plant",
"pathology",
"mycology",
"computer",
"and",
"information",
"sciences",
"proteins",
"life",
"cycles",
"evolutionary",
"systematics",
"animal",
"genomics",
"biochemistry",
"eukaryota",
"plant",
"pathogens",
"genetics",
"biology",
"and",
"life",
"sciences",
"proteases",
"genomics",
"evolutionary",
"biology",
"larvae",
"organisms"
] |
2019
|
Infection mechanisms and putative effector repertoire of the mosquito pathogenic oomycete Pythium guiyangense uncovered by genomic analysis
|
Malaria transmission remains high in Sub-Saharan Africa despite large-scale implementation of malaria control interventions . A comprehensive understanding of the transmissibility of infections to mosquitoes may guide the design of more effective transmission reducing strategies . The impact of P . falciparum sexual stage immunity on the infectious reservoir for malaria has never been studied in natural settings . Repeated measurements were carried out at start-wet , peak-wet and dry season , and provided data on antibody responses against gametocyte/gamete antigens Pfs48/45 and Pfs230 as anti-gametocyte immunity . Data on high and low-density infections and their infectiousness to anopheline mosquitoes were obtained using quantitative molecular methods and mosquito feeding assays , respectively . An event-driven model for P . falciparum sexual stage immunity was developed and fit to data using an agent based malaria model infrastructure . We found that Pfs48/45 and Pfs230 antibody densities increased with increasing concurrent gametocyte densities; associated with 55–70% reduction in oocyst intensity and achieved up to 44% reduction in proportions of infected mosquitoes . We showed that P . falciparum sexual stage immunity significantly reduces transmission of microscopic ( p < 0 . 001 ) but not submicroscopic ( p = 0 . 937 ) gametocyte infections to mosquitoes and that incorporating sexual stage immunity into mathematical models had a considerable impact on the contribution of different age groups to the infectious reservoir of malaria . Human antibody responses to gametocyte antigens are likely to be dependent on recent and concurrent high-density gametocyte exposure and have a pronounced impact on the likelihood of onward transmission of microscopic gametocyte densities compared to low density infections . Our mathematical simulations indicate that anti-gametocyte immunity is an important factor for predicting and understanding the composition and dynamics of the human infectious reservoir for malaria .
The 2007 call issued by Bill and Melinda Gates for malaria eradication [1] has rapidly gained momentum . Malaria has plummeted globally and numerous countries continue to make significant progress toward elimination [2–5] . Nevertheless , more than a dozen countries in Sub-Saharan Africa , representing approximately 80% of the remaining global malaria burden , still need to considerably improve malaria control interventions substantially to reach levels of transmission that are suitable for elimination strategies [6] . In these countries where climate [7] and ecological conditions favor the vector’s life cycle , malaria transmission remains high and is likely compounded by under-resourced health care systems , resilient vector populations , and , obligatory for ongoing transmission , a reservoir of infectious individuals [8–11] . Although prioritizing vector control , early diagnosis and treatment of symptomatic malaria currently remain the foundation of malaria control , actual assessment and management of the human infectious reservoir may accelerate transmission reduction in highly endemic countries and contribute to sustaining elimination progress where transmission is low [12 , 13] . Considerable efforts in diagnostics have been made to improve malaria case management and elimination strategies . More sensitive molecular methods now detect previously undiagnosed low-density infections [14–16] while point-of-care diagnostics such as rapid diagnostic tests play an important role in case management and may contribute considerably to elimination strategies once their sensitivity is improved [17] . Parasite detection or quantification form imperfect proxies for the infectious reservoir of malaria that can currently only be assessed by mosquito feeding assays [18–20] . P . falciparum sexual stage immunity may restrict parasite development within the mosquito’s midgut [21–25] , and may thus affect the composition and dynamics of the infectious reservoir that is not only dictated by parasite or gametocyte densities [26] . The infectious reservoir for malaria negatively associates with age [11 , 27 , 28] , with conflicting evidence on the contribution of submicroscopic infections to transmission [11 , 29–33] . The impact of sexual stage immunity on the human infectious reservoir has never been studied and is not incorporated in mathematical models that play an increasingly important role in campaign strategies and policy making decisions [17 , 34–36] . Naturally acquired sexual stage immunity is predominantly antibody-mediated rather than cellular [37 , 38] . Antibodies against antigens that are shared between gametocytes and gametes are acquired upon natural gametocyte exposure and , once ingested as part of the mosquito blood meal , may interact with gamete surface-expressed antigens such as Pfs48/45 and Pfs230 [24 , 39–45] and thereby reduce or completely block human-to-mosquito transmission . Functional sexual stage immunity , transmission reducing activity ( TRA ) , can be estimated using an in vitro standard membrane feeding assay ( SMFA ) [46] in which a mixture of cultured gametocytes and purified antibodies is fed to mosquitoes . However , the SMFA may only detect the highest level of TRA in endemic sera , partly due to its high stringency and reliance on unnaturally high gametocyte densities and on no local malaria vector for African settings , Anopheles stephensi [47] . It was recently shown that functional TRA that affects the transmission from natural infections may be missed by the SMFA [48] . The SMFA may thus not be as sensitive to detect natural transmission reducing activity of field serums that were induced against concurrent lower or submicroscopic gametocyte densities . In the current study , we use direct membrane feeding ( DMFA ) measurements that assess the infectiousness of naturally acquired gametocytes to locally relevant vectors . DMFA experiments were performed during natural infections in all age groups from the entire range of infectious gametocyte densities and allow for the first time a direct assessment of the impact of TRA on natural malaria transmission [25 , 41 , 49 , 50] . Despite the unique richness of the present data , which include hundreds of concurrent measurements of parasite densities , serological markers , and human infectiousness , the underlying dynamics between longitudinal sampling points are complicated: seasonality impacts the timing of new infection events; age-dependent blood-stage immunity modulates the distribution of oscillating parasite densities; and recent gametocyte exposure drives changes in sexual-stage antibody concentrations . In order to assist in the interpretation of statistical relationships observed in these data , we have taken an existing model of infection and immunity dynamics in Burkina Faso [51] and extended it to include boost and decay of sexual-stage antibodies . The goal of the present modeling work is not to identify the optimal model structure or parameterization for gametocyte density and sexual-stage antibody dynamics; that will require significantly more temporal resolution in future longitudinal surveys . Rather , the goal throughout will be to compare which features of the data might follow directly from a set of plausible assumptions about unobserved dynamics , and to highlight how this should impact their interpretation .
The study received ethical clearance from the Ethical Review Committee of the Ministry of Health of Burkina Faso ( MS/MESSRS/N° 2007–035 ) . Study procedures , risks and benefits were explained to participants and written informed consent obtained from adults and parents/guardians of children prior to enrolment . Negative and positive control plasma samples [25] used in the ELISA were anonymized and provided by Sanquin blood bank ( Nijmegen , the Netherlands ) to the Department of Medical Microbiology , Radboud University Medical Center , Nijmegen , for malaria research purpose and approved by institutional Review Board for use in the present study . The serological and oocyst data were collected through a longitudinal study on the infectious reservoir [11] . Data are deposited in the Dryad repository: ( http://dx . doi . org/10 . 5061/dryad . v60jk42 ) [52] . The study design and procedures for data analysis and modeling are summarized in Fig 1; clarifying where conventional statistical models were used to present and analyze study data and where data were incorporated into the EMOD malaria model to simulate their impact on malaria transmission and the human infectious reservoir for malaria . Study participants of all ages living in a hyper-endemic and seasonal malaria transmission setting of Burkina Faso were randomly recruited from four age groups ( < 5 years , 5–14 years , 15–30 years and above 30 years of age ) in proportions that reflect local demographic proportions ( 20% , 30% , 25% and 25% respectively ) . Individuals were invited to participate; and the first to arrive were enrolled until samples sizes for the different age groups were reached . To study infectivity to mosquitoes in relation to parasite density , age , season and sexual stage immunity , participants were invited to donate blood samples . These samples were used for parasite detection , mosquito feeding assays and assessment of specific anti-gametocyte antibodies ( prevalence and ELISA optical density as proxy for antibody density ) . These assessments were performed at the start of the wet season , the peak of the wet season and during the following dry season . As described elsewhere [11] , we were not able to estimate samples sizes based on experimental mosquito’s infection rates due to the paucity of age and season-related mosquito feeding data . Serological sampling for the current study was never presented but part of a previous study that compared parasite prevalence between age groups . For that study question , that informed sample size , pre-existing age profiles of parasite rates determined by QT-NASBA in the present study area [15] were used . An enrollment and follow-up of 50 volunteers < 15 years and 50 volunteers ≥ 15 years of age would allow over 85% power to detect a significant difference in gametocyte prevalence between the two age groups at the two-tailed 5% significance level . All statistical analyses were performed in R version 3 . 3 . 1 . Gametocyte positive samples in the Pfs25 mRNA QT-NASBA were considered submicroscopic if negative by microscopy . Median densities ( optical density ) of Pfs230 and Pfs48/45 antibodies were categorized by age group ( 1–4 , 5–14 or 15+ years ) , season ( start-wet , peak-wet or dry ) and gametocyte density ( sub-microscopic or microscopic ) to describe antibody density dynamics . The proportion of seropositive individuals and infected mosquitoes were categorized by age group ( 1–4 , 5–14 , 15–30 and 30+ years ) to describe antibody prevalence and mosquito infection dynamics . A Wilcoxon rank sum test was used to compare the mean oocyst density between samples seronegative versus seropositive to Pfs48/45 and Pfs230 antigens at different gametocyte densities ( Submicroscopic , 0 . 01–16 , 17–100 , ≥100 gametocytes/μL ) . A statistical test was considered significant if p ≤ 0 . 05 . A generalized linear mixed effect model ( GLMM ) was used to assess the impact of P . falciparum sexual stage immunity on oocyst intensity . The intensity of P . falciparum oocysts per individual mosquito midgut and per participant was used in this model as a proxy of human-to-mosquito transmission capability . Because natural malaria infections in endemic areas are unique to individuals , sophisticated regression analyses that account for confounding factors are required to isolate the effect of specific immune factors on malaria transmission . The GLMM grounded in R software AD Model Builder framework accounting for zero inflation , over-dispersion and repeated measurements was used to estimate the effects of anti-Pfs48/45 and Pfs230 antibodies on oocyst intensity assuming a negative binomial distribution of oocysts counts in the mosquito population . Only gametocyte positive individuals ( i . e . with density ≥ 0 . 01 gametocyte/μL by QT-NASBA ) were included in the model . Factors influencing human-to-mosquito transmission including participant’s age , season and microscopy asexual parasite prevalence were included as covariates in the model in addition to gametocyte density as key predictor . Given that all mosquitoes were exposed to a constant duration of blood meal uptake in a repeated measurement scenario , subject’s identity and number of dissected mosquitoes were included as random effect and offset variables respectively to allow for within-subject correlation and normalization of the variable number of dissected mosquitoes across individuals . Transmission reducing-effects of model were estimated based on corresponding adjusted regression coefficients . Regression coefficients ( β values ) that associated with Pfs48/45 and Pfs230 antibodies were transformed into odd ratios ( eβ ) and the transmission-reducing activity of P . falciparum sexual stage immunity on oocyst intensity defined as TRAoocyst = 1 − eβ . A birth cohort of 100 individuals was simulated over a period of 50 years using the EMOD model configured as in [51] with incidence of new infections matching the seasonal transmission characteristics of Burkina Faso . Simulated sexual-stage antibody concentrations for Pfs48/45 and Pfs230 were boosted in response to simulated gametocyte densities each day before decaying with a 1-month half-life—consistent with the immunoglobulin G half-life [56 , 57] and with repeat observations from this and other cohort studies ( S1 Fig ) . Procedures for antibody half-life estimates are described in S1 Appendix in section 1 and 2 . An alternative approach to estimate antibody half-life that fits a simple equation to time between serological measurements estimated half-lives of 46 . 4 days for Pfs230 and 142 days for Pfs48/45 ( see S1 Appendix – Section 1 ) . The reason for the disparity in estimates for Pfs48/45 is uncertain but may be related to antibody boosting by regular antigen re-exposure that is evident in the study site [11 , 58 , 59] and not incorporated in the simple antibody decay model . Boosting rates were fitted for each antigen to match the observed features of antibody density distributions—median densities and inter-quartile ranges by age , season , and current gametocyte density—under the structural assumptions that higher gametocyte densities may boost more strongly and that age-dependent cumulative exposure may impact the magnitude of the boost and the inter-individual variation within a smoothness constraint ( S1 Appendix ) . An age-matched set of simulated trajectories was sampled at the 3 time points corresponding to the survey data .
The overall prevalence of NANP6 antibodies was 78 . 1% ( 222/284 ) and this antibody prevalence increased significantly with age ( Adjusted β ( regression coefficient ) = 0 . 838 , se = 0 . 288 , p = 0 . 0037 , Table 1 ) . The density of NANP6 antibodies also increased with age ( β = 1 . 032 , se = 0 . 29 , p < 0 . 001 ) and season with the lowest levels recorded at the start ( 0 . 4752 ) of the wet season compared to the peak ( 0 . 7165 ) and dry season ( 0 . 8302 ) ( β = -0 . 387 , se = 0 . 164 , p = 0 . 018 ) . Compared to NANP6 , a smaller proportion of samples showed reactivity to Pfs48/45 ( 22 . 6% ) and Pfs230 ( 33 . 6% ) ( Table 1 ) ; only 12% ( 36/291 ) of the samples reacted to both Pfs230 and Pfs48/45 antigens . Antibody density and seroprevalence were lower for Pfs48/45 compared to Pfs230 . Pfs230 antibody prevalence was significantly higher in adults ( β = 1 . 150 , se = 0 . 445 , p = 0 . 0097 , Table 1 ) , and adults also had higher relative antibody densities ( β = 0 . 5289 , se = 0 . 2502 , p = 0 . 0345 ) . In contrast to Pfs48/45 ( Fig 2g ) , the prevalence and density of Pfs230 antibodies in individuals ≥15 years of age dropped at the peak of the wet season , although not significantly ( Fig 2c ) . The prevalence of Pfs48/45 antibody increased with age up to 30 years but declined in older age ( adjusted β = 0 . 871 , se = 0 . 507 , p = 0 . 086 , Table 1 ) , and prevalence was higher ( adjusted β = 0 . 470 , se = 0 . 3012 , p = 0 . 12 ) at the peak of the wet season ( 31% , 30/96 ) compared either to the start of the wet season ( 17% , 18/102 ) or to the dry season ( 19% , 18/93 ) . Interestingly , although not statistically significant , the decline in Pfs48/45 antibody after 30 years of age coincided with a drop in the mean gametocyte density in that population from onset ( 80 . 5 gametocytes / μL ) to peak transmission ( 58 . 7 gametocytes / μL ) . Pfs48/45 antibody density also showed an age-related trend ( Adjusted β = 0 . 2581 , se = 0 . 2629 , p = 0 . 451 ) and reached a peak during the wet season ( Fig 2e and 2g ) although not significantly ( β = 0 . 1064 , se = 0 . 1752 , p = 0 . 544 ) . Reflecting on the characteristics of the dynamical model for Pfs230 ( Fig 2d ) and Pfs48/45 ( Fig 2h ) that are required to reproduce the age and seasonal trends observed in the surveys , two keys features are apparent . First , a relatively short decay half-life is an important driver of both the seasonal variation and also the inter-individual variation driven by periodic spikes in gametocyte density ( Fig 3 shows an example trajectory ) . Second , a combination of chronic submicroscopic boosting and more efficient boosting with cumulative exposure influences the increasing densities with age , especially for Pfs230 . We further investigated the impact of concurrent gametocyte carriage , either sub-microscopic or microscopic , on measured P . falciparum sexual stage antibody densities to better understand antibody acquisition and decay upon parasite exposure . In Fig 2a , under 5 , 5–14 and 15+ years old submicroscopic gametocyte carriers presented with 0 . 3515 , 0 . 6597 and 0 . 8394 median Pfs230 antibody densities respectively , which were lower though not significantly so , compared to densities in microscopic gametocyte carriers ( 0 . 4597 , 0 . 7808 , 1 . 1183 respectively ) . Similarly , in Fig 2e , under 5 , 5–14 and 15+ years old submicroscopic gametocyte carriers presented with 0 . 4640 , 0 . 525 and 0 . 594 median Pfs48/45 antibody densities respectively whilst microscopic gametocyte carriers had increased though not significantly so Pfs48/45 median antibody densities ( 0 . 5865 , 0 . 4395 and 0 . 9115 respectively ) . GLMM regression models that included age and parasite status showed that microscopic gametocyte carriers were more likely to have higher Pfs230 ( Adjusted OR = 1 . 52 , se = 0 . 302 , p = 0 . 164 ) and Pfs48/45 antibody densities ( Adjusted OR = 1 . 4 , se = 0 . 3282 , p = 0 . 34 ) than submicroscopic gametocyte carriers , although not statistically significantly so . Whilst anti-gametocyte immunity has been hypothesized to be short-lived [39 , 60] anti-Pfs230 and anti-Pfs48/45 antibodies were detected in a considerable fraction of individuals without concurrent gametocytes , plausibly reflecting recent gametocyte exposure . Under 5 , 5–14 and 15+ years old gametocyte-negative individuals presented with 0 . 2031 , 0 . 7868 and 0 . 7936 median Pfs230 antibody densities and 0 . 5662 , 0 . 6332 and 0 . 6524 median Pfs48/45 antibody densities , respectively , which tend to be lower than densities in submicroscopic and microscopic gametocyte carriers ( S2 Fig ) . There was no association or trend between NANP6 antibodies and concurrent gametocyte density ( S3 Fig ) . The observed pattern of a small but consistent shift in antibody densities for concurrently detected gametocytes is an expected consequence of the dynamical model for Pfs230 ( Fig 2b ) and Pfs48/45 ( Fig 2f ) . Quantitatively , it depends on the gametocyte density varying on a timescale that is similar to the antibody decay time and the degree to which higher gametocyte densities resulting in more efficient boosting . P . falciparum gametocytes were detected in 66% of samples ( 188/284 ) by QT-NASBA ( Table 1 ) whilst 32% ( 96/291 ) of the total population was infectious to mosquitoes ( i . e . infected at least one mosquito ) . P . falciparum oocyst counts as enumerated in mosquito’s midguts followed a zero-inflated negative binomial distribution ( S1 Appendix – Fig 3 ) . Among the 11 , 440 dissected mosquitoes , 886 mosquitoes were found to be infected with one or more oocysts ranging from 1 to 97 oocysts . Predictors of oocyst intensity were tested in a GLMM , and the estimated coefficients are shown in Table 2 . The intensity of oocysts significantly decreased with age and was lower in the dry season as compared to the transmission season . In mosquitoes that fed on blood samples from individuals ≥ 15 years of age , the intensity of oocysts was significantly lower compared to oocyst intensity of mosquitoes fed on blood samples from individuals 5–14 ( p = 0 . 003 ) and under-5 years old children ( p = 0 . 047 , Table 2 ) . Similarly , season was related to oocyst with an increased oocyst intensity at the start ( p = 0 . 075 ) and peak of the wet season ( p < 0 . 001 , Table 2 ) . Carriers of gametocytes ( ≥100 gametocytes / μl blood ) or microscopic asexual parasites positively associated with oocyst intensity ( p < 0 . 001 for both covariates , Table 2 ) . By investigating the interactions between oocyst counts and antibody densities , adjusting for gametocyte density , we were able to estimate the transmission reducing activity ( TRA ) of specific antibody densities . Fig 4 illustrates TRA in relation to Pfs230 , Pfs48/45 and NANP6 antibodies at various gametocyte densities determined by QT-NASBA . At gametocyte density ≥100 gametocytes/μL , individuals seropositive to both Pfs230 and Pfs48/45 associated with an order of magnitude reduced oocyst intensities ( 11 and 11 . 3-fold respectively , Fig 4a and 4b ) in comparison to seronegative individuals ( p < 0 . 001 ) . At gametocyte density of 17–100 gametocytes/μL , samples seropositive for Pfs230 significantly reduced oocyst intensity compared to samples seronegative for Pfs230 ( p = 0 . 03 , Fig 4 ) . The same was not true for samples seropositive for Pfs48/45 ( Fig 4 ) . At gametocyte density below 16 gametocytes/μL or at the submicroscopic level , we observed no statistically significant impact for Pfs230 or for Pfs48/45 seropositivity on oocyst intensity ( p = 0 . 937 and p = 0 . 251 respectively ) . The GLMM regression analysis estimated—after adjustment for age , season , gametocyte density and microscopically detectable asexual parasites—a 55 . 1% and 70% reduction in oocyst intensity respectively for Pfs48/45 antibody ( p < 0 . 001 , Table 2 ) and Pfs230 ( p < 0 . 001 , Table 3 ) . No reduction of oocyst intensity associated with NANP6 antibodies was observed ( Fig 4c ) . Simulated gametocyte densities generated an overall fraction of 39 . 9% ( 5998/15000 ) infectious individuals with an overall mean mosquito infection rate of 5 . 6% ( 850/15000 ) and oocyst intensity ranging from 0 to 94 ( including non-infectious individuals ) . Based on the values of TRA estimated from the GLMM regression analysis , the presence of antibody with specificity for Pfs48/45 or Pfs230 resulted in reductions of 26% , 37% respectively and for both Pfs48/45 and Pfs230 a reduction of 44% in mosquito infection rates ( Fig 5 ) . Our findings of an age and season-dependent prevalence and density of antibodies to Pfs48/45 and Pfs230 and a considerable impact of these antibodies on the likelihood of mosquito infection indicate that incorporating sexual stage immunity in transmission models may improve our ability to quantify the contribution of different populations to the infectious reservoir for malaria . Fig 6a incorporates the smearing of true simulated gametocyte densities to account for uncertainty in QT-NASBA derived gametocyte density estimates ( S1 Appendix – Fig 5 [61] ) . This results in a clear over-dispersion of infected mosquitoes spanning a wider range of measured gametocyte density ( 100–104 gametocytes/μL ) . But another important consequence is that high gametocyte density measurements will appear to be less infectious in groups with lower true average density ( e . g . older ages ) , since it is proportionately more likely they result from lower true densities in the tails of the measurement uncertainty distribution [62] . Fig 6b additionally includes the effect of TRA from both Pfs230 and Pfs48/45; the color of the bubbles corresponds to the Pfs230 antibody density distribution ( as an example ) across individuals . The age- and season-dependence of high antibody densities is reflected in the changing shape of the infectious reservoir between Fig 6a and 6b . In general , there was 1 . 14 ( 11 . 3%/9 . 9% ) , 1 . 24 ( 7 . 6%/6 . 1% ) , 1 . 45 ( 3 . 2%/2 . 2% ) and 1 . 5-fold ( 0 . 6/0 . 4 ) decrease corresponding to 12 . 2 , 19 . 3 , 31 and 33% reductions in mosquito infection rate in the <5 , 5–14 , 15–30 , and 30+ year old individuals because of TRA impact , providing an improved fit to the study site data as described elsewhere [11] . There was no seasonal impact of TRA on the proportion of infected mosquitoes . The shape of the association between oocyst prevalence and oocyst intensity ( S1 Appendix – Fig 4 [62] ) was not altered by the presence of antibodies against Pfs230 or Pfs48/45 ( Fig 5 ) . Both oocyst prevalence and oocyst intensity were lower for experiments on samples with detectable antibodies to these antigens with 70% and 55% as best estimate of the reductions in oocyst intensity related to the presence of antibodies against Pfs230 and Pfs48/45 , respectively ( Fig 5 ) .
Using a unique set of longitudinal data on malaria-specific antibody responses and infectivity of gametocytaemic individuals to mosquitoes , we showed that P . falciparum sexual stage immunity significantly reduces the transmission of microscopic but not submicroscopic gametocyte infections . A notable age-dependent decrease in mosquito infection rate is attributed to both Pfs48/45 and Pfs230 immunity in a computational model . Our simulations reflect and further support our field findings that P . falciparum sexual stage immunity associates with recent exposure to gametocyte density and duration of exposure . Whilst the presence of antibodies against P . falciparum sexual stage antigens in endemic populations is well established [24 , 39 , 49] , it is currently unknown how this sexual stage immunity develops in relation to either microscopic or submicroscopic gametocyte carriage and how it subsequently impacts the likelihood of mosquitoes becoming infected by these malaria-infected human hosts . Although there is conflicting evidence on the impact of age and cumulative malaria exposure on sexual stage immunity [63] , sexual stage immunity may influence the composition and dynamics of the human infectious reservoir . We followed an all-age cohort of individuals with variable gametocyte exposure over distinct transmission seasons and directly determined their infectiousness to mosquitoes in relation to the presence or density of sexual stage antibodies against gametocyte/gamete antigens Pfs48/45 and Pfs230 . Pfs48/45 is a pre-fertilization protein on the surface of gametocytes and gametes that has a central role in male gamete fertility [45] while Pfs230 antibodies mediate lysis of gametes in a complement-dependent manner [42] . Both antigens are transmission blocking vaccine candidates that elicit up to 100% functional transmission reducing immunity in animal models [45 , 64] . In the present study , antibodies against Pfs48/45 and Pfs230 were more commonly observed in samples with high concurrent gametocyte densities [39 , 65] . We did not directly assess antibody titer by diluting participant plasma samples but used the optical density in the ELISA assays as indicator of antibody density , as commonly done [66] . Our simulations indicate a short half-life of anti-Pfs48/45 and anti-Pfs230 antibodies [39] . Although long-lived antibody responses to gametocyte antigens have been observed in expatriates who retained functional antibodies for several years after returning from malaria endemic settings [25] , most of the field evidence supports the hypothesis that gametocyte immune responses are more short-lived than asexual responses [49 , 67] . Our estimates of an antibody half-life of approximately 1 month for both anti-Pfs48/45 and anti-Pfs230 antibodies must be interpreted with caution . Although the estimates are in line with that from an independent cohort study from a similar setting in Burkina Faso ( see S1 Fig ) , an alternative , simpler approach to estimating antibody half-life that assumes an antibody decay as function of time rather than accounting for parasite exposure in this time window estimated a longer half-life of anti-Pfs48/45 antibodies . In addition , our ELISA methodology that utilizes native gametocyte antigen may be limited in sensitivity and may thus have classified samples with low antibody densities as antibody negative . Because the functionality of sexual stage antibodies is strongly dependent on density [39 , 68 , 69] we nevertheless consider it unlikely that we have missed functionally relevant antibody densities in the current study . At microscopically detectable gametocyte densities , the presence of anti-Pfs48/45 and anti-Pfs230 antibodies had a pronounced effect on the proportion of mosquitoes that became infected and the oocyst burden in these mosquitoes . Although antibody densities have been previously associated with TRA in both SMFA [39 , 68] and DMFA [69] , these studies were all performed in selected high-density gametocyte carriers . Our study is the first to quantify its impact on the infectious reservoir for malaria by concurrent use of membrane feeding assays , antibody assessments and sensitive gametocyte detection . We observed that the presence of antibody with specificity for Pfs48/45 , for Pfs230 or for either resulted in a reduction of 26 , 37 and 44% in mosquito’s infection rate respectively . We further found that incorporating sexual stage immunity improved the age-related dynamics of the infectious reservoir in simulations . This could be explained by the age-related increase in Pfs48/45 and Pfs230 antibody prevalence and densities . These antibody responses did not explain the comparatively lower mosquito infection rates in the dry season when both antibody prevalence and mosquito infection rates were lower [11] . Older infections surviving the long dry season [70] appear to be less infectious to mosquitoes by a mechanism independent of TRA [71] . We observed a direct relationship between sexual stage antibody responses and recent / concurrent exposure to gametocytes [24 , 60 , 67 , 68] . We thus used gametocyte density and duration of exposure ( rather than age as proxy for cumulative exposure ) , as parameters to the model acquisition of P . falciparum sexual stage immunity . Our simulations closely replicate our field observations suggesting that age is unlikely to play an independent role in the dynamics of Pfs48/45 and Pfs230 immunity . However , given the higher Pfs230 antibody density in adults , we could not fully rule out a role of age on the acquisition of sexual stage immunity . Indeed , an age-dependent expansion of B cell memory may be an explanation for the observed patterns , as could be the persistent exposure to gametocyte antigens due to continued low-density gametocytemia , as observed in 40% of adults in the area [11] . While antibody data from three seasonal sampling points allowed us to infer many characteristics of antibody boosting and waning dynamics , there remains significant uncertainty in relation to dependency of antibody acquisition or maintenance in relation to gametocyte density and duration of gametocyte carriage . Future longitudinal studies with more frequent sampling will help refine our understanding in these areas specifically . To conclude , the present study provides novel insights into how P . falciparum sexual stage immunity is acquired in seasonal transmission settings and how it impacts infectivity to mosquitoes during natural infections and the human infectious reservoir for malaria . Our finding that P . falciparum sexual stage immunity has very limited impact on the transmissibility of submicroscopic gametocytes may contribute to previous findings on their important role in sustaining malaria transmission [11 , 32 , 72] . The finding that P . falciparum sexual stage immunity associates with concurrent submicroscopic and microscopic gametocyte density has implications for the future deployment of transmission blocking vaccines where antibody boosting densities may be dependent on recent high-density gametocyte exposure . Lastly , we show that P . falciparum sexual stage immunity significantly reduces the infectious reservoir in an age-dependent manner , legitimizing its incorporation in mathematical models that may be used for projecting the infectious reservoir in the context of malaria elimination and for testing future transmission-blocking interventions in silico .
|
Submicroscopic gametocyte infections are efficiently transmitted from humans to mosquitoes in settings with efficient malaria vectors and may pose challenges for malaria control and elimination efforts . Our understanding of what mechanisms contribute to submicroscopic gametocytes infectiousness remains limited . Here we assess the impact of naturally acquired anti-gametocyte antibodies on malaria transmission to mosquitoes and on the age-dependent composition of the infectious reservoir and seasonal dynamics . Anti-gametocyte immunity significantly reduces the infectiousness of high gametocyte density infections , contributes to explain the age-related profiles of the infectious reservoir in the study area , whilst submicroscopic gametocyte infections that present with lower anti-Pfs48/45 and anti-Pfs230 antibody responses commonly remain transmissible to mosquitoes . Our findings indicate that sexual stage immunity needs to be incorporated in transmission models to better understand transmission dynamics . Furthermore , tools that boost sexual stage immunity may reduce transmission to mosquitoes and thus aid elimination strategies .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"parasite",
"groups",
"immune",
"physiology",
"enzyme-linked",
"immunoassays",
"oocysts",
"gametocytes",
"immunology",
"tropical",
"diseases",
"parasitic",
"diseases",
"animals",
"parasitology",
"germ",
"cells",
"apicomplexa",
"sexually",
"transmitted",
"diseases",
"antibodies",
"immunologic",
"techniques",
"insect",
"vectors",
"research",
"and",
"analysis",
"methods",
"immune",
"system",
"proteins",
"infectious",
"diseases",
"animal",
"cells",
"proteins",
"immunoassays",
"disease",
"vectors",
"insects",
"arthropoda",
"biochemistry",
"mosquitoes",
"eukaryota",
"cell",
"biology",
"immunity",
"physiology",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"malaria",
"species",
"interactions",
"organisms"
] |
2018
|
Modeling the impact of Plasmodium falciparum sexual stage immunity on the composition and dynamics of the human infectious reservoir for malaria in natural settings
|
Influenza viruses present major challenges to public health , evident by the 2009 influenza pandemic . Highly pathogenic influenza virus infections generally coincide with early , high levels of inflammatory cytokines that some studies have suggested may be regulated in a strain-dependent manner . However , a comprehensive characterization of the complex dynamics of the inflammatory response induced by virulent influenza strains is lacking . Here , we applied gene co-expression and nonlinear regression analysis to time-course , microarray data developed from influenza-infected mouse lung to create mathematical models of the host inflammatory response . We found that the dynamics of inflammation-associated gene expression are regulated by an ultrasensitive-like mechanism in which low levels of virus induce minimal gene expression but expression is strongly induced once a threshold virus titer is exceeded . Cytokine assays confirmed that the production of several key inflammatory cytokines , such as interleukin 6 and monocyte chemotactic protein 1 , exhibit ultrasensitive behavior . A systematic exploration of the pathways regulating the inflammatory-associated gene response suggests that the molecular origins of this ultrasensitive response mechanism lie within the branch of the Toll-like receptor pathway that regulates STAT1 phosphorylation . This study provides the first evidence of an ultrasensitive mechanism regulating influenza virus-induced inflammation in whole lungs and provides insight into how different virus strains can induce distinct temporal inflammation response profiles . The approach developed here should facilitate the construction of gene regulatory models of other infectious diseases .
Invading pathogens induce acute inflammation when molecular signatures are detected by pattern recognition receptors ( PRRs; e . g . , RIG-I like receptors [RLRs] and Toll-like receptors [TLRs] ) expressed on tissue-resident immune cells and non-immune cell types . PRR ligation triggers innate immune responses and leads to the induction of inflammatory and antiviral gene expression , which together function to limit pathogen growth , activate the adaptive immune response , and ultimately resolve the infection [1 , 2] . Precise regulation of PRR-mediated signaling is necessary to both avoid inadvertent tissue damage in response to non-pathogenic stimuli , and to prevent immunopathology resulting from excessive expression of inflammatory molecules . In essence , the ideal inflammatory response must exhibit a balance between appropriate activation against a genuine threat and self-limiting behavior once that threat has been controlled . Despite its importance in maintaining normal tissue homeostasis and limiting pathogen-associated diseases , the mechanisms underlying the regulation of this balance are poorly understood . Influenza A viruses are recognized by both TLRs and RIG-I-like receptors ( RLRs ) [3–7] , and some strains are potent inducers of inflammatory and antiviral gene expression . Generally , lung tissues infected with pathogenic isolates exhibit high virus titers and robust inflammatory gene expression , as has been documented in in vivo studies with the 1918 Spanish influenza virus [8 , 9] , highly pathogenic H5N1 avian influenza viruses [10–12] , and the 2009 H1N1 pandemic influenza virus [13 , 14] . In contrast , seasonal influenza viruses typically replicate less efficiently , elicit more restrained inflammatory responses , and are usually not associated with lethal infections . Recent evidence has implicated the level of virus replication in infected lung tissues as the primary phenotypic variable driving inflammation and lethal outcomes [15 , 16] . Other data indicate that influenza viruses that exhibit significant differences in pathogenicity stimulate qualitatively similar host responses that differ primarily at the level of magnitude and kinetics [17] . However , these studies have not revealed the mechanisms that account for the different profile dynamics observed in infections by high and low pathogenic viruses . Such information would aid in clarifying not only how some influenza viruses induce lethal disease , but also the general mechanisms that regulate inflammatory balance . To characterize the dynamics of influenza virus-induced inflammation , we developed a novel approach to infer gene regulatory models from dynamic gene expression data . Referred to as systems inference microarray analysis , our method builds on current approaches that use co-expression analysis to isolate modules of functional signatures in gene expression data and then extends these methods by fitting the gene expression modules to mathematical equations ( models ) by using segmented regression analysis . Models can be created to look for strain-dependent responses and , unlike traditional differential expression analysis , to predict gene expression under new experimental conditions . By using this method , we set out to determine how influenza viruses that exhibit variable pathogenicity profiles influence the dynamics of the inflammatory response .
To characterize the dynamics of the host immune response to specific virus isolates , we infected mice with 105 PFU of three virus isolates with distinct pathogenicity profiles and harvested lung tissues at 14 time points after infection ( from 0 to 7 days post-infection; n = 3 per virus per time point; see Fig 1 ) for several parallel analyses . These viruses included a low pathogenicity seasonal H1N1 influenza virus ( A/Kawasaki/UTK4/2009 [H1N1]; referred to as ‘H1N1’ ) , a mildly pathogenic virus from the 2009 pandemic season ( A/California/04/2009 [H1N1]; referred to as ‘pH1N1’ ) , and a highly pathogenic H5N1 avian influenza virus ( A/Vietnam/1203/2004 [H5N1]; referred to as ‘H5N1’ ) . An initial inoculation of 105 PFU was used as previous studies indicated that a high virus dose was needed to invoke different pathologies in H1N1 and pH1N1-infected mice [13] . As expected , lung virus titers ( virus titers determined by plaque assay and reported in plaque forming units [PFU] per gram lung; Fig 1 ) indicated a clear hierarchy of mild , moderate , and severe virus-induced disease . Specifically , the H5N1 virus produced the highest lung titers and between days 5 and 7 post-infection , this virus also caused mortality in the animals whose lungs were to be collected on day 7 post-infection ( i . e . , ‘severe’ disease ) . In contrast , all animals infected with the H1N1 or pH1N1 viruses survived the duration of the time course study; however , pH1N1-infected animals were visibly sicker and exhibited higher lung titers relative to those infected with H1N1 at all time points observed after the first 30 h post-infection ( i . e . , ‘moderate’ and ‘mild’ disease , respectively ) . Histopathological analysis of tissue samples collected on days 1 , 2 , and 5 post-infection ( Fig 2 ) also showed that H5N1-infected tissue exhibited the earliest , most severe signs of inflammation and inflammatory immune cell infiltrates followed by pH1N1-infected tissue , whereas H1N1-infected tissue showed mild evidence of inflammation and was most similar to tissue from the control mice ( mock-infected mice ) . We next used co-expression analysis to integrate inflammation-associated gene expression differences between influenza-infected and control lungs into a systems level context . We first asked whether the expression of inflammation-associated genes clustered into modules of co-expressed genes . Tissues from the same animals that were used to determine virus growth were used to evaluate changes in global lung transcriptional profiles . A total of 168 microarrays were developed ( three per time point for H1N1-infected , pH1N1-infected , H5N1-infected and control mice ) . One microarray was removed after reviewing replicate quality . After filtering transcripts for minimally confident variation ( we required at least one time-matched , infected condition compared with mock-infected absolute fold change ≥ 2 and a false discovery rate [FDR]-adjusted P-value < 0 . 01 ) , the log2 of the normalized intensity of the retained transcripts ( 16 , 063 ) for all 167 samples were then clustered by using the Weighted Gene Co-expression Network Analysis ( WGCNA ) algorithm [18] . In all , 45 distinct co-expression modules were identified ( referred to as N1 , N2 , etc . ; S1 Table provides the module assignments for all transcripts ) . To identify the biological role of the host response modules , we performed functional enrichment analysis on each gene module by using DAVID [19] and ToppCluster [20] . Because each module was comprised of positively and negatively correlated transcripts , we used the module eigengene ( i . e . , the first principle component of the gene expression matrix ) to divide each module into two submodules containing transcripts that were positively or negatively correlated with the parent module’s eigengene , denoted as kME+ and kME- ( referred to as module membership ) , respectively . This procedure allowed us to look for biological processes with similar but opposing dynamic responses to the virus infections . Functional enrichment analyses were then applied to each submodule by using two bioformatics platforms to ensure robust results . We identified two submodules ( N1 kME+ and N2 kME- , referred to as simply N1 and N2 in the remainder of the text ) that were enriched for inflammatory response and inflammation-associated pathway signatures by using both bioinformatics platforms , and these two modules became the focus of our study ( Table 1 summarizes the functional enrichment results for the immune and inflammatory related annotations . The complete enrichment results from ToppCluster and DAVID are available in S2 Table and S1 and S2 Files ) . The N1 module was uniquely enriched for cytokine activity and type I interferon ( IFN ) regulating TLR and RLR pathways [21] , as well as transcriptional signatures associated with IFN-regulated activity ( i . e . , the transcription factor binding sites [TFBS] of Irf1 , Irf7 , Irf2 , ISRE , and NF-κB ) . Additionally , N1 was the only module that exhibited significant enrichment with a compendium of established IFN-stimulated genes ( ‘ISGs’; Table 1; see Methods . The list of ISGs is available in S3 File ) . A more recent study identified 147 IFN stimulated genes in immortalized , human airway epithelial ( Calu3 ) cells [22] . Of these , 90 mouse homologs were annotated on the microarrays and 70 of the homolog probes were assigned to the N1 module ( Fisher’s exact test; P-value < 10–16; odds ratio = 36 . 4 ) , further associating N1 interferon-stimulated gene activity . In contrast , the N2 module was only weakly enriched for some cytokine activity related annotations and not enriched for any of the binding sequences of transcription factors that are members of canonical inflammatory pathways ( such as interferons , interferon regulatory factor proteins , or NFκB ) . Instead , it was primarily associated with several annotations related to leukocyte and lymphocyte activity ( see summary of ToppCluster enrichment results in Table 1; see S1 and S2 Files ) . Further analysis with CTen [23] , a platform for associating clustered gene expression data with specific cell types , found N2 to be highly enriched for genes expressed in macrophages in various cellular states ( e . g . , bone marrow-derived macrophages exposed to lipopolysaccharide [LPS] ) ( Fig 3A; additional details available in S3 Table ) . The remaining immune-associated submodules ( the kME+ N22 , N25 , N31 , and N35 submodules; described in Table 1 ) were enriched for several key immune processes such as antigen presentation , and T cell and natural killer ( NK ) cell activity , but their further assessment would be beyond our focus and the scope of this study . Thus , bioinformatics analyses robustly associated the N1 module with inflammation , cytokine production , and type I IFN pathway activity—likely activated in resident lung cells—whereas the N2 module is associated with migration and activation of macrophages in the lung . A closer examination of the expression dynamics of each of the inflammation response-associated modules revealed patterns of expression that were consistent with the biological roles predicted by our bioinformatics analyses . We used the scaled difference of the module eigengene to characterize the expression of all genes within each module . We subtracted the mean of the eigengene of the control samples from the eigengene of time-matched , virus-infected samples and then divided by the largest observed average difference across all conditions . The resulting scaled difference eigengene ( SDE ) represents the fraction of the maximum log fold change in gene expression observed across all experimental conditions ( Fig 3B and 3C . S1 Fig briefly illustrates the eigengene scaling in greater detail . S4 Table provides a heatmap of the gene expression of all probes in each module ) . Within the inflammation-associated modules ( N1 and N2 ) , the H5N1 virus induced the earliest gene expression changes and the highest peak expression levels , corroborating previous observations that H5N1 viruses are strong inducers of inflammatory and IFN response signaling in vivo [10 , 24 , 25] . Consistent with the prediction that N1 is involved in detecting virus in infected tissues , the N1 module eigengene was the most highly correlated with virus titer ( Pearson pairwise correlation , ρ = 0 . 70 ) . Module gene expression dynamics further suggested that the N2 module gene is associated with lymphocyte infiltration . Exudate macrophages [26] and neutrophil [27] have been identified as factors of severe disease during influenza infection . To associate gene expression dynamics with changes of immune cell counts , a new population of mice were infected with the three influenza viruses , five mice per infection group were sacrificed on days 1 , 2 , 3 and 7 and the changes in the number of macrophages and neutrophils was assessed ( see Methods ) . Unlike the previous study by Brandes , et al . [27] , strong neutrophil infiltration was not specific to fatal infections but , instead , infiltration of both cell types had a clear hierarchical relationship with the severity of the infection ( Fig 3D and 3E ) . The N2 module eigengene exhibited a lesser correlation to virus titer ( ρ = 0 . 55 ) , but was tightly correlated to macrophage influx into the lung ( ρ = 0 . 90; P-value < 0 . 01 Student’s t-test ) . Of the 45 module identified in the studied , N2 had the highest correlation to both macrophage and neutrophil influx ( the correlation of macrophage and neutrophil influx and all 45 module eigengenes are provided in S5 Table; ρ = 0 . 80; P-value < 0 . 01 ) . The N1 module on the other hand was weakly but significantly correlated with immune cell infiltration ( ρ = 0 . 67 [P-value = 0 . 03] for neutrophils; ρ = 0 . 60 [P-value = 0 . 05] for macrophages ) , but its eigengene was not the most highly correlated ( 5 other module eigengenes had a greater absolute correlation ) . These results further associate N2 with immune cell-specifically macrophage- infiltration , while the sum of the bioinformatics , virus titer correlations and immune cell infiltration evidences associated N1 with inflammation and type I IFN pathway activity . A further advantage of a network approach is that the functional relevance of genes might be inferred from their positions within the co-expression network [28] . We used the module membership ( the correlation between the gene’s expression and the module eigengene , kME ) to isolate potential regulators of the N1 module . Among the top intramodular hub genes ( i . e . , genes with the highest module memberships , see S4 Table ) , we found Mnda , Herc6 and Cd274 and several interferon regulated , virus replication inhibitory genes such as Oas2 and Oas3 [29] . Herc6 is involved in ubiquitination [30] . Mnda is significantly up-regulated in monocytes exposed to interferon α [31] while Cd274 is a transmembrane protein expressed on antigen presenting cells and modulates activation of T cells , B cells and myeloid cells . We also observed that interferon stimulated genes tended to have higher intramodular hub rankings , suggesting a regulatory role for interferon ( Wilcoxon rank sum test , P-value < 10–12 ) . We then considered the module membership rankings of transcription factors known to regulate interferon . Of the established interferon regulatory factors that are members of the N1 module ( e . g . , Irf1 , Irf2 , Irf7 , Irf9 , Stat1 , and Stat2 . Nfkb1 and Nfkb2 were not assigned to N1 ) , Irf1 and Irf7 had the highest module memberships ( kME = 0 . 94 and 0 . 93; ranking = 118 and 225 , respectively ) . Irf7 expression was also several orders of magnitude greater than Irf1 ( S4 Table ) . Together , these findings corroborate our bioinformatics analyses by suggesting that N1 is regulated by interferon , and N1 expression likely results in enhanced cytopatchic effects and regulation of the lymphocyte immune response . Network analysis further suggests that Irf7 may play a regulatory role upstream of interferon transcription . Previous studies have suggested that highly pathogenic influenza virus infections induce an irregular or disproportionate inflammatory response relative to seasonal influenza viruses , and that these differences occur early in the host response [24] . For this reason , we sought to further explore the possibility of isolate-specific or isolate-independent response patterns of the cytokine-associated N1 module . We wanted to infer mathematical relationships that could describe when inflammatory-associated gene expression occurs and what magnitude of expression is expected . By using the eigengene as a representation of the scaled gene expression dynamics , we attempted to infer simple mathematical models that can be related to common signaling mechanisms . Surprisingly , when we plotted the N1 SDE for each isolate against the corresponding virus titer , we observed a consistent profile for all three viruses; regardless of intrinsic virulence , the fold change in N1 gene expression remained initially low and rapidly increased only after a virus titer of approximately ~108 PFU/g ( of lung ) was reached ( Fig 4A ) . Following activation , N1 gene expression increased as a function of virus concentration at the same apparent rate for all infection conditions , and more complicated dynamics were observed only during the later phase of the infection when virus clearance was observed ( i . e . , when the virus titers began to decrease ) . These observations suggest that IFN-regulated ( N1 ) gene expression was induced by an ultrasensitive response mechanism controlled at the level of the virus titer rather than the virus’s intrinsic virulence . Ultrasensitive responses characterize the dynamics of several signaling pathways that regulate essential and often toxic biological processes such as the cell cycle [32] and apoptosis [33] . As shown in Fig 4B , ultrasensitive responses are typified by an attenuated response to low levels of stimulation but a strong response occurs once a threshold level of stimulus is reached . Cooperativity [33] and positive feedback [34] are two mechanisms that produce ultrasensitive responses . To formalize the hypothesis that the inflammatory gene response follows an ultrasensitive response profile , we selected a segmented linear model ( SLM , defined in Fig 4C ) to be a simplified representation of the ordinary differential equations normally used to model ultrasensitive responses , and we fit the N1 SDE to an SLM that was strictly a function of the virus titer . The optimal fit showed a threshold of 107 . 78±0 . 14 PFU/g is required for N1 module activation to occur , after which the SDE’s rate of activation ( a2 ) was 0 . 5±0 . 07 log10 ( PFU/g ) -1 with an intercept ( b2 ) of -3 . 8 ( unitless ) ( see the Methods and S2 Fig for additional details ) . Below this threshold , the model predicted minimal gene expression ( a1 = 0 . 17; b1 = 0 . 03 ) . The SLM goodness of fit on the training data was an adjusted R2 = 0 . 72 while an adjusted R2 = 0 . 41 was observed when the data was fit to a linear model . A Davie’s test confirmed that a segmented model was a significantly better fit than a linear correlation model ( P-value < 2 . 2e-16 ) . While the H1N1-infected lung tissue did not exceed an average peak virus titer of 107 . 4 PFU/g ( peak titer occurs at 48 hpi in Fig 4A ) , we observed increased transcriptional activity in H1N1-infected mouse lung tissues after this time point , suggesting either that the actual peak virus titer occurred between 48 hpi and the subsequent time point ( 60 hpi ) , or that the model-predicted threshold was slightly over-approximated . We next sought to validate the threshold model by attempting to predict cytokine-associated gene expression in influenza virus-infected lung when only the virus titers are known . For this , we selected the H5N1 virus , which has previously been associated with an excessive cytokine response [10] . We infected mice with 103 PFU of the H5N1 virus ( a 100-fold reduced dose compared with that used in the experiments to fit the model ) , determined lung virus titers at the same time points used for the initial experiment ( S3 Fig ) , and then evaluated the segmented linear model’s ability to predict cytokine-associated gene expression . First , we confirmed that the original transcripts assigned to the N1 module were again co-expressed , and thus we used the same transcripts originally assigned to the N1 module to determine the eigengene ( see S4 Fig for an analysis of the conservation of the N1 module between the two experiments ) . In this experiment , as expected , the peak average virus titer ( 109 . 3±0 . 21 PFU/g ) for the 103 PFU dose was delayed compared with that for the 105 PFU dose ( S3 Fig , compare to Fig 1 ) . Moreover , the SDE exhibited an 18-h delay in activation and a 42-h delay in peak expression compared with the 105 PFU N1 eigengene ( Fig 4D ) . Importantly , based on the virus titers alone , the fitted segmented linear model accurately predicted N1-like SDE behavior ( R2 = 0 . 71 for all time points and R2 = 0 . 87 for time points up to peak expression; Fig 4E and S5 Fig ) , and this could be further demonstrated at the individual gene level for specific N1-associated transcripts ( e . g . , Herc6 , Stat1 and Irf7; Fig 4F ) . These observations provide strong evidence that activation of inflammatory-associated gene expression is dictated by a specific virus concentration in infected tissue , and further suggest the novel possibility that the pulmonary innate inflammatory response has a nonlinear , ultrasensitive-like activation profile that promotes tolerance to low concentrations of virus . Although transcriptional activation of IFN-stimulated and inflammatory gene expression is a reasonable measure of the effects of inflammation response stimulation , we reasoned that a bona fide ultrasensitive mechanism that regulates this response should be reflected in other aspects of the associated signaling pathway ( s ) . Indeed , of the 17 cytokines associated with the N1 module , 15—including key inflammatory proteins , such as interleukin 6 ( IL-6 ) and monocyte chemoattractant protein-1 ( MCP-1 ) —were significantly correlated with the N1 module eigengene ( Pearson’s ρ≥0 . 5; FDR-adjusted P-value < 0 . 01; see S6 Table ) . In addition , when the protein expression levels of these 15 cytokines were plotted against the corresponding titer data , we observed dynamics similar to that of the inflammatory-associated N1 gene module . Initially , protein expression was low but strongly increased only after the virus titers exceeded the threshold of ~108 PFU/gram determined in the gene regulation model ( Fig 5A ) . In contrast , most of the protein levels of the other measured cytokines with transcripts that were not assigned to the N1 module did not show any obvious relationship to the proposed threshold response ( S6 Fig ) . The only major exceptions were LIF , RANTES , and IL18 ( S6 Fig ) . LIF’s gene transcript was not annotated on the arrays whereas IL18’s transcript was not identified as differentially expressed and therefore was not included in the clustering study . The RANTES transcript was included in the clustering study and assigned by the WGCNA algorithm to N2 although the transcript’s correlation to the N1 eigengene suggests it could also have been assigned to the N1 module ( Pearson correlation = 0 . 84 and 0 . 89 to the N1 and N2 eigengene , respectively ) . Some proteins appeared to conform to the threshold model in pH1N1-infected , but not H1N1- or H5N1-infected , mice . These may be cytokines that have strain-dependent responses and do not conform to the model . Overall , changes in N1-associated cytokine protein levels in influenza virus-infected mouse lungs were consistent with the proposed virus-titer regulated threshold-mechanism underlying the IFN-mediated response . We then searched for evidence of threshold-like behavior in the upstream signaling events leading to the activation of IFN-mediated , inflammation-associated gene expression: namely IFN-α/β protein expression , and IRF3 and STAT1 transcription factor phosphorylation ( Figs 5B , S7 and S8 . Images of representative blots are available in S9 Fig ) . Significant increases in the concentration of IFN-α and phosphorylated STAT1 ( pSTAT1 ) were detectable in infections with all three virus isolates and occurred at time points after the threshold level of virus was exceeded ( Fig 5B ) . For the H1N1 data , significant levels of pSTAT1 were observed at 36 hpi which—as noted previously—corresponds to the time immediately after the virus titers in H1N1-infected mice reached their peak ( see previous discussion ) . On the other hand , significant increases in phosphorylated IRF3 ( pIRF3 ) were observed only in pH1N1 and H5N1 infections , whereas significant increases in IFN-β were observed only in the H5N1 infection . Changes in the levels of IFN-β and pIRF3 occurred after significant increases in pSTAT1 and IFN-α occurred . As such , the change in the percentage of pSTAT1 was more closely correlated to the N1 eigengene than was that of pIRF3 ( correlation = 0 . 77±0 . 06 and 0 . 67±0 . 09 respectively ) , and significant increases in pSTAT1 corresponded to time points at which the mean virus titer exceeded the threshold level identified in the gene expression analysis ( ~107 . 78 PFU/g ) . The greater correlation of IFNα and STAT1 activation to the inflammation-associated , N1 module’s gene expression dynamics and the enrichment of the N1 module for the IRF7 binding sequence ( Table 1 ) suggest that the primary driver of the threshold-regulated , inflammatory gene response originates along the IRF7 → IFN-α → STAT1 axis ( Fig 5C ) .
Our data reveal that the activation of the IFN-associated inflammatory and antiviral response program in influenza-infected mouse lung is characterized by an ultrasensitive response driven by the virus load . The power of the threshold model is illustrated by its ability to accurately predict gene expression in infected mice , and the data further suggest that the molecular basis of threshold behavior originates upstream of IFN-α production . Threshold-like and ultrasensitive mechanisms are hypothesized to be necessary for effective management of critical cellular machinery in noisy environments , and are recognized players in the activation of the cell cycle [34] , mitogen-activated protein kinase signaling [35] , and apoptosis [33 , 36] . However , while a role for threshold behaviors have been postulated to be essential for filtering noise or errant signaling in complex biomolecular environments [35] , our study is the first to directly link threshold-like behavior to the virus-induced innate immune response . The ultrasensitive response observed in this study provides additional insight into the mechanisms that drive severe pathologies during influenza infection . Several works have suggested that viral load is a key determinant of pathology [12 , 15] while other works suggest that highly pathogenic influenza viruses induce early , strong inflammatory responses that are independent of the viral load [9 , 24 , 37] . Recently , it was observed that fatal influenza infections in mice coincide with a strong influx of neutrophils in what the author’s describe as a viral load-independent , “feedforward” inflammatory circuit [27] . The ultrasensitive response suggested by our study consolidates these hypotheses by suggesting that viral load drives cytokine production ( and in turn immune cell infiltration ) in a nonlinear manner which is capable of producing states of high and low innate immune responses . Characterization of key aspects of the inflammatory response , such as the onset and peak inflammatory gene expression , require a high temporal resolution of the virus growth and host response dynamics; an experimental design that was unique to our study . The ultrasensitive response model does not negate the significance of neutrophil infiltration [27] in determining fatal infections but suggests that viral load drives the high and low innate immune states . The observed threshold may represent the transition to immunopathology; as indicated by the histopathology results ( Fig 2 ) . Moreover , the influenza virus’ NS1 protein is crucial for inhibiting the interferon-mediated antiviral response [38] . The NS1 protein of three viruses used in this study have the SUMO1 acceptor site that indicates interferon antagonism capability [39–41] . Additional studies with NS1-mutated viruses and other pathogens may better reveal strain-dependencies for the observed thresholding behavior . The ultrasensitive response further suggests that the innate immune response has a limited capacity to respond to influenza virus infection and supports the development of immunomodulatory therapies . Interestingly , after the threshold was exceeded , the rate of activation for inflammatory and interferon-associated gene expression ( N1 ) was conserved for a moderately pathogenic and deadly viruses ( Fig 4A ) . The conserved rate of activation implies that the immune response detects the virus concentration but not the virus growth rate; suggesting the innate immune response is naturally limited in its ability to respond to high growth influenza viruses . Additionally , studies in knockout mice indicate that type I IFN-associated pathways are essential for protection during primary infection [42] and that earlier initiation of these pathways coincides with increased survival in mice infected with highly pathogenic isolates [43] . In combination with these studies , the findings here suggest a novel means of protecting high risk groups by treating them with compounds that target the molecular mechanisms responsible for the threshold behavior . Lowering the threshold required to induce the cytokine response may be a means of providing protection from severe influenza infection . Since these compounds would target host proteins , such treatments would be effective against various influenza virus strains . Data from the viruses studied here suggest that post-threshold , inflammatory gene expression primarily reflects interferon-regulated tissue damage , but time-course data from additional highly pathogenic viruses are needed to assess the degree to which interferon activity is associated with virus growth suppression .
The A/California/04/09 H1N1 virus ( pH1N1 ) was received from the Centers for Disease Control and Prevention . The A/Kawasaki/UTK-4/09 H1N1 virus ( H1N1 ) served as a reference for a seasonal influenza , whereas a fatal human isolate , A/Vietnam/1203/04 H5N1 virus ( H5N1 ) , served a highly pathogenic virus . All mouse experience were performed in accordance to the University of Tokyo's Regulations for Animal Care and Use . These regulations were approved by the Animal Experiment Committee of the Institute of Medical Science , the University of Tokyo ( approval number: PA10-13 ) . The committee acknowledged and deemed acceptable the legal and ethical responsibilities for the animals , as detailed in the Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education , Culture , Sports , Science and Technology , 2006 . All experiments with H5N1 viruses were performed in biosafety level 3 containment laboratories at the University of Tokyo , which are approved by the Ministry of Agriculture , Forestry , and Fisheries , Japan . Five-week old C57BL/6J , female mice were obtained from Japan SLC . For all experiments , mice were anesthetized with isoflurane and intranasally inoculated with either 103 or 105 PFU of virus . Initially , 42 mice were inoculated with 105 PFU of the H1N1 , pH1N1 , or H5N1 virus or mock-infected with PBS ( a total of 168 mice ) . At 14 time points , 3 mice per group were humanely sacrificed and their lungs harvested . The lungs were sectioned and used to assess virus titers ( right-upper lobe ) , cytokine levels ( right lower ) , and the initial gene expression that was used to train the segmented linear model ( left-lower ) . Separately , 42 mice were infected with 103 PFU of the H5N1 , sacrificed at the same 14 time points , and lungs sections obtained as described previously to provide the model validation data . The same inoculation method was used for all mice in this study . The numbers of mice used for flow cytometry , western blot , and interferon protein assay experiments are specific in the corresponding sections . For cytokine and chemokine measurements , mouse lungs were treated with the Bio-Plex Cell Lysis Kit ( Bio-Rad laboratories , Hercules , CA ) according to the manufacturer’s instructions . Concentrations of other cytokines were determined with the Bio-Plex Mouse Cytokine 23-Plex and 9-Plex panels ( Bio-Rad laboratories ) Array analysis was performed by using the Bio-Plex Protein Array system ( Bio-Rad laboratories ) . Virus titers were determined by plaque assay using MDCK cells . Mouse lung tissues were placed in RNALater ( Ambion , CA ) , an RNA stabilization reagent and stored at -80°C . All tissues were thawed together and homogenized for 2 minutes at 30 Hz using a TissueLyser ( Qiagen , Hilden , Germany ) as per the manufacturer’s instructions . From the homogenized lung tissues , total RNA was extracted with the RNeasy Mini Kit ( Qiagen , Hilden , Germany ) in accordance with the manufacturer’s instructions . Cy3-labeled cRNA preparations were hybridized onto Agilent-014868 Whole Mouse Genome 4x44K microarrays for 17 h at 65°C . Feature Extraction Software version 7 ( Agilent Technologies ) was used for image analysis and data extraction , and Takara Bio provided whole array quality control metrics . Differential expression was assessed by using a linear regression model . By using the limma package [44] 26 version 3 . 14 . 1 from BioConductor , probe intensities were background corrected by using the “norm-exponential” method , normalized between arrays ( using the quantile method ) , and averaged over unique probes IDs . Replicate quality was assessed using hierarchical clustering , resulting in the removal of a single array ( the array corresponded to a sample collected at 3 hpi in H5N1-infected mice . ) Probe intensities were fit to a linear model that compared data from infected samples to time-matched data collected from uninfected mice . Probes were annotated by matching to the probe names in the mgug4122a version 2 . 1 mouse annotation database available from BioConductor . All arrays in this study have been deposited on the GEO Expression Omnibus ( GSE63786 ) . Unsigned co-expression networks were constructed by using the blockwiseModules program from the WGCNA package version 1 . 23 . 1 [18] in R . The analysis was performed with several different parameterizations to ensure robust clustering . For the results reported in this text , we removed all probes that did not have a confident fold-change greater than 2 ( FDR < 0 . 01 ) for at least one infected-tissue to control-tissue , time-matched comparison . We then clustered the log2 of the normalized intensities for all 167 microarrays ( corresponding the three samples for each time-point for each infected or control population with the exception of data from H5N1-infected mice at 3 hpi which had two samples ) . Based on the scale-free topology characteristics curve , a power of n = 7 with no reassignment after clustering ( reassignThreshold = 0 ) and a maximum cluster size of 6000 probes was used . We generally observed that allowing gene reassignment between the modules led to poorer clustering based on the distribution the gene’s module memberships ( the correlation between a gene and the eigengene of the module to which it had been assigned . S10 Fig illustrates the distribution of the gene kMEs for modules N1 and N2 ) . We then repeated the clustering using different powers ( ranging from 7 to 11 ) , allowing different cluster sizes , different subsets of the expression data ( e . g . , clustering data from each infection separately or together ) , or relaxing the differential expression condition . In all clusterings performed , the two modules discussed in the text were identifiable . Fisher exact tests between each clustering run were used to determine whether the initial modules were significantly conserved under different parameterizations . We also considered if the N1 module would be isolated when using signed versus unsigned network construction . We constructed a signed co-expression network and found that 92% of the kME+ N1 genes are again clustered and confirmed that the gene expression dynamics were maintained ( see S11 Fig ) . ToppCluster [20] and DAVID [19] were used for gene ontology and pathway enrichment , and ToppCluster was also used for transcription factor binding site enrichment analyses . DAVID uses clusters of related annotations constructed from several annotation databases ( e . g . , pathway and gene ontology annotations ) to determine the function of a set of genes and scores the enrichment by averaging the unadjusted P-values ( determined by Fisher’s Exact test ) of the annotations within the cluster . ToppCluster uses hypergeometric tests to determine the enrichment between a set of genes and gene lists contained in 18 categories ( databases ) detailed in the ToppGene Suite [45] . The databases include cis-regulatory motif data [46 , 47] , referred to as transcription factor binding sites ( TFBS ) in the text . Both tools were used with their default settings and the gene universe was considered to be all annotated mouse genes . For each module , we considered the enrichment of all genes assigned to the module and the kME+ and kME- subsets . Generally , the enrichment analysis of the whole module gene set reiterated the enrichment results of the kME+ and kME- subsets albeit with slightly lower but still significant enrichment . Since both tools returned similar GO and pathway enrichment results , we summarized the functional and pathway enrichment results in Table 1 using the results from DAVID . The enrichment of interferon stimulated genes was determined by using a list of interferon stimulated genes from the Interferon Stimulated Gene Database [48] that was downloaded on May 9 , 2012 ( see S3 File ) . For each module , all module genes and the kME+ and kME- subsets were tested for enrichment using Fisher’s exact test in R . The p values were adjusted to control the false discovery rate . CTen [23] was used to determine enriched cell signatures in select co-expression modules . The enrichment score reported is the—log10 of the false discovery rate . Model fitting and validation was performed in R using the ‘segmented’ package [49] . Five mice per time point per infection were infected with 105 PFU of the described virus . Five uninfected ( naïve ) mice served as negative controls . Whole lungs were collected from mice , and incubated with Collagenase D ( Roche Diagnostics; final concentration: 2 μg/mL ) and DNase I ( Worthington; final concentration: 40 U/mL ) for 30 minutes at 37°C . Single-cell suspensions were obtained from lungs by grinding tissues through a nylon filter ( BD Biosciences ) . Red blood cells ( RBCs ) in a sample were lysed with RBC lysis buffer ( Sigma ) . Samples were resuspended with PBS containing 2 mM EDTA and 0 . 5% bovine serum albumin ( BSA ) , and cell number was determined by using a disposable cell counter ( OneCell ) . To block nonspecific binding of antibodies mediated by Fc receptors , cells were incubated with purified anti-mouse CD16/32 ( Fc Block , BD Biosciences ) . Cells were stained with appropriate combinations of fluorescent antibodies to analyze the population of each immune cell subset . The anti-F4/80 ( BM8; eBioscience ) antibodies were used . All samples were also incubated with 7-aminoactinomycin D ( Via-Probe , BD Biosciences ) for dead cell exclusion . Data from labeled cells were acquired on a FACSAria II ( BD Biosciences ) and analyzed with FlowJo software version 9 . 3 . 1 ( Tree Star ) . Three mice per time point per infection group were infected with 105 PFU of the described virus . The primary antibodies of mouse anti-STAT1 ( phospho Tyr701 ) mAb ( ab29045 , abcam ) , rabbit anti-IRF3 ( phospho Ser396 ) mAb ( 4947 , Cell Signaling ) , and mouse anti–β-actin ( A2228; Sigma-Aldrich ) were used; the secondary antibodies were HRP-conjugated anti-mouse IgG antibody ( GE Healthcare ) and HRP-conjugated anti-rabbit IgG antibody ( GE Healthcare ) . Mouse lungs were collected and homogenized with RIPA buffer ( Thermo Scientific , Rockford , IL , USA ) containing proteinase inhibitor ( Roche , Mannheim , Germany ) and phosphatase inhibitor cocktails ( Sigma-Aldrich , Saint Louis , Missouri , USA ) . The lysates were then briefly sonicated and centrifuged . Each sample was electrophoresed on sodium dodecylsulfate polyacrylamide gels ( Bio-Rad Laboratories , Hercules , CA , USA ) and transferred to a PVDF membrane ( Millipore , Billerica , MA , USA ) . The membranes were blocked with Blocking One ( Nacalai Tesque , Kyoto , Japan ) for 30 min at room temperature , and then were incubated with the primary antibodies overnight at 4° C , followed by the secondary antibodies . They were then washed 3 times with PBS plus Tween 20 ( PBS-T ) for 5 min and incubated with secondary HRP-conjugated antibodies ( as described above ) for 30 min at room temperature , followed by three washes with PBST . Specific proteins were detected by using SuperSignal West Femto Maximum Sensitivity Substrate ( Thermo Scientific , Rockford , IL , USA ) . Photography and quantification of band intensity were conducted with the VersaDoc Imaging System ( Bio-Rad Laboratories , Hercules , CA , USA ) . The quantity of target bands from each sample was standardized by their respective β-actin . Three mice per time point per infection group were infected with 105 PFU of the described virus . Half the lung of each mouse was dissolved in 1 mL of RIPA buffer . We measured the Interferon-alpha and Interferon-beta by using ELISA kits ( #12100 , #42400 , PBL Assay Science , NJ , USA ) according to the manufacturer’s instructions . Plates were read at an absorbance of 450 nm using a Versa Max plate reader ( MolecularDevices , Menlo Park , CA ) . Additional gene set overlap tests were performed in R with all of the genes annotated on the array as the reference ( background ) set . Statistical tests to compare means within the western blot , flow cytometry , immune cell count and protein assay data sets were performed in R using the ‘multcomp’ package [50] .
|
Vaccines suffice for protecting public health against seasonal influenza viruses , but when unexpected strains appear against which the vaccine does not confer protection , alternative treatments are necessary . In this work , we used gene expression and virus growth data from influenza-infected mice to determine how moderate and deadly influenza viruses may invoke unique inflammatory responses and the role these responses play in disease pathology . We found that the relationship between virus growth and the inflammatory response for all viruses tested can be characterized by ultrasensitive response in which the inflammatory response is gated until a threshold concentration of virus is exceeded in the lung after which strong inflammatory gene expression and cytokine production occurs . This finding challenges the notion that deadly influenza viruses invoke unique cytokine and inflammatory responses and provides additional evidence that pathology is regulated by virus load , albeit in a highly nonlinear fashion . These findings suggests immunomodulatory treatments could focus on altering inflammatory response dynamics to improve disease pathology .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
An Ultrasensitive Mechanism Regulates Influenza Virus-Induced Inflammation
|
Automatic responses enable us to react quickly and effortlessly , but they often need to be inhibited so that an alternative , voluntary action can take place . To investigate the brain mechanism of controlled behavior , we investigated a biologically-based network model of spiking neurons for inhibitory control . In contrast to a simple race between pro- versus anti-response , our model incorporates a sensorimotor remapping module , and an action-selection module endowed with a “Stop” process through tonic inhibition . Both are under the modulation of rule-dependent control . We tested the model by applying it to the well known antisaccade task in which one must suppress the urge to look toward a visual target that suddenly appears , and shift the gaze diametrically away from the target instead . We found that the two-stage competition is crucial for reproducing the complex behavior and neuronal activity observed in the antisaccade task across multiple brain regions . Notably , our model demonstrates two types of errors: fast and slow . Fast errors result from failing to inhibit the quick automatic responses and therefore exhibit very short response times . Slow errors , in contrast , are due to incorrect decisions in the remapping process and exhibit long response times comparable to those of correct antisaccade responses . The model thus reveals a circuit mechanism for the empirically observed slow errors and broad distributions of erroneous response times in antisaccade . Our work suggests that selecting between competing automatic and voluntary actions in behavioral control can be understood in terms of near-threshold decision-making , sharing a common recurrent ( attractor ) neural circuit mechanism with discrimination in perception .
A hallmark of behavioral flexibility is our ability , given the same sensory input , to resolve the conflict between an automatic response and a more appropriate volitional one [1–3] . This ability requires at least two executive processes . First , an automatic ( habitual or reflexive ) response needs to be withheld by top-down inhibitory control [2–11] . Second , a flexible mapping between the sensory input and motor response must be executed based on a rule signal [12–15] . In laboratory , competition between automatic and volitional responses is often investigated using the antisaccade task [16] , where a visual target suddenly appears in the periphery and one is required to make a saccade in the opposite direction from the target , rather than toward it as a prepotent reflex . Antisaccade has been used in clinical studies for testing the development of executive control or for probing its abnormalities associated with attention deficit hyperactivity disorder ( ADHD ) and other neurological and psychiatric disorders [17–22] . In addition to dorsal lateral prefrontal cortex ( DLPFC ) [23–26] , extensive studies have identified several other correlated brain regions , which include frontal eye field ( FEF ) , supplementary eye field ( SEF ) , superior colliculus ( SC ) , anterior cingulate cortex ( ACC ) , parietal cortex and basal ganglia ( see Munoz & Everling 2004 [16] and Pierrot-Deseilligny et al . 2005 [27] for review ) . These studies revealed a diversity in neuronal activities within or across brain regions during antisaccadic eye movement , suggesting that they are accomplished through complex interactions between multiple brain regions . Despite extensive studies , the neural circuit mechanism underlying antisaccade remains poorly understood . A number of computational models have been proposed to reproduce some of the basic features observed in antisaccade or similar anti-reach movements [28–34] . Because the neural processes required by these movements were thought to be comparable to those proposed for Stroop tasks , most models are conceptually similar to the classic model in Cohen et al . 1990 [1] ( see also Miller & Cohen 2001 [2] ) . These models typically assume a competition between a fast automatic response pathway and a slower voluntary response pathway , and a top-down control signal dictates which one wins the competition . An error ( a saccade toward rather than away from the target ) is attributed to a failure to suppress the automatic response , therefore is inevitably associated with a short reaction time . However , experiments with the antisaccade task have shown diversity in erroneous responses; there are fast errors associated with express saccade , but slow errors with reaction times comparable to those of correct antisaccade were also observed . In this work , we propose that competition takes place both at the level of interplay between automatic and voluntary responses , and within a neural circuit module for flexible sensorimotor remapping ( Fig 1A and 1B ) . We implemented this mechanism in a biophysically realistic model of spiking neurons , using a previously proposed decision-making circuit [35 , 36] and an inhibitory control circuit [37 , 38] . We show that the model reproduces salient neurophysiological observations of diverse neural activity patterns ( so far unaccounted for by previous simpler and more abstract models ) , as well as broad reaction time distributions for erroneous responses in antisaccade trials . This work revealed a new component of inhibitory control process ( responsible for slow behavioral errors ) which share a similar recurrent ( attractor ) neural circuit mechanism as that for near-threshold discrimination in perception [35 , 39] .
In the proposed model , an antisaccadic eye movement was produced through competitions between the automatic response driven by the visual input and the voluntary response generated internally by the top-down controls . The competition occurred not just in one , but in multiple regions in the brain ( Fig 1A and 1B ) . The action-selection module ( Fig 2A ) , which produced the neuronal command that drove the eye movement , received the target signal directly from the visual neurons ( Vis ) and the signal representing the voluntary action from the remapping module ( Fig 2B ) . The function of the remapping module is to convert the target signal into a desired saccade command based on the task instruction ( pro- or antisaccade ) . In a prosaccade trial , the remapping module was only weakly activated and the level of the gaze-holding control was low , so the target signal from the visual neurons Vis quickly triggered a prosaccade ( Fig 2C left ) . In an antisaccade trial , the top-down controls took the effect by suppressing the action-selection module temporarily and activating the saccade command conversion in the remapping module . The target signal from the visual neurons Vis could not trigger an erroneous prosaccade due to the temporary suppression of the action-selection module . In the mean time the remapping module converted the target signal into a saccade command toward the end point that is opposite to the target . The remapping module took time to generate a decision signal and sent it to the action-selection module ( Fig 2C right ) . By the time the suppression of the action-selection module from the top-down control was removed , the antisaccade signal was able to drive a correct antisaccade response . The action-selection module chose between the signals that represented different actions and generated a corresponding motor command . We hypothesized that the function of action-selection module is performed , in part , in SC and FEF . Based on observations from various experiments on SC and FEF , we required the action-selection module to exhibit several basic functions: 1 ) The module needs to have two separate populations of saccade neurons . Each population triggers saccade eye movements to a different direction , i . e . right vs . left . A saccadic eye movement is defined as the population firing rate of one of the neural populations reaches a preset ( 100Hz ) firing rate before the other does . 2 ) There is a strong competition between the two populations of saccade neurons , i . e . when one population is strongly activated , the other has to be suppressed . 3 ) Both populations of saccade neurons can be suppressed by fixation neurons when a gaze holding is required . To accommodate these functions , we created an action-selection module which consisted of the following neural populations ( Fig 2A ) : 1 ) The excitatory populations SacL and SacR which signalled saccades to the left and right , respectively . 2 ) The inhibitory population I0 that provided mutual inhibition between SacL and SacR . The circuit of SacL , SacR and I0 was built based on a similar circuit design with that used in the decision layer [35] . The two saccade populations exhibited a winner-take-all competition and only one saccade population could be activated in each trial . 3 ) The excitatory populations FNL and FNR which contained fixation neurons for SacL and SacR , respectively . 4 ) The inhibitory populations I1 -I4 which provided feedforward inhibition between the saccade neurons ( SacL and SacR ) and fixation neurons ( FNL and FNR ) . The circuit formed by SacL , SacR , FNL , FNR and I1-I4 was built based on the neural interactions reported in Munoz & Istvan ( 1998 ) [40] . Fixation neurons were driven by the fixation signal and gaze-holding signal from the top-down gaze-holding control . Each of the saccade neural populations ( SacL & SacR ) was further divided into two sub-populations , BN and BUN , which were used to mimic the observed burst neurons and build-up neurons , respectively [41] . The difference between BN and BUN was that BN neurons received strong feedback excitation by forming strong recurrent connections with all neurons in the Sac population while BUN received weaker feedback excitation by forming recurrent connections only with BN neurons ( Table 1 ) . When the input was weak , BUN neurons displayed a graded activity that resembled the observed build-up activity . When the input was large enough , BN neurons became activated and exhibited an all-or-none burst activity due to the strong recurrent excitation . Because there was mutual excitation between the two sub-populations , the burst activity in BN also strongly activated BUN neurons . As a result , both BUN and BN neurons exhibited strong burst activity before saccades as observed . In the paper , when we display the activity of the saccade neurons Sac , they are always represented by BUN neurons . Note that in the model BN neurons sent an efferent copy to neurons in the decision layer . The purpose of adding this projection was to shut down the neurons in the decision layer after an motor action was executed in order to generate a more realistic appearance of the neuronal activity in the decision layer . The projection did not affect the behavior performance or the conclusion of the present study . The remapping module consisted of two , visual and decision , layers ( Fig 2B ) . The visual layer responded to the visual target and projected to the downstream decision layer . Empirical data suggest that the function of remapping module is , in part , performed in SEF and LIP [42 , 46 , 73–75] . We assumed that the projection from the remapping to the decision layers consisted of all possible visuomotor mappings that were capable of converting the target signal into a saccade command toward any end point . In the study we only modelled the two most relevant mappings: ( i ) the default direct map that directly transfers the target signal to a saccade command toward the target and ( ii ) the inverted map that converts the target signal into a saccade command toward the end point opposite to the target . The direct map was the strongest pathway by default , but the subject could learn to suppress it and to facilitate other maps based on the task instruction . The decision layer received inputs from these maps and made a probability decision on the stronger one through neural competition . The design of the decision layer followed that of the attractor neural network model of perceptual decision [35 , 36] . The layer consisted of a large population of excitatory neurons and a population of inhibitory interneurons . Among the excitatory neurons , two sub-populations of neurons , DecL and DecR , were selected to receive input from the visual layer and project to SacL and SacR in the action-selection module , respectively . DecL and DecR competed with each other through the inhibitory interneurons . Neurons are heavily connected to each other and form strong feedback excitations within each decision population ( DecL or DecR ) . There are weak excitatory connections between the two decision populations and the net interactions between them are inhibition due to the strong feedforward inhibition provided by the inhibitory interneurons ( population I ) . The described circuit is consistent with the general organization of the cortical circuits in which the connection probability between pyramidal neurons decreases with the distance and the inhibitory interneurons provide feedback and feedforward inhibition to the pyramidal neurons . The neurons in the decision layer exhibited reverberatory excitation within each selective pool and winner-take-all competition between selective pools , which was essential in producing the observed longer mean reaction time and higher error rate in antisaccade trials than in prosaccade trials . The probability of making either choice and the response time depended on the levels of the inputs to DecL and DecR ( Fig 3 ) . Therefore , we first needed to individually decide the proper levels of inputs to the decision layer for prosaccades and antisaccades . The decision layer took inputs from Dir neurons in the direct map and Inv neurons in the inverted map . The neurons in both maps received the same visual signal but different baseline input , which was modulated by the top-down remapping control ( see Materials and Methods ) . We tested the performance and the mean reaction time of the decision layer under various combinations of input levels by adjusting the background inputs to Dir and Inv neurons ( Fig 3 ) . Each combination of input was tested and averaged over 1000 trials . The optimal combination we found for prosaccade determined the default background input in the absence of the top-down control while the combination we found for antisaccade determined the levels of the background input , or kDir and kInv , under the modulation of the top-down control . We illustrate how the neural network model performs prosaccade and antisaccade using two example trials ( Fig 2C ) . We first assume that the visual target appears on the left of the screen in both trials . In a prosaccade trial , the visual target activates left visual neurons in VisL , DirL and InvL . The activated VisL strongly excites SacL neurons . In the remapping module , DirL develops a stronger response than InvL does by default due to the stronger background excitation in Dir neurons . As a result , DecL wins the competition against DecR easily and sends the signal to SacL . A leftward prosaccade is triggered by either the early signal from VisL or by the late signal from DecL . In contrast , in an antisaccade trial , a subject suppresses the tendency to make a prosaccade by applying a strong top-down control which is assumed to be carried out , in part , by DLPFC ( see Discussion ) . The top-down control causes two effects: 1 ) temporal suppression of SacL and SacR through the fixation neurons and 2 ) bias in the remapping module due to the facilitated inverted map and the suppressed direct map . See Materials and Methods for more detail . The left visual target , although strongly activates VisL as in the prosaccade trial , cannot activate SacL due to the suppression from the top-down holding control . In the remapping module , InvL sends a stronger signal to the decision layer than DirL does . In consequence , DecR wins the competition with probability higher than DecL does and triggers a rightward antisaccade by activating SacR . Note that although the activity of InvL is stronger than that of DirL , the difference in the magnitudes between the two populations is smaller than that in prosaccade trials . Therefore in antisaccade trials the activity of DecR ramps up slowly with a probability much less than 1 . This produces a slower mean response time and a higher error rate in antisaccade than in prosaccade . Next , we performed full model simulations and examined how the circuit works to produce prosaccades , antisaccades and erroneous responses . In a prosaccade trial ( Fig 4A ) , due to the weak level of the gaze-holding control , the activity of saccade neurons ( SacL ) in the action-selection module rose quickly in response to the visual input from the left target . At the same time , the left visual input strongly activated the direct pathway ( DirL to DecL ) which in turn projected to the same saccade neurons . As a result , the saccade neurons in SacL received strong input and quickly triggered a prosaccade . In an antisaccade trial ( Fig 4B ) , the strong level of the gaze-holding control temporally suppressed the responses of the saccade neurons SacL & SacR to any input including the visual input from the left target . In the meaning time , the remapping control activated InvL neurons while suppressed DirL neurons . As a result , the downstream decision neurons received a stronger input in DecR than in DecL . The neurons in DecR won the competition against DecL and activated the downstream saccade neurons ( SacR ) which triggered an antisaccade to the right . Based on the simulated neural activities , we suggest that the SEF neurons that involve in the direct or inverted pathways can be identified by comparing their related strength of visual responses during prosaccade and antisaccade trails ( as depicted in Fig 4A and 4B ) When modelling a cognitive function , it is insightful to study how the nervous system makes errors in addition to how it makes correct responses . Indeed , we found that the proposed model makes two types of erroneous responses in antisaccade: “slow error” and “fast error” ( Fig 4C and 4D ) . In some antisaccade trials , the system exhibited a weak gaze-holding control which could not efficiently suppress the response of the saccade neurons ( SacL ) to the left target . Therefore , a quick prosaccade was triggered even before the upstream decision layer ( DecR and DecL ) reached any decision ( Fig 4C ) . In other antisaccade trials , due to the stochastic nature of the neural competition , activity of DecL neurons ramped up against DecR and an incorrect decision was made . In this case , even if the system exhibited a strong gaze-holding control and successfully suppressed the response of SacL neurons to the onset of the left target , the wrong decision made in the upstream decision layer could still produce an erroneous response . The reaction times in this type of error trials were comparable to those of correct antisaccade trials because in both case the system went through the neural competition in the decision layer ( Fig 4D ) . Since the model we used for the remapping module was originally used for near-threshold perceptual discrimination [35] , this result suggests that a common mechanism for near-threshold decision-making in perceptual discrimination and conflict resolution between competing automatic and voluntary actions in behavioral control . The model further suggests that erroneous saccades with slow or fast reaction times are due to different mechanisms as revealed by the activity of neurons in the decision layer ( Fig 4C and 4D ) , which presumably corresponds to the movement neurons in SEF . We next verified that in the model the level of gaze-holding control indeed affected the types of error the system made . We first defined the “fast error” as the erroneous saccade with a reaction time shorter than that of all correct antisaccades ( 178 ms in the case of the gap paradigm demonstrated here ) while the “slow error” are the remaining erroneous saccades with longer reaction times . We found that the trials with fast errors tended to have a weaker gaze-holding control ( Fig 5A ) , whereas slow errors had comparable levels of gaze-holding control with that of correct antisaccade trials , indicating that the slow errors were produced due to a mechanism ( neural competition in the upstream decision layer ) other than a weak gaze-holding control . The low level of gaze-holding control results in a high level of the saccade neuron firing rate during the gap period ( Fig 5B ) . In consequence , an erroneous saccade is more likely to be triggered immediately after the target onset . Therefore , the model predicts a negative correlation between the level of top-down control and the saccade neuron activity during the gap period for fast errors in antisaccade trials . The model qualitatively produced diverse neuronal activities observed in various brain regions [41–44] . We first examine the model-produced neuronal activities here and compare the model with observations in a later section . Neurons in the visual layer exhibited quick firing activity following target onset and we found that these neurons gave rise to stronger responses to the visual input in prosaccade trials than in antisaccade trials if we recorded from the direct map ( Dir ) ( Fig 6A ) . In contrast , neurons in the same layer exhibited stronger activity in antisaccade trials than in prosaccade trials if recorded from the inverted map ( Inv ) ( Fig 6B ) . This activity is consistent with several empirical studies in which some visual neurons in SEF were found to exhibit a stronger response to the visual target in antisaccade than in prosaccade , while a few other visual neurons in SEF exhibited an opposite trend [42 , 46] ( S1 Fig ) . Neurons in the decision layer can be viewed as movement neurons because they develop strong activity toward the onset of the motor responses . Interestingly , if we recorded neurons from the side that corresponded to the correct response ( DecR for antisaccade and DecL for prosaccade ) , correct antisaccade produced stronger responses than correct prosaccades did ( Fig 6C and 6D ) . This is because in prosaccade trials the direct visual activation from VisL neurons to SacL neurons also contributed to the generation of saccades . Therefore a prosaccade could often be triggered when the activity of neurons in DecL was still weak . Interestingly , it was reported that most movement neurons in SEF exhibited stronger activity in antisaccade than in prosaccade [42 , 46] . This is consistent with the response of the neurons in the decision layer we report here ( S2 Fig ) . In the action-selection module , we observed that the saccade neurons ( Sac ) developed two waves of activities in antisaccade trials . The neurons ( SacL ) that received the visual input exhibited a moderate response immediately following the target onset but became suppressed due to the inhibition from the gaze-holding control . On the other hand , the neurons on the opposite side ( SacR ) started to develop a strong activity upon the arrival of the antisaccade signal from the upstream decision layer ( Fig 6E ) . Such two waves of activity could be identified in several empirical observations of neurons in FEF or SC in which neurons with the target in their response field showed a transient but strong response with an early onset , while a slightly weaker activity with a later onset was developed in neurons in the opposite side when a correct antisaccade was initiated [43 , 44 , 47 , 48] ( S3 Fig ) . However , in the model the early response was weaker than the latter response . Interestingly , in a study using visual search task combined with prosaccade and antisaccade [44] , two types of neurons were observed in FEF . Type I neurons responded to the target ( singleton ) and the saccade endpoint with a “two-wave” activity in antisaccade-like trials . This type of neurons are similar to the build-up neurons ( BUN ) in the Sac populations in our model . Type II neurons only responded to the saccade endpoint with a monotonic ramping activity which resembled the activity of neurons in the decision layer in our model . Therefore , the mappings between the model components and the brain regions may not be one-to-one but multiple-to-multiple , dependent on the specific task being investigated . Futhermore , by comparing saccade neurons that triggered correct prosaccade ( SacL ) and correct antisaccade ( SacR ) , we found that prosaccades produced stronger responses than antisaccades did before the target onset ( Fig 6F ) . This trend is consistent with several earlier studies which reported that saccade neurons in FEF and SC exhibited stronger population activity in prosaccade than in antisaccade before the saccade onset [41 , 43] ( S3 Fig ) . However , around and after the saccade onset , observed SC neurons still displayed stronger prosaccade than antisaccade activity . The observation was not reproduced in our model . Next , we examined the reaction time distributions produced by the model in the gap , no-gap and overlap paradigms ( Fig 7 ) . The mean reaction time of correct antisaccade was longer than that of correct prosaccade in all three paradigms . The mean reaction time of failed antisaccade trials ( erroneous prosaccade ) were shorter than that of correct ones . Notably , reaction times of the erroneous antisaccade covered a broad range with some erroneous responses as fast as the fastest prosaccades and others as slow as the slowest antisaccades . The Gap paradigm produced the highest percentage of fast errors in antisaccade among the three paradigm while the Overlap produced the least . Furthermore , the reaction times of prosaccade and erroneous antisaccade also exhibited bimodal distributions . This bimodal feature was most significantly in the gap paradigm , becomes less apparently in no-gap paradigm and disappeared in overlap paradigm . All of these features were observed in earlier empirical studies of prosaccade and antisaccade using the Gap and Overlap paradigms [41 , 45] ( S4 Fig ) . However , the detailed shape of reaction time distribution is strongly subject-dependent and the bimodality may not be observed in all subjects . It is interesting to see how the model predicts when the levels of the remapping control and/or the holding control change . To this end , we performed simulations under the following four conditions of control levels: 1 ) strong holding / strong remapping , 2 ) strong holding / weak remapping , 3 ) weak holding / strong remapping and 4 ) weak holding / weak remapping . The strong and weak remapping controls were simulated using levels that are 20% more or 20% less than the normal level used in Fig 7B , respectively . The strong gaze-holding control was also modelled as 20% more than the normal level while the weak gaze-holding control is simply modelled as the level used in prosaccade trials . We examined the reaction time distribution and the error rate of antisaccades and found that the level of the remapping control mainly affected the percentage correct and the mean reaction time of correct antisaccade ( Fig 8 ) . A strong or a weak remapping control led to 90% or 60% of correct antisaccade trials in the case of strong gaze-holding control . The mean reaction time of correct antisaccades was ∼ 225 ms for the strong remapping control and ∼ 264 ms for the weak remapping control . On the other hand , the level of the gaze-holding control significantly affected the percentage correct but not the mean response time . The relative numbers of fast errors versus slow errors were affected by both control types , but in opposite ways ( Fig 8 ) . A strong gaze-holding control resulted in lower ratio of fast errors because of the suppression of automatic responses while a strong remapping control increased the ratio of the fast errors due to the fewer slow errors made by the decision layer . This model predicited association between the ratio of fast/slow errors and the deficit of top-down control may be used for clinical assessment ( see Discussion ) .
In the present study we constructed a spiking neural network model which demonstrates how the brain selects between competing automatic and voluntary responses in the antisaccade task . Our model distinguishes previous models [28–33] in three basic respects . First , a near-threshold decision circuit makes a decision on the saccade direction , whereas the motor output is determined in a separate , action-selection module . Second , competition between the automatic response and voluntary response occur in two stages in more than one brain regions through different neural pathways . Third , top-down control has two components; the stop , or action-holding , control does not race against the go process , but exhibits proactive inhibition through tonic activation of fixation neurons . Each of the three assumptions play different roles in reproducing the diverse behavioral and neuronal activity observed in the antisaccade task . The decision circuit reproduces the basic behavioral trait of antisaccade , which is characterized by the slow and less accurate responses . The two-stage competition further produces express saccades in prosaccade trials , fast errors in antisaccade trials and diverse neuronal activity observed in different brain regions ( see the detailed discussion below ) . The two-component top-down control provides predictions of the model by showing that the manipulation of the individual components can change the error rate and the ratio between the fast and slow errors in the antisaccade trials ( see the detailed discussion below ) . The two-stage competition scheme is essential in reproducing behavioral and neuronal observations in the antisaccade task . In the scheme ( 1 ) The target signal coming down the direct visual input pathway competes with the voluntary signal under the modulation of gaze-holding control and ( 2 ) the target signal coming down the remapping pathway competes with the inverted saccade signal ( produced by the remapping top-down control ) in the decision layer with a mechanism resembles near-threshold decision-making . A successful antisaccade requires that the voluntary response ( driven by the top-down controls ) wins both competitions . On the other hand , erroneous prosaccades made in antisaccade trials can be caused by two distinct mechanisms . Errors with fast response times are produced due to a weak gaze-holding control . In this scenario , the direct visual stimulus from the target activates the saccade neurons in the action-selection module before the remapping module is able to generate the signal that inverts the saccade direction . Errors with slower response times are produced due to the failure of the remapping module in inverting the saccade direction even when the gaze-holding control successfully suppresses the response of the action-selection module to the direct visual input from the target . This approach utilizes neural competition in the attractor network ( implemented in the remapping module ) in which a smaller difference in the inputs to the two competing neural pools results in a smaller percentage correct and a longer mean reaction . It has been shown that the erroneous responses in such a network have a similar mean reaction time to that of correct responses [35] . Hence , it is suitable to account for slow errors with long reaction times comparable to those of correct antisaccade . It remains to see whether this model conclusion is generally valid , for instance by examining in future work whether slow errors are present in Stroop task [1 , 49 , 50] and other behavioral paradigms that engage inhibitory control . The two competition pathways also explain why in the superior colliculus ( and in the frontal eye field ) the prosaccade responses are stronger than antisaccade responses while in the supplementary eye field the neural responses are diverse ( some are stronger in prosaccades and others are stronger in antisaccades ) and the diversity distributed differently across visual and movement neurons . Our modelling result implies that when using the antisaccade task as a diagnostic tool , in addition to measuring the reaction times and the percentage of correct antisaccades , assessing the erroneous responses including the ratio between the fast and slow errors may also provide valuable information . For example , when a subject has an moderate impairment in the holding control , it does not change the mean reaction time for the correct antisaccade and only slightly reduces the percentage correct . However , if we measure the percentage of the fast errors , the impact of the impaired holding control becomes very significant ( Fig 8 ) . A new insight from the model is that the top-down control consists of two components: remapping and action holding controls . The holding control exhibits an inhibitory effect on the action selection module . This inhibition is thought to be exhibited by the prefrontal cortex [4–9] . However , several recently studies on DLPFC deactivation demonstrated more complex effects , including excitatory influence of DLPFC on the antisaccade tasks [11 , 48] . The observations implied that , if there is inhibitory top-down control involved in the antisaccade task , it may be carried out by brain regions other than DLFPC . In the proposed model , the gaze holding and the remapping controls exhibit inhibitory and excitatory effects , respectively , over the action-selection module . Therefore , these two controls may not originate from a single prefrontal region , but rather from a distributed network across multiple brain regions that extend beyond DLPFC [10] . A recently large-scale neural network model with a similar design of multiple control mechanisms has been independently developed and shown to be able to simulate various inhibitory control tasks including antisaccade [31] . The model consists of two main control pathways: 1 ) from the conflict detecting anterior cingulate cortex to the hyperdirect pathway in basal ganglia and 2 ) from the rule-dependent dorsolateral prefrontal cortex ( DLPFC ) to the indirect pathway in basal ganglia and to frontal eye fields . Although the first mechanism is functionally similar to our top-down inhibition on the action selection module , the latter one is different from ours . In our model , we have a top-down re-mapping signal that facilitates the inverted pathway in the antisaccade trials . As a result , the competition ( or the “conflict” ) between the direct and inverted pathways increases and the resulting mean reaction time in the antisaccade trials is slower than that of prosaccade trials . In Wiecki & Frank 2013 , the larger antisaccade response times are mainly due to the slow responses of DLPFC because of the larger membrane time constant but not the result of the stronger competition . Unlike in the present model which produces two types of errors , in their model the error in antisaccade is mainly due to failing to suppress prosaccades and therefore the response times are faster than those of the correct antisaccade trials . The present work focuses on neural processes that resolve conflict in a trial and that lead to diverse erroneous responses in antisaccade . We do not model the slow neural processes that exhibit the influence across trials . For example , erroneous antisaccades may be frequently made in prosaccade trials when preceded by a block of consecutive antisaccade trials [51] . This type of trial-history effect can be reproduced in our model by introducing a slow and history-dependent component in the remapping control ( as the task-set inertia ) and/or in the gaze-holding control ( as the persistent response-system inhibition ) [52] . We also do not model trial-to-trial adaptation [53] which requires conflict monitoring in ACC [54] . It is interesting to incorporate a history-dependent modulation in the control system and/or an ACC module with a learning process in an extension of our model in the future . Another interesting open question for future research is the rule representation itself . Although there are some physiological studies of it [55] , we are still not clear about the rule circuit ( presumably in the prefrontal cortex ) and how it actually implements the control signal . One may argue that the spiking neuron model is not necessary for our circuit model because the top-down control and the competition pathways mainly work at the circuit level , at which the detailed neuron dynamics does not seem to be important . However , our neuron model is not just “spiking” , but also endowed with conductance-based synapses which captures temporal dynamics of synapses . In particular , the slow reverberatory excitation mediated by the NMDA receptors is crucial for the winner-take-all decision circuit [35] used in the proposed model . Indeed , recent studies have demonstrated the rule of NMDA receptors in antisaccade [56 , 57] . Our biologically-based model allows us to further investigate how manipulation of specific receptors may influence the performance of antisaccade in future studies . In the present study we do not explicitly include basal ganglia in the model . Basal ganglia controls eye movements by inhibiting or disinhibiting neurons in SC and hence may participate in the antisaccadic eye movement [58–62] ( however , see Condy et al . 2004 [63] ) . Indeed , in Wiecki & Frank model [31] , the authors included basal ganglia which plays a central role in action control . So , why can our model , without including the basal ganglia , still reproduce a broad range of behavioral and neuronal activities that observed empirically ? From the functional point of view , the inhibitory control in the basal ganglia is replaced by the holding control and fixation neurons in the action selection module in our model . Furthermore , comparing to the fixation neurons which perform functions specifically for gaze holding , the inhibitory pathways ( indirect and hyperdirect ) in basal ganglia is multi-functional and is more associated with higher brain functions including inhibition of planned responses , enhancement of action precision , avoidance of aversive stimuli , etc [58 , 59 , 64–66] . Although a number of studies discovered differential responses of striatal neurons to prosaccade and antisaccade [60 , 61] , a brain lesion study did not found a significant impact of the basal ganglia lesion to the performance of antisaccade [63] . Moreover , studies on psychiatric disorders that associated with basal ganglia abnormality , e . g . Parkinson’s and Huntington’s diseases , found mixed results in their effects on antisaccade ( see [17] for an extensive review ) . In addition to basal ganglia , other brain regions , such as inferior parietal lobule , middle occipital gyrus and cuneus , have also been suggested to be involved in antisaccade [67] . Further studies on precise lesions or deactivation of related brain regions are necessary to reveal actual contributions of these regions to the performance of antisaccade . Our model can be viewed as providing a minimum or a principle circuit that includes only major functional modules necessary for performing antisaccade . Some of the elements in the model may be actually implemented in multiple locations in the brain as redundancy . In conclusion , the proposed model suggests a two-stage competition mechanism for how the brain selects between the automatic and voluntary responses in the antisaccade task . In this mechanism , the voluntary responses driven by top-down control competes against the automatic responses in two different pathways . Failing in either of the pathways results in erroneous responses . The proposed model is able to reproduce a wide range of neuronal and behavioral features observed in various studies and provides insights into how competing responses are selected at the neural circuit level and why the subjects make errors . This work demonstrates that a common mechanism , with a combination of NMDA receptor mediated slow reverberatory excitation and winner-take-all-competition , is at the core of both near-threshold perceptual decision-making and behavioral control .
We developed a model for resolving a conflict between automatic and voluntary responses . For the sake of concretness , we tested the model by simulating a pro- versus anti-saccade task ( Fig 1C ) . In the beginning of a trial , the subject fixated at the spot , or the fixation signal , located at the center of the screen . The color of the fixation signal served as a cue indicating the type of the response ( green for prosaccade and red for antisaccade ) . After a delay , a target appeared on the either side of the screen . In prosaccade trials , the subject had to make a saccade into the target as soon as it appears . In antisaccade trials , the subject had to make a saccade to the side that opposites to the target . In the present study we tested the model using three different paradigms of antisaccade: Overlap , Gap and NoGap ( Fig 1D ) . This three paradigms were different in the timing of the fixation signal offset . In Gap task , the fixation signal turned off 200 ms before the onset of the target . In Overlap task , the fixation signal was not turned off during the entire course of the trial . In the NoGap task , the fixation signal was turned off at the same time with the target onset . In the simulations the stimuli including the fixation signal and the target signal were modelled as excitatory inputs to the fixation neurons and visual neurons , respectively . The cue determines the levels of the top-down controls in the model . See the sections “The network model” and “The visual stimuli and top-down controls” below for details . A saccadic eye movement was triggered in the simulation if the population firing rate of one of two ( right and left ) burst-neuron populations in the action-selection module exceeded 100 Hz . The population firing rate was calculated using a 20 ms sliding time window . The reaction time was defined as the interval between the onset of the target and the saccade . Single neuron and synapse models followed those used in previous studies [35 , 36] . Briefly , each neuron in the circuit model was simulated using the leaky integrate-and-fire model with conductance-based synapses . The membrane potential V ( t ) for each neuron obeys the following equation: C m d V ( t ) d t = - g L ( V ( t ) - V L ) - I s y n ( t ) , where Cm is the membrane capacitance , gL is the leak conductance , VL is the resting potential and Isyn is the total synaptic current . When the membrane potential V ( t ) of each neuron reaches a threshold Vthreshold = −50 mV , a spike is emitted and V ( t ) is set to the reset potential Vreset = −55 mV for a refractory period Tr = 2 ms . For inhibitory neurons , we used the following parameters: Cm = 0 . 2 nF , gL = 20 nS and VL = −70 mV . For excitatory neurons , we used Cm = 0 . 5 nF , gL = 25 nS and VL = −70 mV . The synaptic current Isyn ( t ) includes external inputs from the outside of the modelled circuit ( stimuli , top-down controls etc ) , background noise and internal input from other neurons in the modelled circuit . We modelled three types of synaptic receptors: AMPA , NMDA and GABAA . They are described by: I s y n ( t ) = g AMPA s AMPA ( t ) ( V ( t ) - V E ) + g NMDA s NMDA ( t ) ( V ( t ) - V E ) 1 + [ M g 2 + ] e - 0 . 062 V ( t ) / 3 . 57 + g GABA s GABA ( t ) ( V ( t ) - V I ) , where VE ( = 0 ) and VI ( = −70 mV ) are the reversal potentials , [Mg2+] ( = 1 . 0 mM ) is the extracellular magnesium concentration , g is the synaptic efficacy and s is the gating variable . Subscripts in g and s denote the receptor type . The gating variable obeys d s ( t ) d t = ∑ k δ ( t - t k ) - s τ for AMPA and GABAA receptors and d s ( t ) d t = α ( 1 - s ( t ) ) ∑ k δ ( t - t k ) - s τ for NMDA receptors with α = 0 . 63 . The decay constant τ was 2 ms , 5 ms and 100 ms for AMPA , GABAA and NMDA receptors , respectively; δ ( t − tk ) is the delta function and tk is the time of the kth presynaptic spike . We implemented short-term facilitation ( STF ) in several synaptic connections in the action-selection module ( indicated by * in Table 1 ) . The gating variable s was multiplied by the STF factor F , which obeys the following dynamics [68]: d F d t = α F ( 1 - F ) ∑ k δ ( t - t k ) - F / τ F , where the dimensionless factor αF was 0 . 15 and the decay constant τF was 1000 ms . Several brain regions have been found to play roles in antisaccadic eye movement . They include DLPFC [23–25 , 69 , 70] , FEF [26 , 43 , 44 , 47 , 71] , SEF [24 , 42 , 46 , 71] , SC [11 , 41 , 48] , ACC [24 , 71 , 72] , parietal cortex [67 , 71 , 73–75] and basal ganglia [59–62] ( but see [63] ) . Some of the brain regions , such as frontal eye field and superior colliculus , exhibit similar neuronal responses during the antisaccade task , possibly reflecting a redundancy of neural representations . Based on this consideration , our goal is not to simulate activities in every correlated brain region , but to model the core neural processes that is sufficient to reproduce the diverse neuronal and behavioral responses . The network model consisted of two major modules: the action-selection module ( Fig 2A ) and the remapping module ( Fig 2B ) . The visual target signal reached the action-selection module which produced saccadic eye movements through two pathways: 1 ) a direct projection from the visual neurons Vis to the action-selection module and 2 ) a remapping pathway that went through the remapping module before reaching the action-selection module . The function of the remapping module was to map the visual signal input to the desired saccade direction , based on a previously proposed attractor network model of perceptual decision [35] , and produces neural commends that drove the downstream action-selection module . The model was endowed with a cue-dependent top-down control component which modulated the action selection and the remapping modules in order to produce prosaccade or antisaccade . The functions of each module and their relationship are illustrated and analysed in Results . The parameters of synaptic connections of each neural populations are listed in Tables 1–3 . All synaptic connections in the model were all-to-all , i . e . every neuron in the source population connected to every neuron in the target population . The model was driven by four types of inputs: background noise , target signal , fixation signal and top-down controls , which were all modelled as synaptic input with Poisson statistics to related neural populations through AMPA mediated currents . The level of background noise and the target signal are the same in all trial types . The fixation signal is elicited by the fixation point , hence is different between the Overlap , NoGap and Gap trials . The top-down controls , which consist of the gaze-holding control and the sensorimotor remapping control , are used to initiate an antisaccade . The background noise was used to maintain each neuron at a desired baseline activity . The level of the background noise input to each neural population is listed in Table 4 . The target signal was sent to the visual neurons ( Vis ) in the action-selection module ( Fig 2A ) and to the visual neurons ( Dir and Inv ) in the remapping module ( Fig 2B ) . To simulate the strong visual responses and the quick adaptation ( at the hundred-millisecond time scale ) commonly observed in many visually responded neurons , we modelled the firing rate of the target signal as f ( t ) = ( f max - f min ) exp ( - t / τ ) + f min , where t = 0 corresponds to the onset of the target signal . At t = 0 , the input firing rate jumps to fmax but decays exponentially to fmin . In the study we set fmax = 28 , 000 Hz and fmin is 38 . 8% of the peak value fmax . The decay time constant τ was 100 ms . The input conductance of the target signal was 0 . 3 nS for all populations ( Vis , Dir and Inv ) that receive the target signal input . Note that here and in the following , the input firing rate represents the total rate from a large number of upstream neurons . The fixation signal drove fixation neurons in the action-selection module . The input firing rate was set to a constant value of 320 Hz with a synaptic conductance 2 . 0 nS . The fixation signal is elicited by the fixation point and is therefore turned off at the same time with the fixation point offset . The top-down control depended on the rule signal ( Crule ) and consisted of two components: a gaze-holding control Ch and a sensorimotor remapping control Crem . When an antisaccade was required , Crule = 1 . Otherwise , Crule = 0 . In the beginning of each simulated trial , the rule signal Crule is set to 0 or 1 according to the cue ( the color of the fixation signal ) . Note that in the present study we focus on how conflict responses are resolved by the neural competition in multiple brain regions . The model was developed based on the assumption that the subjects already learned the association between the cue ( the color of the fixation signal ) and the task ( prosaccade or antisaccade ) . Such associative learning can be realized by the mechanisms of flexible sensorimotor mapping proposed in other published studies ( for example , the model proposed in Fusi et al 2007 [15] ) . The gaze-holding control mimics a subject’s effort to withhold a gaze before the onset of the target and to suppress an express saccade triggered by the direct visual input . The gaze-holding control was modelled as a constant input to the fixation neurons ( both FNL and FNR ) in the action-selection module with a synaptic conductance of 2 . 0 nS . The strength of the constant input varies from trial to trial and was given by C h = C h 0 + k C rule + δ , where Ch0 ( = 960Hz ) is the mean strength of the gaze-holding control when a subject is actively fixating . The mean holding strength increased during an antisaccade trial because the subject tended to make more effort to suppress unwanted express saccades . This increase is described by the second term kCrule where k = 140 Hz . The trial-to-trial variability was modelled in the third term . In each trial , the value of δ was determined by randomly drawing a number from a Gaussian distribution with a zero mean and a standard deviation of 240 Hz . To avoid drawing an extremely strong or a negative control level from the Gaussian distribution , we set an upper limit ( 400 Hz ) and a lower limit ( -960 Hz ) for δ . The gaze-holding control was turned on at the start of a trial and was turned off 150 ms after the onset of the target signal in all trial types . We noted that this offset latency , together with the trial-to-trial variability of the strength of gaze-holding control , were important in producing a desired reaction time distribution , especially in prosaccade trials . Therefore the related parameters were determined by matching the behavior outcome of the model to that of the typical experimental observations [41 , 43] . The sensorimotor remapping control mimics a subject’s action to switch the response rule from prosaccade to antisaccade . In a prosaccade trial , no remapping control is needed and the Dir and Inv neurons only receive the target stimuli and the default background input as indicated in Table 3 . In an antisaccade trial , the subject needs to suppress the direct map and facilitate the inverted map in the remapping module . Therefore , the remapping control was modelled as a reduced background input to the neurons ( Dir ) in the direct map and an increased background input to the neurons ( Inv ) in the inverted map . The amount of change in the background inputs is calculated using the following equations: C rem Dir = k Dir C rule and C rem Inv = k Inv C rule , where kDir = − 1093Hz and kInv = 2000 Hz . C rem Dir and C rem Inv were then added to the background inputs indicated in Table 4 . According to the equations , in prosaccade trials neurons in the direct map receive a larger background excitation than neurons in the inverted map whereas in antisaccade trials it is opposite . The input firing rates ( sensory and top-down control ) described above seem to be arbitrary . However , they all fall into a physiologically reasonable range . Assuming that there are 240 neurons ( same with the most populations in the model ) in each upstream input neuron pool , the maximum sensory input of 28 , 000 Hz corresponds to 116 . 7 Hz per visual input neuron . Indeed , a large number of studies ( Carandini & Ferster , 2000 [76] for example ) on the mammalian visual cortex have reported such a strong response ( > 100 Hz ) to salient visual stimuli . Regarding the input associated with the gaze-holding or remapping controls , the mean strength of the controls in the antisaccade trials corresponds to a much smaller per-neuron firing rate ( ∼ 4-8 Hz ) . This is also a reasonable range considering that the top-down inputs are assumed to partially originate from the prefrontal cortex . Several studies have discovered the rule-dependent or cue-dependent neurons in the prefrontal cortex with a firing rate between 5Hz and 20 Hz , and rarely exceeding 50Hz [23] . In the model we did not manually add trial-to-trial variability to the remapping control as we did for the gaze-holding control . This is because the remapping module performed probabilistic decisions and already exhibited a high degree of trial-to-trial variability [35] . Adding more variability to the top-down remapping control , or equivalently , adding variability to the activity of Dir and Inv neurons did not help producing a better result . Once the subject applied the remapping control based on the rule signal , whether the response rule could be successfully switched was determined stochastically in the decision layer , where the trial-to-trial variability originated .
|
We propose a novel neural circuit mechanism and construct a spiking neural network model for resolving conflict between an automatic response and a volitional one . In this mechanism the two types of responses compete against each other under the modulation of top-down control via multiple neural pathways . The model is able to reproduce a wide range of neuronal and behavioral features observed in various studies and provides insights into not just how subjects make correct responses and fast errors , but also why they make slow errors , a type of error often overlooked by previous modeling studies . The model suggests critical roles of tonic ( non-racing ) top-down inhibition and near-threshold decision-making in neural competition .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"neural",
"networks",
"decision",
"making",
"reaction",
"time",
"nervous",
"system",
"brain",
"social",
"sciences",
"neuroscience",
"cognitive",
"neuroscience",
"cognition",
"vision",
"sensory",
"physiology",
"computer",
"and",
"information",
"sciences",
"animal",
"cells",
"basal",
"ganglia",
"neural",
"pathways",
"visual",
"system",
"cellular",
"neuroscience",
"psychology",
"eye",
"movements",
"neuroanatomy",
"cell",
"biology",
"anatomy",
"physiology",
"neurons",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"sensory",
"systems",
"sensory",
"perception",
"cognitive",
"science"
] |
2016
|
Conflict Resolution as Near-Threshold Decision-Making: A Spiking Neural Circuit Model with Two-Stage Competition for Antisaccadic Task
|
Pyridoxal 5′-phosphate ( PLP ) , the active form of vitamin B6 , has been implicated in preventing human pathologies , such as diabetes and cancer . However , the mechanisms underlying the beneficial effects of PLP are still unclear . Using Drosophila as a model system , we show that PLP deficiency , caused either by mutations in the pyridoxal kinase-coding gene ( dPdxk ) or by vitamin B6 antagonists , results in chromosome aberrations ( CABs ) . The CAB frequency in PLP-depleted cells was strongly enhanced by sucrose , glucose or fructose treatments , and dPdxk mutant cells consistently displayed higher glucose contents than their wild type counterparts , an effect that is at least in part a consequence of an acquired insulin resistance . Together , our results indicate that a high intracellular level of glucose has a dramatic clastogenic effect if combined with PLP deficiency . This is likely due to an elevated level of Advanced Glycation End-products ( AGE ) formation . Treatment of dPdxk mutant cells with α-lipoic acid ( ALA ) lowered both AGE formation and CAB frequency , suggesting a possible AGE-CAB cause-effect relationship . The clastogenic effect of glucose in PLP-depleted cells is evolutionarily conserved . RNAi-mediated silencing of PDXK in human cells or treatments with PLP inhibitors resulted in chromosome breakage , which was potentiated by glucose and reduced by ALA . These results suggest that patients with concomitant hyperglycemia and vitamin B6 deficiency may suffer chromosome damage . This might impact cancer risk , as CABs are a well-known tumorigenic factor .
It is now widely accepted that chromosome aberrations ( CABs ) can contribute to cancer development . Deletions , duplications and chromosome exchanges such as dicentrics and translocations can ultimately result in loss of genetic material ( loss of heterozygosity ) , DNA amplification and formation of aberrant gene fusions , thus promoting carcinogenesis [1]–[3] . Tumor development has been also associated with chromothripsis , a phenomenon of massive DNA fragmentation followed by multiple chromosomal rearrangements involving between one and a dozen of chromosomes [4]–[6] . It is currently unclear whether cells with chromothripsis are generated by a single event or result from multiple successive events involving more than one cell cycle [7]–[9] . Abundant evidence indicates that CABs are mainly generated by unrepaired or improperly repaired double strand breaks ( DSBs ) . DBSs can be induced by external agents such as ionizing radiations and chemical mutagens or by endogenous factors such as the free radicals generated by the oxidative metabolism or errors in DNA replication [10]–[13] . DSBs are repaired through two distinct but interconnected mechanisms - non-homologous end joining ( NHEJ ) and homologous recombination ( HR ) - both of which are mediated by evolutionarily conserved proteins . NHEJ joins broken chromosome ends directly and relies on the activities of the Mre11-Rad50-Nbs ( MRN ) complex , the Ku heterodimer , and the Ligase 4 complex . HR and its variant single strand annealing ( SSA ) are based on recombination with homologous genomic sequences , and exploit a variety of factors including the MRN complex , RAD51 , BRCA1 , BRCA2 , BLM and ATM [10] . Mutations in ATM ( Ataxia Telangiectasia Mutated ) , MRE11 , NBS1 ( Nijmegen Breakage Syndrome ) , BRCA1 ( Breast Cancer 1 ) , BRCA2 and Ligase 4 cause human syndromes characterized by both CABs and cancer predisposition , highlighting the connection between CABs and cancer [14] , [15] . Several studies have shown that inadequate intake of micronutrients results in DNA damage and cancer in humans [16] , [17] . A micronutrient that protects from DNA damage and is beneficial for cancer prevention is Pyridoxal 5′-phosphate ( PLP ) [16] , [18]–[20] . PLP is the metabolically active form of vitamin B6 generated by pyridoxal kinase; it acts as a cofactor for more than 140 enzymes , which catalyze a myriad of biochemical reactions . It has been estimated that PLP is involved in 4% of all catalytic activities and it is known to play essential roles in wide range of metabolic and developmental processes including amino acid , fatty acid and neurotransmitter metabolism [19]–[21] . There is also evidence that PLP quenches the oxygen reactive species acting as a potent antioxidant [22]–[24] and antagonizes Advanced Glycation End-products ( AGE ) formation [19] , [25] , [26] . Based on its wide range of functions it is not surprising that PLP is beneficial for many human diseases . Indeed , many epidemiological studies indicate that PLP protects from cancer , diabetes , cardiovascular diseases and neurological disorders [19] , [20] . However , the mechanisms underlying the effects of PLP at the molecular and cellular levels are still poorly understood . Here we show that in both Drosophila and human cells an elevated intracellular level of glucose has a dramatic clastogenic effect if combined with PLP deficiency; some cells exhibit an extensive chromosome damage that is reminiscent of chromothripsis . In addition , we show that PLP deficiency greatly potentiates AGE formation . Our findings suggest vitamin B6 deficiency coupled with hyperglycemia results in chromosome damage , which might promote carcinogenesis .
We identified a mutation ( dPdxk1 ) in the Drosophila gene encoding pyridoxal kinase ( Pdxk ) by a cytological screen of 1680 third chromosome lines bearing recessive mutations that cause death at late larval stages ( see Materials and Methods ) . Pdxk plays a critical role in the formation of pyridoxal 5′-phosphate ( PLP ) , the active form of vitamin B6 [19] . Mitotic cells from colchicine-treated dPdxk1 mutant brain displayed ∼6% chromosome aberrations ( CABs ) ; the frequency of aberrations in wild type cells is ∼0 . 5% ( Fig . 1A–C ) . Genetic analyses placed dPdxk1 in the 67A9-67B2 polytene chromosome interval that contains only 11 genes ( figure S1A ) . A mutation in one of these genes , l ( 3 ) 67Ab , failed to complement dPdxk1 and has been thus renamed dPdxk2 . The frequencies of CABs observed in dPdxk1/Df ( 3L ) AC1 ( 67A2-67D11 ) and dPdxk2/Df ( 3L ) AC1 hemizygotes were significantly higher than those seen in the respective dPdxk1 and dPdxk2 homozygotes ( Figures 1C and S1A ) , suggesting that both mutant alleles are hypomorphic . DNA sequencing showed that both dPdxk1 and dPdxk2 carry lesions in the Pdxk-encoding CG34455 gene , which specifies a 304 aa protein ( http://flybase . org ) . dPdxk1 carries an A→G transition ( # 338 ) in a splicing acceptor site , which is predicted to lead to a truncated protein of 83 aa; dPdxk2 carries a T→C transition ( # 700 ) that leads to a phenylalanine to serine substitution in the active site of Pdxk . To confirm the identity of the dPdxk gene we performed complementation tests with three different transgenes: one contained the endogenous promoter and the CG34455 genomic sequence fused in frame with the GFP sequence; the other transgenes were both placed next to the tubulin promoter and contained either the CG34455 or the human PDXK cDNA fused in frame with the 3HA sequence ( Figure S1B ) . All transgenes rescued the CAB phenotype of dPdxk mutants; the transgene placed under the control of the endogenous promoter also rescued the dPdxk1 lethal phenotype . To ask whether the CAB phenotype was due to PLP deficiency , we grew dPdxk mutant flies in food supplemented with 10−2 M PLP in 4% sucrose , or in a food supplemented with 4% sucrose only ( protocol 1 , Figure 1B ) . The PLP-containing food completely suppressed the CAB phenotype . However , to our surprise , dPdxk mutant larvae grown in the food with only sucrose displayed a 3-fold increase in CAB frequency compared to mutant larvae grown in normal food ( Figure 1D ) . Because larval feeding does not allow precise control of sugar and PLP intake , we incubated dissected brains from third instar larvae for 4 h in saline/FBS ( 0 . 7% NaCl supplemented with 10% fetal bovine serum ) containing defined quantities of sugars or drugs ( protocol 2 , Figure 1B ) . This analysis revealed that 1% glucose causes substantial increases in the CAB frequency in both dPdxk1 ( 3 . 5-fold ) and dPdxk2 ( 4 . 1-fold ) mutant brains ( Figure 1E ) . The effects of 1% fructose were even more dramatic , as it caused 6 . 9 and 10 . 3-fold increases in the CAB frequency in dPdxk1 and dPdxk2 brains , respectively . 1% glucose or fructose did not induce CABs in wild type brains ( see Figure 2B below and data not shown ) . Most importantly , the clastogenic effects of sugars were drastically reduced by 1 mM PLP ( Figure 1E ) . In addition to glucose and glycogen Drosophila cells and hemolymph contain trehalose , a disaccharide formed from two glucose moieties . Trehalose is a particularly stable high-energy storage molecule that can be transported and accumulated to high concentrations without toxic effects [27] . We thus focused on glucose and measured its concentration in dPdxk mutant tissues . Using a Hexokinase-based detection method , we found that the hemolymph of dPdxk1 larvae contains nearly twice as much glucose as that of wild type larvae ( Figure 2A ) . We next analyzed the dose-effect relationships between the intracellular glucose concentration ( IGC ) in larval brain cells and the CAB frequency . Dissected brains were incubated for 4 h in saline/FBS with increasing glucose concentrations and then examined for both the IGC and the presence of CABs ( protocol 2 , Figure 1B ) . In dPdxk1 brains exposed at high glucose concentrations the frequency of cells with more than 5 CABs was quite high ( Figure 2B ) and in many cases the CAB number per cell could not be assessed . Thus , in order to render the data comparable , we considered the frequency of cells with CABs instead of the CAB frequency ( number of CABs/number of cells scored ) as in Figure 1C–E . In dPdxk1 mutant brains , the frequency of cells with CABs increased with the IGC ( Figure 2B ) . In wild type brains treated with 1% or 5% glucose , the frequencies of cells with broken chromosomes were comparable to that of non-treated controls , and only treatments with 10% glucose resulted in a significant CAB increase ( Figure 2B ) . Importantly , wild type and dPdxk1 brains with comparable IGCs ( wt brains in 10% glucose and dPdxk1 brains in 5% glucose ) displayed significantly different frequencies of cells with CABs , with mutant brains showing a higher frequency of metaphases with damaged chromosomes than controls ( Figure 2B ) . These results indicate that CABs are not caused by a high IGC only , but by the simultaneous occurrence of an elevated IGC and a low PLP level . Thus , in the presence of low PLP levels ( as shown in later figure ) , sucrose , glucose and fructose behave as potent clastogens . Interestingly , in brains with high IGC approximately 10% of the metaphases with more than 5 CABs showed an extensive chromosome fragmentation ( Figure 1A ) . The frequency of chromosome breaks and rearrangements in the latter cells is clearly much higher than that expected from a Poisson distribution of CABs . The occurrence of this massive chromosome damage might reflect an increase of IGC beyond a critical threshold in cells where PLP is strongly reduced . Additional evidence for a high IGC in dPdxk1 mutant brains was provided by the analysis of mitosis in larval brains . We noticed that preparations of dPdxk1 mutant brains incubated in 1% glucose ( protocol 2 , Figure 1B ) , and treated for 90 min with 10−5 M colchicine before fixation , contain several anaphases . Immunostaining for tubulin revealed that these preparations display many mitotic spindles with a slightly reduced microtubule ( MT ) density but otherwise normal ( Figure 3A–C ) . In contrast , preparations of colchicine-treated wild type or dPdxk1 mutant brains did not exhibit any recognizable spindle structure ( Figure 3A ) . We thus asked whether PLP addition alleviates the colchicine resistance of the dPdxk1 mutant spindles incubated in 1% glucose . We found that treatments with 1% glucose ( protocol 2 , Figure 1B ) result in 40 . 3 % colchicine-resistant spindles . However , if brains were incubated in both 1% glucose and 1 mM PLP , the frequency of resistant spindles was only 20 . 1% ( Figure 3D ) . These observations suggest that addition of exogenous PLP compensates the PLP deficiency caused by the dPdxk1 mutation , reducing the glucose level and its effects on spindle MTs . Tubulin tyrosination and detyrosination is a well-known reversible enzyme-mediated post-translational modification that affects MT stability . MTs that end with a tyrosine residue at the C-terminus of α-tubulin ( Tyr-MTs ) are more dynamic and more resistant to nocodazole-induced depolymerization than detyrosinated MTs ( usually called Glu-MTs because their C-terminal residue is Glu instead of Tyr ) [28] , [29] . Early studies showed that human umbilical vein endothelial cells display colchicine-resistant MTs and reduced proliferation when cultured in high glucose . The MTs of these cells were characterized by frequent loss of the terminal tyrosine residue and their colchicine resistance was corrected by tyrosine addition [30] , [31] . Based on these results , we incubated for 4 hours dPdxk1 mutant brains in saline/FBS containing 1 . 7 mM tyrosine and 10−5 M colchicine ( added 90 min before fixation ) . These brains showed a substantial reduction in the frequency of mitotic figures with undepolymerized spindles compared to brains treated in the same way but without tyrosine addition ( 18 . 3 vs 40 . 3 % ) ; addition of glycine ( 1 . 7 mM ) instead of tyrosine did not affect the frequency of colchicine-resistant spindles ( Figure 3D ) . Thus , in line with the studies on human umbilical vein endothelial cells , addition of tyrosine corrects colchicine resistance also in Drosophila dPdxk mutant cells that accumulate an elevated glucose amount . We also studied the effect of tyrosine on CAB formation . We examined the CAB frequencies in dPdxk mutant brains incubated in either 1% glucose and 1 . 7 mM tyrosine or in 1% glucose only ( protocol 2 , Figure 1B ) . We found that the frequency of CABs in tyrosine treated brains ( 18 . 0 %; 8 brains; 506 metaphases ) is not significantly different from that of brains exposed to glucose only ( 20 . 9 %; 8 brains; 367 metaphases ) . These results suggest that colchicine-resistant spindles and CABs are unrelated outcomes of PLP deficiency , as tyrosine affects spindle resistance but not CAB formation . There are several well-known PLP inhibitors , some of which are used in pharmaceutical treatments . These drugs include the vitamin B6 analog 4-deoxypyridoxine ( 4-DP ) , isoniazid ( tuberculosis treatment , antidepressant ) , penicillamine ( antirheumatic ) , and cycloserine ( tuberculosis treatment , antidepressant ) [32] , [33] . Brains from wild type larvae incubated for 4 hours in the presence of 4-DP , isoniazid or penicillamine ( protocol 2 , Figure 1B ) showed higher levels of CABs than untreated controls ( Figure 4 ) . When these brains were also exposed to 1% glucose , the CAB frequency was further and significantly increased ( Figure 4 ) . Similarly , brains from larvae grown in the presence of cycloserine displayed a significant increase in CABs compared to untreated controls , and this effect was potentiated by glucose addition ( Figure 4 ) . These results indicate that glucose induces CABs when vitamin B6 activity is reduced by drug treatments . Studies on mammalian systems have shown that glucose accumulation within the cell might depend on either lack of insulin ( type-1 diabetes ) or defects in the insulin-signaling pathway ( type-2 diabetes ) . Insulin promotes phosphorylation of AKT , which leads to phosphorylation and inactivation of glycogen synthase kinase 3 ( GSK3 ) allowing glycogen formation ( Figure 5A , top ) . In the absence of insulin , the active form of GSK3 phosphorylates and inhibits glycogen synthase ( GS ) , the enzyme that catalyzes glycogen synthesis ( Figure 5A , bottom ) [34] . Thus loss of AKT or lack of AKT phosphorylation should result in glycogen synthesis inhibition and glucose accumulation . Flies and mammalian systems employ similar mechanisms ( Figure 5A ) for regulation of carbohydrate homeostasis ( reviewed in [35] , [36] ) . The Drosophila genome encodes eight insulin-like peptides ( DILPs ) that are considered orthologous to mammalian insulin . These peptides are expressed in tissue- and stage-specific manner during development [37]–[39] . DILP2 is the closest homologue of human insulin and the most highly expressed DILP in the two bilaterally symmetric clusters of brain cells dubbed median neurosecretory cells ( mNSCs ) or insulin producing cells ( IPCs ) ; mNSCs/IPCs produce Drosophila insulin and are the main insulin suppliers during larval growth [37]–[39] . Using an antibody that specifically recognizes the mNSCs/IPCs [40] we found that dPdxk1 mutant brains show a normal concentration of DILP-2 ( Figure 5B ) . We next tested whether exogenous insulin has the ability of stimulating phosphorylation of Akt at Ser 505 in dPdxk1 mutant brains; this residue is homologous to mammalian Ser 473 whose phosphorylation promotes full Akt activation [41] ( Figure 5 ) . After stimulation with human synthetic insulin , dPdxk1 mutant brains displayed a limited but statistically significant reduction in the Akt phosphorylation level compared to controls ( Figures 5C , D and S2 ) . This suggests that glucose accumulation in dPdxk1 mutants is at least in part due to a defect in insulin signaling . PLP is cofactor of several enzymes involved in thymidylate ( dTMP ) biosynthesis . It has been previously shown that mutants in BUD16 , the yeast gene that encode pyridoxal kinase , are defective in dTMP synthesis and incorporate more uracil nucleotides in their DNA than nonmutant controls [18] . Thus , in the attempt of defining the primary defect leading to CABs , we asked whether dPdxk1 mutants accumulate uracil . We used HPLC/MS to determine uracil concentration in larval extracts and found that dPdxk1 mutant larvae exhibit lower dTTP and higher dUTP levels compared with wild type controls ( Figure 6A ) . HPLC/MS also showed that the PLP level in dPdxk1 mutant larvae is approximately 50 % of that found in wild type controls ( Figure 6A ) . Because nucleotide unbalance can affect DNA synthesis and cause CABs [42] we asked whether brain cells of dPdxk1 mutants were sensitive to hydroxyurea ( HU ) . HU is expected to cause a DNA replication stress because inhibits ribonucleotide reductase and thus decreases the production of deoxyribonucleotides . We treated brains for 15 min with 2 mM HU and then placed them in saline for 2 . 5 hours before fixation . This treatment did not substantially reduce the mitotic index in control and mutant brains . HU-treated dPdxk1 brains displayed a CAB frequency that was only slightly higher than the sum of the frequencies observed in HU-treated controls and in untreated dPdxk1 mutants ( Figure 6B ) . Thus , dPdxk mutations cause little or no increase in the sensitivity of Drosophila cells to HU . It has been reported that PLP counteracts AGE ( Advanced Glycation End products ) formation [25] , [26] . AGEs are heterogeneous molecules formed after nonenzymatic glycosylation ( glycation ) of proteins , lipids , or nucleic acids . AGE formation has been associated with the production of DNA damaging reactive oxygen species , and with the progression of several disorders including diabetes complications , neurodegenerative and cardiovascular diseases [43] , [44] . Immunostaining with an anti-human AGE antibody revealed that both untreated and glucose-treated dPdxk1 mutant brains ( protocol 2; Figure 1B ) consistently exhibit higher frequencies of AGE-positive cells than wild type controls . Here again , PLP addition prevented AGE formation ( Figure 7A , B ) . We next treated dPdxk1 brains with α-lipoic acid ( ALA ) , an antioxidant compound that antagonizes AGE formation and cooperates with vitamin B6 in ameliorating insulin resistance in pre-diabetic rats [45] , [46] . We incubated dPdxk1 brains in 10 mM ALA , with or without glucose addition ( protocol 2 Figure 1B ) . In all cases , ALA reduced the frequencies of AGE positive cells . Interestingly , ALA also reduced the frequency of CABs in both untreated and glucose-treated dPdxk1 mutants ( Figures 7B , C ) . We finally asked whether glucose causes CABs in human cells with reduced PLP levels . RNAi-mediated PDXK depletion in HeLa cells ( 38 % of the control level; Figure 8A ) resulted in a dramatic increase of CABs ( Figures 8B–E ) . In addition , PDXK RNAi cells grown in media containing final glucose concentrations of 0 . 9 % or 2 . 45 % showed significant CAB increases compared to cells grown in standard medium ( 0 . 45 % glucose ) . Addition of PLP ( at a final concentration of 2 mM ) to either the standard or the 2 . 45 % glucose medium strongly reduced the CAB frequency ( Figure 8E ) . Interestingly , PDXK RNAi cells exposed to high glucose concentrations displayed several metaphases with extensive chromosome fragmentation . These shattered metaphases ( Figure 8D ) were similar to the metaphases with multiple breaks observed in glucose-treated Drosophila Pdxk mutant brains ( Figure 1A ) . The frequency of such metaphases was higher in fructose-treated cultures ( 2% fructose plus 0 . 45% glucose ) than in cultures exposed to 2 . 45% glucose ( Figure 8E ) . Thus , fructose appears to be more efficient than glucose in the induction of chromosome shattering in both Drosophila and human PLP-deficient cells . We also found that HeLa cells behave as Drosophila brain cells in their responses to the vitamin B6 analog 4-DP and ALA . 4-DP caused extensive chromosome breakage in HeLa cells , which was potentiated by a high glucose concentration . Treatment of PDXK-depleted or 4-DP treated cells with ALA reduced the CAB frequency ( Figure 8C ) .
Studies on BUD16 yeast mutants showed that PLP is required for dTMP biosynthesis [18] . Cells bearing mutations in BUD16 displayed excessive uracil incorporation into DNA compared to wild type , a condition that may lead to DSBs via uracil excision and production of abasic DNA sites . However , a deficiency of uracil glycosylase , the main enzyme that removes uracil from DNA , did not suppress the mutagenic effects of BUD16 , indicating that uracil excision is not a major cause of DNA damage in cells with low PLP levels . Moreover , BUD16 mutant cells showed an extreme sensitivity to hydroxyurea ( HU ) , which inhibits ribonucleotide reductase that catalyzes the de novo synthesis of dNTPs . Based on these results , it has been suggested that PLP deficiency in yeast leads to DNA lesions and gross chromosomal rearrangements by causing a strong nucleotide imbalance that impairs DNA synthesis [18] . We have shown that Drosophila cells bearing mutations in dPdxk also exhibit a dUTP excess compared to wild type controls . However , dPdxk mutant cells do not appear to be particularly sensitive to HU . Untreated dPdxk1 mutant cells and HU-treated wild type cells showed 6% and 4% CABs , respectively; HU-treated dPdxk1 mutant cell displayed 16% CABs , a frequency that is only slightly higher than the frequency expected ( 10% ) if HU and dPdxk1 mutations acted independently . In contrast HU-treated brains of mutants in tim2 , which encodes a replisome-associated factor , showed a ∼10-fold increase in the CAB frequency compared to the sum of the frequencies observed in untreated tim2 mutant cells and HU-treated controls [47] . The finding that HU does not greatly exacerbate DNA damage in Drosophila dPdxk mutants suggests that the primary cause of CAB formation in these mutants is not nucleotide unbalance . We have found that in both untreated and glucose-exposed dPdxk mutant brains there is a substantial increase in AGE formation . Most importantly we found that ALA reduces both AGE formation and the CAB frequency . Given that AGE formation is accompanied by the production of DNA damaging reactive oxygen species [43] , [44] and that ALA is a potent antioxidant compound [46] , we suggest that the DNA damage leading to CABs in dPdxk mutants , and especially in dPdxk mutants exposed to sugars , is at least in part a consequence of AGE formation . It is also possible that the lesions that cause CABs are generated by the simultaneous presence of AGE-linked reactive oxygen species and an unbalanced nucleotide pool . In this context , we propose that the higher frequency of CABs observed in fructose-treated brains compared to those exposed to glucose ( Figure 1E ) might reflect the higher efficiency of fructose in the initiation of the Maillard reaction that leads to AGE formation [48]–[50] . Our results clearly show that the hemolymph and the brain cells of dPdxk mutants contain a higher glucose concentration than their wild type counterparts . We have also shown that Pdxk mutant brains incubated in glucose-enriched saline accumulate more glucose than wild type brains . Our results suggest that the increase in glucose concentration observed in Pdxk mutant brains is at least in part due to an acquired insulin resistance . It has been recently shown that Drosophila larvae reared on a high-sugar diet exhibit diminished insulin-induced Akt phosphorylation at Ser 505 compared to controls grown in normal medium [41] . However , the reduction in Akt phosphorylation observed in larvae grown in a sugar-rich medium [41] is substantially stronger than that found in Pdxk mutants . A possible reason for this difference in the response to insulin stimulation might be the time of exposure to high sugar . Larvae grown in sugar-rich medium were exposed to sugar throughout development . In contrast , it is likely that in dPdxk mutant larvae the maternal supply of Pdxk was progressively diluted during development becoming critically depleted only in late larval stages [51] . However , while this interpretation explains the differences between our data and those of Musselman and coworkers [41] , it does not explain how dPdxk mutant acquired insulin resistance . We propose that PLP deficiency leads to an initial glucose accumulation through an as yet unknown mechanism and that this accumulation leads to insulin resistance . According to this interpretation , dPdxk mutant brains incubated in glucose- or fructose-enriched saline accumulate more sugar than wild type brains through two mechanisms , one linked to PLP deficiency per se and the other linked to insulin resistance . The mitotic spindles of dPdxk mutant brains incubated in 1% glucose displayed an unexpected resistance to colchicine-induced depolymerization . This resistance was attenuated by treatment with tyrosine but not with glycine . Early work on human umbilical vein endothelial cells ( HUVEC ) has suggested a possible interpretation of these results . Culture of HUVECs in high glucose resulted in frequent loss of the terminal tyrosine residue from MTs , colchicine-resistant MTs and reduced cell proliferation [30] , [31] . Colchicine resistance was attributed to tubulin detyrosination at MT ends , a reversible modification that renders MTs less dynamic and less resistant to nocodazole-induced depolymerization than tyrosinated MTs [28] , [29] . Consistent with this interpretation La Selva and coworkers [31] showed that the defect in cell proliferation was corrected by tyrosine addition to the tissue culture medium . These authors did not attribute the loss of terminal tyrosine to tubulin glycation because they found that L glucose , which is not metabolized but can form Amadori bonds with protein amino groups , did not inhibit HUVEC proliferation . There is abundant evidence that tubulin glycation can occur . For example in vitro glycation of rat brain tubulin increases with glucose concentration , and diabetic rats display a dramatic increase in glycosylated tubulin and defective tubulin polymerization [52] . Moreover recent work has shown that tyrosine glycosilation can occur also in humans [53] . Thus , although to the best of our knowledge there is no experimental evidence that tyrosine glycosylation can regulate MT dynamics , one can envisage that high intracellular glucose concentrations might cause tyrosine glycation . Under this assumption , the high glucose concentration in dPdxk1 mutant cells would cause tyrosine glycation leading to detyrosinated MTs and colchicine resistance; addition of an excess of tyrosine would dilute the glycosylated tyrosine moiety , partially restoring the sensitivity of MTs to colchicine . Whatever the mechanism underlying the colchicine resistance of PLP depleted cells grown in high sugar , our finding has a potential translational impact and merits further study . An extrapolation of our results to human tumor therapy predicts that patients with vitamin B deficiency and hyperglycemia might be resistant to chemotherapy with MT-depolymerizing agents . Our experiments on HeLa cells have shown that the clastogenic effects of glucose and fructose in vitamin B6-deficient cells are evolutionarily conserved . HeLa cells in which the PDXK level was reduced to 38% of the control level by RNAi , displayed a 22-fold increase in the frequency of metaphases with CABs compared to non-RNAi cells . Both RNAi and non-RNAi cells were grown in standard DMEM medium , which contains 0 . 45% glucose and thus exceed the normal glucose concentration in blood , which is around 0 . 1 % . PDXK RNAi cells grown in DMEM containing final glucose concentrations of 0 . 9 % or 2 . 45 % ( or 0 . 45 % glucose and 2% fructose ) showed significant CAB increases compared to those grown in standard medium . Addition of PLP to either the standard or the 2 . 45% glucose medium strongly reduced the CAB frequency in PDXK RNAi cells . In addition , the PLP inhibitor 4-DP behaved as strong clastogen in HeLa cells , and its effect was potentiated by glucose addition to the medium . Together these results indicate that even moderate reductions in PLP level , such as that caused by a 62% diminution of PDXK , can results in a high CAB frequency . However , is possible that the elevated CAB frequency observed in PDXK RNAi cells is at least in part a consequence of the concomitant PLP deficiency and high glucose concentration ( 0 . 45% ) in the standard DMEM medium used to grow HeLa cells . The central issue raised by these results is whether the clastogenic effects of sugars can occur in human patient with a vitamin B6 deficiency . We believe that this is quite possible , as deficiency of this vitamin can be caused by many dietary , genetic and pharmacological factors [54] and blood glucose level over 0 . 5% ( 500 mg/dL ) have been observed in patients with hyperglycemic crises [55] . It is also conceivable that even relatively limited glycemia increases would cause CABs in people with particularly severe PLP deficiencies . An interesting observation made on both Drosophila and human cells is that PLP deficiency accompanied by high sugar results in several metaphases with extensive chromosome fragmentation . The frequency of CABs in these metaphases is much higher than that expected from the Poisson distribution of CABs , suggesting that these CABs might result from the synergistic combination of two at least in part independent events such as PLP reduction and sugar increase . Given that PLP and sugar concentration can fluctuate we speculate that during interphase a small fraction of the cells that suffer PLP deficiency/high sugar can rapidly revert to relatively normal metabolic condition through either a PLP increase or a sugar decrease . This situation would results in chromosome shattering followed by DNA repair in a relatively normal cellular environment and might therefore give rise to cells with multiple and stable rearrangements such as those observed in chromothripsis [4]–[6] , [56] . Inadequate intake of vitamin B6 has been associated with cancer risk [16] , [57]–[59] and recent studies have shown that a high expression level of PDXK has a positive impact on survival of non-small cell lung cancer ( NSCLC ) patients [60] . In addition , growing evidence indicates that diabetes patients have a higher risk of various types of cancer [61]–[65] . Our findings provide an important link between these studies , suggesting that vitamin B6 deficiency accompanied by hyperglycemia might lead to chromosome damage and thus trigger carcinogenesis [1]–[3] . Our work further suggests that patients with hyperglycemia who also take drugs that antagonize PLP , should compensate by taking extra amounts of vitamin B6 . Conversely , patients chronically treated with drugs that antagonize PLP should keep under control the level of sugar in their blood .
dPdxk1 was isolated by a cytological screen of larval brain squashes from a collection of 1680 EMS-induced late lethals generated in Charles Zuker's laboratory ( University of California , San Diego ) . dPdxk2 [or l ( 3 ) 67Ab] , Df ( 3L ) 29A6 , Df ( 3L ) AC1 and Df ( 3L ) ED4416 were all obtained from the Bloomington Stock center . The dPdxk mutations and the deficiencies were balanced over TM6B or TM6C , which carry the dominant larval marker Tubby ( http://flybase . bio . indiana . edu/ ) ; homozygous and hemizygous mutant larvae were recognized for their non-Tubby phenotype . Germline transformation and complementation analysis are illustrated in Figure S1 . All stocks were maintained on standard Drosophila medium at 25°C . Flies were grown in standard Drosophila medium containing 4 . 5 % sucrose , and experiments were performed at 25 C° . To analyze the effects of dPdxk mutations and drug treatments we followed two different protocols ( Figure 1B ) . In both protocols , wild type or dPdxk mutant larvae were grown in fly medium for 6 days with or without addition of sugars or drugs . In protocol 1 , brains dissected from 6-day third instar larvae were incubated for 90 min in 2 ml of saline ( 0 . 7% NaCl ) and 10−5 M colchicine and then fixed . In protocol 2 , brains dissected from third instar larvae were incubated in 2 ml of saline supplemented with 10% fetal bovine serum ( FBS , Gibco BRL ) for 4 hours with or without addition of sugar or drugs and then fixed; 90 min before fixation colchicine ( final concentration , 10−5 M ) was added to the saline/FBS to collect metaphases . In the experiments with hydroxyurea ( HU ) , we used a protocol different from those described in Figure 1B . Brains form six-day old third instar larvae were incubated for 15 min in saline with 2 mM HU ( Sigma ) , washed , placed in saline for 2 . 5 h and then fixed; 1 h before fixation , brains were treated with 10−5M colchicine . Drosophila metaphase chromosome preparations were obtained as previously described [66] and mounted in Vectashield H-1200 with DAPI ( Vector Laboratories ) to stain the chromosomes . Brain preparations for immunofluorescence and tubulin immunostaining were carried out according to Bonaccorsi et al . [67] . To stain the AGEs , brain squashes were incubated overnight at 4°C with a rabbit anti-human AGE antibody ( 1∶200 in PBS; ab23722 , Abcam , UK ) , which was detected by a 1-hour incubation at room temperature with Alexa 555-conjugated goat anti-rabbit IgG ( H+L ) ( 1∶300 in PBS , Molecular Probes ) . Immunostained preparations were mounted in Vectashield medium H-1200 with DAPI . Observations were carried out using a Zeiss Axioplan fluorescence microscope equipped with CCD camera ( Photometrics CoolSnap HQ ) . Glucose concentration in Drosophila hemolymph and brains was measured using the Infinity Glucose Hexokinase reagent ( Thermo scientific ) . To measure glucose in the hemolymph , samples of 10 larvae were washed in NaCl 0 . 7% , 0 . 1% Triton X-100 and then in d-H2O . The hemolymph of these larvae was collected and its glucose content measured following the protocol of Rulifson et al . [38] . The values reported in Figure 2A are the means ±SE of 8 samples of 10 larvae . To measure glucose in brains , samples of 20 brains were placed in 40 µl of 10−3 M EDTA , 10−2M KH2PO4 , and the complete protease inhibitor cocktail ( Roche ) , mechanically homogenized and then centrifuged at 14 , 000 rpm for 10 min . The supernatant was collected with a micropipette and used for glucose measurement according to Rulifson et al . [38] . The measures reported in Figure 2B are the means ±SE of 4 samples of 20 brains . Insulin stimulation of Akt phosphorylation was performed using recombinant human insulin ( Sigma , I0516 ) . Before stimulation , larvae were starved for 5 hours in an empty vial humidified with a drop of saline devoid of FBS . We then followed the protocol described by Musselman et al . [41] . However , at the end of the insulin stimulation procedure , instead of larvae , we homogenized samples of 20 isolated brains , which were then used for Western blotting analysis . To quantitate nucleotides and PLP in larval extracts we used the HPLC/MS method described by D'alessandro et al . [68] . Briefly , third instar larvae ( 20 per sample ) were washed in saline , homogenized , and then resuspended in 80 µl methanol; 100 µl of chloroform were then added to each tube . After 30 min mixing , 20 µl of ice-cold ultra-pure water was added to the tubes , which were centrifuged at 1 , 000 g for 1 min and then transferred to −20°C . After thawing , liquid phases were recovered and mixed to an equivalent volume of acetonitrile . The tubes were then centrifuged at 10 , 000 g for 10 min; the supernatants were recovered into 2 ml tubes , dried to obtain visible pellets , and resuspended in 200 µl of 5% formic acid in water . For metabolite separation we used an Ultimate 3000 high-resolution fast HPLC system ( LC Packings , DIONEX , Sunnyvale , USA ) , with a Dionex Acclaim RSLC 120 C18 column “2 . 1 mm×150 mm , 2 . 2 µm” . A 0–95 % linear gradient of solvent A ( 0 . 1% formic acid in water ) to B ( 0 . 1% formic acid in acetonitrile ) was employed over 15 min followed by a solvent B hold of 2 min , returning to 100% A in 2 min and a 6 min post-time solvent A hold . ESI mass spectrometry was performed as described previously using a High Capacity ion Trap HCTplus ( Bruker-Daltonik , Bremen , Germany ) [68] . Validation of HPLC/MS-eluted metabolites was performed by comparison with the standard metabolites . ANOVA statistical analysis was carried out using GraphPad Prism 5 software . Extracts for Western blotting of Drosophila proteins were prepared by lysing samples of 20 brains in 150 mM NaCl , 50 mM Tris-Hcl pH 7 . 5 , 30 mM NaF , 25 mM b-glycerophosphate , 0 . 2 mM Na3VO4 , Triton X-100 1% , and Complete protease inhibitor cocktail ( Roche ) . Extracts were immunoblotted according to Somma et al . [69]; blotted proteins were detected using rabbit anti-DILP2 ( 1∶2000; a gift of E . Hafen ) , rabbit anti-Phospho ( Ser 505 ) -Drosophila Akt ( 1∶1000; #4054 , Cell Signaling ) , or rabbit anti-pan-Akt ( 1∶1000; #4691 , Cell Signaling ) . To determine the phosphorylation level of Akt we performed three different experiment similar to that shown in Figure 5C . We thus analyzed twelve 20-brain independent samples for wild type or dPdxk mutant third instar larvae . In each experiment we determined the intensities of P-Akt bands normalized to both total Akt and the loading control [Giotto ( Gio ) a Drosophila Phosphatidylinositol transfer protein; ref [70] . Measurements were performed on unsaturated bands using Image J software ( http://rsb . info . nih . gov/ij/ ) for band quantification and normalization . HeLa cells were grown in DMEM ( Gibco BRL ) with 10% fetal bovine serum ( FBS , Gibco BRL ) in a humidified 5% CO2 atmosphere . PDXK siRNAs ( SIHK1569 , Sigma ) were transfected using Lipofectamine 2000 ( Invitrogen ) according to the manufacturer's instructions . Mock-transfected and siRNA-transfected cells where grown for 24 hours in normal medium , which was then supplemented with glucose , fructose , PLP ( 2 mM ) , ALA ( 10 mM ) or 4-DP ( 30 mM ) . In all cases , 72 hours after the beginning of treatments colcemid ( 0 . 05 µg/ml , Gibco BRL ) was added to the cultures for 3 hours before fixation according to Revenkova et al . [71] . Chromosome preparations were mounted in Vectashield H-1200 with DAPI . Human cell extracts were prepared according to Cherubini et al . [72] , and Western blotting was performed as described in Somma et al . [69]; PDXK was detected using a mouse anti-PDXK antibody ( 1∶500; 89006590 , Abnova )
|
We show that the active form of vitamin B6 ( Pyridoxal 5′-phosphate , PLP ) plays an important role in the maintenance of genome integrity . We found , using Drosophila as a model system , that PLP deficiency results in chromosome breaks and rearrangements ( collectively dubbed chromosome aberrations , abbreviated with CABs ) . Most importantly , we observed that in PLP deficient cells , sucrose , glucose , or fructose strongly enhance the frequency of CABs . The mutagenic effects of sugars in the presence of PLP deficiency are evolutionarily conserved , as PLP depletion or inhibition in human cells results in CAB formation , which is potentiated by glucose or fructose . These results suggest that patients with concomitant hyperglycemic crises and vitamin B6 deficiency may suffer genetic damage , which might promote cancer and diabetes complications . Our work further suggests that patients treated with PLP antagonist drugs should keep under control the level of sugar in their blood and compensate their vitamin B6 level .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"genetics",
"of",
"disease",
"cancer",
"genetics",
"genetic",
"mutation",
"genetics",
"biology",
"cytogenetics"
] |
2014
|
Sugar and Chromosome Stability: Clastogenic Effects of Sugars in Vitamin B6-Deficient Cells
|
Schistosomiasis is an important neglected tropical disease caused by digenean helminth parasites of the genus Schistosoma . Schistosomes are unusual in that they are dioecious and the adult worms live in the blood system . MicroRNAs play crucial roles during gene regulation and are likely to be important in sex differentiation in dioecious species . Here we characterize 112 microRNAs from adult Schistosoma mansoni individuals , including 84 novel microRNA families , and investigate the expression pattern in different sexes . By deep sequencing , we measured the relative expression levels of conserved and newly identified microRNAs between male and female samples . We observed that 13 microRNAs exhibited sex-biased expression , 10 of which are more abundant in females than in males . Sex chromosomes showed a paucity of female-biased genes , as predicted by theoretical evolutionary models . We propose that the recent emergence of separate sexes in Schistosoma had an effect on the chromosomal distribution and evolution of microRNAs , and that microRNAs are likely to participate in the sex differentiation/maintenance process .
Human schistosomiasis is a neglected tropical disease ( NTD ) caused by blood flukes of the genus Schistosoma . Schistosomiasis is estimated to affect over 200 million people in developing tropical and subtropical countries , with over 90% of cases being confined to Africa [1] , [2] . Schistosoma mansoni is primarily responsible for intestinal and hepatic schistosomiasis in Africa , the Arabian peninsula , parts of South America and the Caribbean Islands [3] . Unlike most other flatworms ( phylum Platyhelminthes ) , Schistosoma species are dioecious; that is , they have two differentiated sexes . The emergence of sexual dimorphism in these species is believed to be associated with adaptation to warm-blooded vertebrates from a hermaphrodite ancestor in cold-blooded vertebrates [4] . Schistosoma mansoni has seven pairs of autosomes and one pair of sexual chromosomes with a ZW system , i . e . , females are the heterogametic sex [5] . Like other species with a ZW-based system of sex determination , there is no apparent global dosage compensation in females [6] . The origin of Schistosoma sexuality has attracted much attention [4] , [7]–[9] . Moreover , as the eggs laid by the female worms are primarily responsible for the pathology associated with schistosomiasis , the mechanisms associated with pairing and egg-laying , including expression of sex specific genes are of great interest . In the last years , different groups have characterized , using genomic and proteomic approaches , gene products with differential expression between males and females in Schistosoma species [10]–[13] . Since both sexes are necessary for the colonization of the host , sex-biased genes are potential targets for the infection control of schistosomes . MicroRNAs are short endogenous RNA molecules that regulate gene expression by targeting mature mRNA transcripts [14] . This mechanism of post-transcriptional regulation is conserved in animals and is likely to be involved in all aspects of cellular function [15] . The microRNA content of Schistosoma japonicum , which affects large endemic areas around the river Yangtze in China , has been studied in detail by several groups [16]–[21] . In addition , the previous characterization of microRNAs from other non dioecious species of flatworms , such as Echinococcus granulosus and the non-parasitic Schmidtea mediterranea [22] , [23] , provides a background against which to identify Schistosoma-specific microRNAs with a potential role in sexual development and host-parasite interactions . Current knowledge of S . mansoni microRNAs is limited and mostly based on computational predictions [24] , [25] . Moreover , S . mansoni provides an excellent model to study the evolution and function of sex-biased microRNAs . Here we use deep sequencing of RNA libraries to explore the microRNA content of S . mansoni , identify microRNAs specific to the schistosomes , and study the potential impact of sex-biased microRNAs in sexual differentiation .
We have used small RNA deep sequencing to identify a total of 112 microRNAs in adult S . mansoni ( Table 1 , Supplementary Files S1 and S2 ) . Valid microRNA candidates were required to have reads mapping to both arms of the precursor sequences ( representing mature microRNA and microRNA* sequences ) except for those with a previously validated homolog ( see Materials and Methods ) . Our microRNA annotation procedure was intentionally conservative: we may not have detected some bona fide microRNAs , but our predictions are of high confidence . De Souza Gomes et al . [25] computationally identified 42 microRNA loci in S . mansoni , significantly expanding the previous set of 6 microRNAs [16] , [24] . We confirmed 20 of these ( Table 1 ) , all conserved in other species . We failed to detect the remaining 23 . All but two of the unconfirmed microRNAs were not conserved in other flatworms . A second recent work characterized 211 novel microRNAs in S . mansoni by cloning of small RNA sequences from adults and schistosomulas [26] . However , the majority of the candidate microRNAs map to many positions in the genome and only a few are reported to be within putative precursor hairpin structures . Indeed , only two of the reported 211 microRNAs were confirmed in our analyses ( mir-71a and let-7 ) . We identify 92 microRNAs not previously annotated in S . mansoni , eight with obvious homologs in other species ( Table 1 ) . Amongst these , we show that the deeply-conserved mir-124 locus produces microRNAs from both genomic strands in S . mansoni ( Supplementary Files S2 ) . The remaining 84 novel predictions had no detectable similarity with any known microRNA . To characterize the microRNAs conserved in the parasitic Schistosoma genus , we compared our sequenced microRNAs with those already described for S . japonicum [17]–[19] , [21] . We found that 26 out of our 112 microRNAs were conserved between these two species ( Figure 1 ) . In order to determine how many of those 26 are specific to the Schistosoma lineage , we searched the genomes of the flatworms Schmidtea mediterranea [22] and Dugesia japonica [27] for homologous sequences . Strikingly , all known microRNAs conserved between Schistosoma species were also conserved in other flatworms . We also detected a set of S . japonicum homologs for 12 of our newly identified S . mansoni sequences that were not conserved in other platyhelminthes ( Figures 1A and B ) . Hence , these 12 microRNAs are the first instances of schistosome-specific microRNAs . Homology searches in other animals showed that three microRNAs are specific to platyhelminthes: mir-755 , mir-2162 and mir-8451 ( Figure 1A ) . Thirteen microRNAs are protostome-specific and the remaining 10 are conserved across the animal kingdom ( Figure 1A ) . A total of 71 microRNAs identified in this study are not detected in any other species , and are therefore likely to be S . mansoni specific . To explore potential sex-biased expression , we compared the relative expression levels of the 112 microRNAs in males and females ( Figure 2A ) . We found 13 microRNAs that are differentially expressed between males and females . A significant excess [10] show increased expression in females ( Figure 2A , Table 2 ) . We further quantified the relative expression level of all 3 male-biased and 4 of the female-biased microRNAs by real-time PCR ( see Materials and Methods ) . The observed fold changes in our qPCR experiments were consistent with those observed in our RNAseq analysis ( Table 3 ) , although the two least biased microRNAs by RNAseq show small and non-significant changes by qPCR . We next evaluated whether microRNA loci on the sex chromosomes are biased towards differential expression between sexes . The current assembly of the S . mansoni genome does not differentiate between Z and W chromosomes . At this stage , we cannot therefore evaluate the two sexual chromosomes separately . In Figure 2B , we plot the relative enrichment of sex-chromosome-linked microRNAs for male and female-biased expression . We observed that the sex chromosome has fewer female-biased microRNAs ( ∼3-fold change ) than expected by chance , although the difference is only marginally significant ( p = 0 . 10 ) . The data therefore indicate that microRNA genes with female sex-biased expression may have a tendency to move out of sex chromosomes ( see Discussion ) . The mir-71/mir-2 microRNA cluster is highly conserved in invertebrates , and it has been shown that this cluster is duplicated in Platyhelminthes [25] , [28] . Interestingly , one of the clusters ( mir-71/2a/2b/2e ) is on the sex chromosomes while the other ( mir-71b/2f/2d/2c ) is on chromosome 5 [25] , [29] . This has led some authors to postulate that the mir-71/mir-2 clusters may be involved in sexual maturation in Schistosoma [25] . Our analysis reveals that all microRNAs in the autosomal cluster have female-biased expression , while the sex chromosome cluster does not show any bias ( Figure 2A ) . Interestingly , the two clusters emerged by a duplication in the ancestral lineage leading to Schistosoma , and the multiple copies in other Platyhelminthes [29] , [30] came from independent duplication events ( Supplementary Files S4 ) . This example may shed some light on how sex chromosomes evolved in dioecious species ( see Discussion ) .
Although the computational prediction of microRNAs has been useful to understand the biology of small RNAs in S . mansoni [25] , sequencing is required to validate the existence of these microRNAs as well as for detecting new sequences . Our work has confirmed the existence of 20 of the microRNAs predicted by de Souza Gomes et al . [25] and we have expanded the S . mansoni microRNA set to 112 loci . We specifically detect microRNAs expressed in sexually mature adults , and microRNAs specifically expressed in other developmental stages ( such as schistosomulas ) may have escaped our analysis . The use of deep sequencing also permits the characterization of microRNAs produced from both strands of the same locus , and we have identified sense and antisense microRNA production from the mir-124 locus . However , there is no evidence that this microRNA is also bidirectionally transcribed in other species . Indeed , bidirectionally transcribed microRNAs are rare and poorly conserved; only two cases of conserved bidirectional microRNAs are known in protostomes: iab-4 and mir-307 [31] . We observe an excess of microRNAs that exhibit female-biased expression ( Figure 2A ) . This is in agreement with the overall female-bias observed for protein-coding genes in both S . mansoni [10] and S . japonicum [11] , although a recent expression analysis in S . japonicum showed no gender bias [13] . Recently , a work in the parasitic nematode Ascaris suum showed that microRNAs are differentially expressed between males and females [32] . Although the differences were small and the targeting properties of male and female microRNAs similar , they reported a tendency of male microRNA to target extracellular proteins [32] . Another work in the bird Taeniopygia guttata ( zebra finch ) , suggests that the male-biased expressed microRNA mir-2954 specifically target genes in the Z chromosome , and may be involved in sexual dimorphism in song behavior [33] . Together , these papers and our work point to a general mechanism of microRNA-modulation of sex-specific gene expression . A recent work shows that some microRNAs are specifically expressed in Schistosoma japonicum eggs [21] . In that work , the authors also measured the microRNA expression levels in males and females . We reanalyse their data and find that two out of our three male-biased microRNAs ( mir-1 and mir-61 ) have a consistent bias in Schistosoma japonicum , while the third is not present in their dataset . Also , five of our female-biased microRNAs also showed a female bias in their work ( mir-71b , mir-2c , mir-2d , bantam and mir-31 ) . These findings further validate our results and show that the sex-biased pattern of microRNA expression is evolutionarily conserved between these two species . Sex-biased gene expression affects the genetic composition of chromosomes , since selection has different effects on sex-biased genes depending on whether they are located on sex chromosomes or on autosomes ( reviewed in [34] ) . Likewise , sex chromosomes have distinctive evolutionary patterns , which affect the genes encoded within [35] . We may therefore expect to see signatures of chromosome evolution in sex-biased microRNAs . Indeed , we observed that microRNAs that are female sex-biased are depleted in the sexual chromosomes ( Figure 2B ) . This may indicate a loss of sex-biased genes at the sex chromosomes . One interesting example is the female-biased mir-71b/2f/2c/2d autosomal microRNA cluster , which has a paralogous copy in the sex chromosome with no biased expression . If a microRNA gene that is selectively advantageous for females becomes part of a sex chromosome ( by sexualisation of the chromosome , or otherwise ) , selection over this gene will be less efficient in the heterogametic sex ( females in our case ) , generating a conflict between expression pattern and chromosomal location . Duplication of a gene into an autosome has been recently proposed as a way to escape such conflict ( reviewed in [36] ) , and the mir-71/mir-2 cluster duplication appears to be an example of this . Although a more comprehensive analysis of duplicated microRNAs with sex-biased expression is required to confirm this , our analysis shows that sex-biased expressed microRNAs have an impact in shaping the genome during evolution . The study of microRNAs in the schistosomes is of both evolutionary and biomedical interest . First , the recent evolution of a sexual reproductive system from a hermaphrodite ancestor can give clues about how sexuality emerged in other species . Our data are consistent with the acquisition of sex-biased expression of conserved microRNAs soon after the species become dioecious . Second , the characterization of Schistosoma-specific microRNAs may provide new targets for infection control . In this work , we characterize for the first time 12 microRNAs conserved between S . mansoni and S . japonicum but not in other platyhelminthes ( nor in other animals ) . Two of these sequences also showed female-biased expression ( mir-8437 and mir-8447 ) . However , some important questions remain to be answered: Are microRNAs conserved in the two studied schistosomes also conserved in other Schistosoma species or in other trematodes ? Is sex-biased expression of microRNAs associated with sex-biased expression of their targets ? Do other dioecious flatworms have sex-biased expression of microRNAs ? The genomic sequencing of more platyhelminthes and characterization of their microRNAs will help us to answer these questions .
For the detection of expressed microRNAs , female mice ( BKW strain ) were infected with Belo-Horizonte strain Schistosoma mansoni parasites by paddling in water containing 200 cercariae . Seven weeks after infection , adult schistosomes were collected from the mice post-mortem by hepatic portal perfusion . RNA was extracted from adult schistosome samples with the miRVana miRNA isolation kit ( Ambion ) . We used two sequencing technologies , AB SOLiD and Illumina MiSeq , to sequence S . mansoni small RNA libraries . The extremely deep coverage provided by SOLiD sequencing provides high sensitivity for the discovery of novel microRNAs . We further used Illumina MiSeq sequencing of gender-specific libraries to compare the expression level of microRNAs between males and females . Library construction was performed as previous described [37] using the SOLiD Small RNA Expression Kit ( Ambion ) . SOLiD sequencing was performed at the Center for Genomic Research at the University of Liverpool . We obtained a total of 124 , 341 , 126 SOLiD sequence reads from two libraries . For the gender-specific differential expression of microRNAs , we prepared RNA libraries with the miRVana kit from separate male and female samples ( provided by Andrew MacDonald and Rinku Rajan at the University of Edinburgh ) , and prepared libraries for Illumina MiSeq sequencing according to the manufacturer's instructions . MiSeq samples were sequenced in the Genomics Core Facility at the University of Manchester . High-throughput datasets were deposited in Gene Expression Omnibus ( GEO ) at NCBI ( accession number: GSE49359 ) . Sequencing reads from male and female libraries were separately mapped to the S . mansoni reference genome ( assembly 5 . 1 available at http://www . genedb . org/Homepage/Smansoni; [38] , [39] ) with Bowtie 0 . 12 using the sequential trimming strategy implemented in SeqTrimMap 1 . 0 [40] allowing 2 mismatches . Sequences mapping to potential rRNAs or tRNAs were first removed . Putative tRNAs were predicted in the genome sequence with tRNAscan-SE using default parameters [41] and ribosomal RNAs ( rRNAs ) were extracted from the SILVA database ( http://www . arb-silva . de/ , release 108 ) . Mapped reads with a length of 19–25 nucleotides , matching five or fewer positions in the genome ( a total of 63 , 771 , 124 sequence reads ) , were used to detect microRNAs as previously described [37] , [40] . We further used BLAST [42] to search microRNA candidates against the S . mansoni genome and discarded those with more than 5 hits ( E-value <e-10 , 80% query coverage ) to remove potential repetitive elements: 44 candidate sequences did not pass this filter . MicroRNA candidates were manually inspected . Potential homologs of known microRNAs ( detected by BLASTN against all hairpin sequences from miRBase version 17 [43] ) with reads mapped from our datasets but which did not pass our criteria were also retained . Homolog of our microRNA candidates were predicted in the genome sequences of S . japonicum ( version 2 ) , S . mediterranea ( v . 3 . 1 ) , Caenorhabditis elegans ( v . 7 . 1 ) , Tribolium castaneum ( v . 3 . 0 ) , Drosophila melanogaster ( v . 5 . 1 ) , Homo sapiens ( v . 37 . 1 ) and Gallus gallus ( v . 2 . 1 ) , with parameters: −W 4 , −r +2 , −q −3 . Only sequences with a predicted hairpin structure and conserving at least one mature sequence were considered as putative homologs . We mapped reads produced from MiSeq sequencing reactions to our annotated S . mansoni microRNAs and discarded all reads that map to more than one microRNA locus . Read counts were transformed with the ‘upper quartile’ normalization using the edgeR package [44] , following the suggestions in [45] . Other normalization procedures ( ‘TMM’ , ‘LOWESS’ and ‘no normalization’ ) did not change the results ( Supplementary Files S3 ) . Fold changes in expression levels are given in logarithms in base 2 . We consider a dispersion of expression between experiments of 0 . 2 and a false discovery rate of 10% . With these parameters , microRNAs showing a two-fold difference in their expression levels are considered to be sex-biased , as routinely suggested [46] , [47] . The expression level in males and females of seven microRNAs ( mir-1b , mir-61 , mir-281 , mir-36b , mir-71b , bantam and mir-8437 ) were further validated by quantitative PCR . We used custom made TaqMan assays manufactured by Life Technologies . Fluorescent quantification was done in a Chromo 4 qPCR system ( BioRad ) for a log fluorescent threshold of 0 . 05 , using mir-36a ( the microRNA showing the least bias in the MiSeq experiments ) as the non-sex-biased control microRNA . For each amplification , we performed three technical replicates to estimate the significance of the observed differences . Laboratory animal use was within a designated facility regulated under the terms of the UK Animals ( Scientific Procedures ) Act , 1986 , complying with all requirements therein . The experiments involving mice in this study were approved by the Natural History Museum Ethical Review Process and work was carried out under Home Office project licence 70/6834 .
|
Schistosomiasis is the second most common disease caused by a parasite , affecting over 200 million people . The parasites involved are flatworms of the genus Schistosoma . Unlike most non-parasitic flatworms , Schistosoma species have separate sexes , and the emergence of sex has been associated with the development of a parasitic lifestyle . The identification of gene products that are expressed in a sex-biased fashion permits the study of the origin of sexual dimorphism and , in the case of the schistosomes , the evolution of a human parasite . Here we investigated the differential expression of microRNAs in male and female individuals of the species Schistosoma mansoni . MicroRNAs are crucial gene regulators . We observed that many new microRNAs emerged in the evolutionary lineage leading to the schistosomes . However , many sex-biased microRNAs were present in the hermaphrodite ancestor of the flatworms , and therefore acquired sex-biased expression later on . Our results suggest that changes in microRNA expression patterns were associated with the emergence of separate sexes in the schistosomes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2013
|
Sex-Biased Expression of MicroRNAs in Schistosoma mansoni
|
Aedes albopictus is one of the most invasive human disease vectors . Its control has been largely based on insecticides , such as the larvicide temephos . Temephos resistance has been associated with the up-regulation , through gene amplification , of two carboxylesterase ( CCE ) genes closely linked on the genome , capable of sequestering and metabolizing temephos oxon , the activated form of temephos . Here , we investigated the occurrence , geographical distribution and origin of the CCE amplicon in Ae . albopictus populations from several geographical regions worldwide . The haplotypic diversity at the CCEae3a locus revealed high polymorphism , while phylogenetic analysis showed an absence of correlation between haplotype similarity and geographic origin . Two types of esterase amplifications were found , in two locations only ( Athens and Florida ) : one , previously described , results in the amplification of both CCEae3a and CCEae6a; the second is being described for the first time and results in the amplification of CCEae3a only . The two amplification events are independent , as confirmed by sequence analysis . All individuals from Athens and Florida carrying the CCEae3a-CCEae6a co-amplicon share a common haplotype , indicating a single amplification event , which spread between the two countries . The importance of passive transportation of disease vectors , including individuals carrying resistance mechanisms , is discussed in the light of efficient and sustainable vector control strategies .
Aedes albopictus is vector of several important arboviruses , and has recently become a major threat to human health [1 , 2] . Its role in several disease outbreaks has been documented , for example dengue in Hawaii ( 2001–02 ) , Gabon ( 2007 ) and Japan ( 2014 ) [3–5]; chikungunya in Italy ( 2007 ) and Reunion Island ( 2005–06 ) [6 , 7]; Zika in Gabon ( 2007 ) [8] . This mosquito species has also drawn attention by being one of the most invasive human disease vectors worldwide . It originated from India and South-East Asia , and quickly invaded almost all continents [9] . Its presence was first reported in Europe ( Albania ) in 1979 [10] , in USA ( Texas ) in 1985 [11] and in Africa ( Cape Town ) in 1989 [12] . The successful spread of Ae . albopictus throughout the globe has been facilitated by human activities , like trade of tires and other goods , as well as tourism [2]; these indeed allow transportation of desiccated eggs and larvae to new places . Its ability to adapt to different environments has also been associated with specific biological traits , such as the production of diapausing eggs , which enables its survival at cooler temperatures [9] . The control of Ae . albopictus is largely based on habitat management campaigns [13] , repellents ( spatial or personal ) and insecticides ( larvicides and adulticides ) [14] , while the use of alternative means such as Wolbachia , Sterile Insect Techniques and genetic manipulation approaches are also currently being investigated [15 , 16] . Among a limited number of mosquito larvicides ( including bacterial toxins and insect growth regulators—IGRs ) , temephos is an organophosphate ( OP ) that has been used extensively for the control of Aedes mosquitoes ( Ae . aegypti and the often sympatric Ae . albopictus ) in several continents and countries [17] . However , resistance against this insecticide has been reported [18 , 19] . We recently showed [20] that resistance against temephos in an Ae . albopictus population from Greece is associated with the upregulation , through gene amplification ( i . e . multiple gene copies ) of two carboxylesterase genes ( CCEs ) , namely CCEae3a and CCEae6a , which are closely located on the genome [21] . Notably , the same orthologous genes ( CCEae3a and CCEae6a ) have also been associated with temephos resistance in Ae . aegypti , the primary dengue and yellow fever vector worldwide [22] . CCEae3a protein was shown to be localized in malpighian tubules ( MT ) and nerve tissues of the Ae . albopictus larvae , as well as being able to sequester and metabolize temephos oxon , the activated form of temephos [23] . OP resistance based on sequestration and enhanced metabolism resulting from CCE gene amplifications has also been described in other insects and mosquito species , such as the aphid Myzus persicae [24] and members of the mosquito Culex pipiens complex [25 , 26] . At the world scale , only a few CCE genes have been recorded amplified in cases of resistance in insects , which tends to indicate that advantageous mutations could be limiting . Although the mechanism by which esterase genes are amplified has not been established yet , it has been suggested that certain genome regions are probably a "hot spot" for recombination and amplification . Regions showing homology with repetitive elements have been found in DNA flanking the amplified CCEs , suggesting that these may have a role in the amplification process , as they may be functionally related to transposable elements [27] . Resistance to OPs in Culex mosquitoes has been shown to occur via over-expression , through gene amplification , of two esterase loci , Est-2 and Est-3 , which may be amplified singly ( e . g . the estβ1 gene ) or more commonly are co-amplified as allelic pairs in resistant mosquitoes [27 , 28] . However , the amplified alleles can differ , which indicates that the amplification process happened several times independently . These amplified CCE alleles have been described in different geographical places . Some of them remained localized in a relatively limited area and appeared as independent events . Others have spread to distant regions from a single evolutionary origin; the same common haplotypes are indeed found in mosquitoes from different continents [29–32] . It appears that once amplification has occurred , it can easily reach other geographic areas by migration , and then invade thanks to local insecticide selection [33] . For example , the worldwide most common allele is Ester2 ( or estα2-estβ2 co-amplicon ) , which occurs in >80% of insecticide resistant strains [34] , suggesting that it may confer higher fitness than other allelic variants [33 , 35] . The distribution and origin of amplified CCEs associated with insecticide resistance has not been studied in Aedes mosquitoes . Here , we investigated the occurrence , frequency and geographical distribution , as well as the phylogenetic relationship and origin of the CCEae3a-CCEae6a amplicon ( s ) /loci in Ae . albopictus populations from 16 different places across the globe .
Ae . albopictus field mosquitoes used in this study were collected from Mexico ( Apocada , Reynosa and Tapachula ) , U . S . A ( Florida and Atlanta ) , Brazil ( Rio de Janeiro ) , Belize ( Orange walk town ) , Gabon ( Franceville , Cocobeach , Lope ) , Switzerland ( Ticino ) , France ( Montpellier ) , Italy ( Lombardy ) , Greece ( Agios Stefanos , Koronida ) , Taiwan ( Taipei ) , China ( Beijing ) , Sri Lanka ( Peradeniya ) , Australia ( Hammond ) , Bangladesh ( Panchagarh ) , Lebanon ( Beirut ) and Japan ( Tokyo ) . In addition individuals from two laboratory colonies were used: i ) the Tem-GR strain , derived from an Ae . albopictus population collected in 2010 in Athens ( Greece ) and selected with temephos using standard WHO larval bioassays [20] , and ii ) the Malaysia-Lab strain , a susceptible laboratory strain originally collected in Malaysia [36] . Ae . albopictus adults or larvae stored in ethanol were first dried , and then genomic DNA was extracted from each individual using the Cethyl Trymethil Ammonium Bromide ( CTAB ) method described in Navajas et al . [37] . The DNA pellet was dissolved in 20μl of sterile water . Individuals were identified to species based on a species ID PCR [38] . In each PCR reaction , reference Ae . albopictus and Ae . aegypti samples were used as controls . CCEae3a and CCEae6a gene copy number variation ( CNV ) was assessed using quantitative PCR ( qPCR ) on individual Ae . albopictus specimens ( S1 Table ) . Amplification reactions ( 25μl final volume ) were performed on a MiniOpticon Two-Color Real-Time PCR Detection System ( BioRad ) using 2μl of genomic DNA ( diluted 5 times ) , 0 . 4μM primers ( two different primer pairs per target gene ) ( S2 Table ) and Kapa SYBR FAST qPCR Master Mix ( Kapa-Biosystems ) . Two housekeeping genes , histone3 ( NCBI: XM_019696438 . 1 ) and the ribosomal protein L34 ( NCBI: XM_019677758 . 1 ) , were used as reference genes for normalization[39] . Fivefold dilution series of pooled genomic DNA from the temephos susceptible Malaysia-Lab strain and the temephos selected TemGR strain were used to assess the efficiency of the qPCR reaction for each gene specific primer pair . A no-template control ( NTC ) was included to detect possible contamination and a melting curve analysis was performed to check the presence of a unique PCR product . Differences in CCEae3a and CCEae6a gene copy numbers were estimated relative to the temephos susceptible Malaysia-Lab strain , following Pfaffl [40] . CCEae3a ( Vector base , AALF007796 ) is predicted to encompass three exons and two introns . To identify the most variable part of the gene the full intron1 was amplified using forward primer 5’-ACGGTCCTCGATACATAGTG-3’ and reverse primer 5’-TAGCCTCATTGCTGGTTAGC-3’ ( hybridizing respectively at the end of exon1 and at the beginning of exon2 ) and the full intron2 was amplified using forward primer 5’-AGAGTGCGTTACGGATCAAG-3’ and reverse primer 5’-CACTGGCTTCCAGGAGATAC-3’ ( hybridizing respectively at the end of exon2 and at the beginning of exon3 ) . The PCR reactions ( 25μl final volume ) were performed using 2μl genomic DNA from individual Ae . albopictus mosquitoes , 0 . 4μM primers , 0 . 2mM dNTPs , 5μl of 10X buffer and 1U of Kapa Taq DNA Polymerase ( KAPABIOSYSTEMS ) . The PCR conditions were 95°C for 5min followed by 29 cycles of 94°C for 30sec , 48°C for 30sec , 72°C for 1min and a final extension of 72°C for 10min . PCR products were purified using a PCR purification kit ( Macherey Nagel ) and sent for sequencing using the forward primer ( Macrogen Sequencing Facility , Amsterdam ) . To assess the diversity of CCEae3a , the 709bp fragment of the gene ( including the last 314bp of exon1 , the whole intron1 and the first 192bp of exon2 ) was sequenced . PCR products from homozygous individuals were sequenced directly using the forward primer ( 5’-ACGGTCCTCGATACATAGTG-3’ ) ; for heterozygotes , the PCR products were cloned using the pGEM-Teasy vector ( Promega ) according to manufacturer’s instructions to separate the different alleles , and six clones for each individual were sent for sequencing ( Macrogen sequencing facility , Amsterdam ) , with the T7 universal primer . Sequences were examined and aligned using the BioEdit software . Phylogenetic relationships between the different CCEae3a haplotype sequences were determined using the Phylogeny . fr platform ( ‘‘one click mode” ) [41] . Briefly , sequences were aligned using the MUSCLE 3 . 8 . 31 algorithm , and alignment was then refined using the Gblocks 0 . 91b software to exclude poorly aligned parts . Subsequently the PhyML 3 . 1/3 . 0 ( aRLT ) software was used to assess the clade support , by computing the maximum likelihood tree and aLRT test ( approximate Likelihood Ratio Test ) [42] . Finally the tree was drawn using the TreeDyn 198 . 3 software [43] . The work described in this manuscript is in no way linked ( directly or indirectly ) to ethical concerns . No data from humans have been collected . All research activities respect fundamental ethics principles , including those reflected in the Charter of Fundamental Rights of the European Union ( 2000/C 364/01 ) . The work is compatible with EU and international law , as a number of entomological monitoring activities ( and transport of gDNA in Ethanol ) is contacted worldwide and in Europe ( European Mosquito Control Association , http://www . emca-online . eu ) .
A total of 385 mosquitoes from 16 countries and 22 different collection sites ( Fig 1 and S1 Table ) were first confirmed to be Ae . albopictus ( species ID PCR ) and subsequently tested for CCEae3a and CCEae6a CNV via qPCR . Out of 35 individuals tested from Florida , three showed amplification of both CCEae3a and CCEae6a ( Florida 5 , 21 and 28 ) , while four showed amplification of CCEae3a only ( Florida 9 , 24 , 26 and 35 ) ( Fig 2 ) . Amplification of both esterases was also detected in Greece: two individuals out of 10 from Agios Stefanos ( Ag . stef 1 , 2 ) , and four out of 10 from Koronida ( Koronida 1 , 8 , 9 , 10 ) . Amplification of both esterases was also confirmed in all tested individuals of the temephos resistant TemGR strain ( Fig 2 ) . None of the 330 individuals tested from the remaining 14 countries showed amplification of CCEae3a or CCEae6a ( S1 File ) . The two predicted introns of the CCEae3a locus were sequenced using individuals from the TemGR and Malaysia-Lab strains . Intron1 sequences were longer and more variable between individuals from these two strains , while intron2 sequences were identical . A 709 bp region including part of exon1 , the whole intron1 and part of exon2 ( Fig 3 ) was thus used to examine the haplotype diversity between individuals from the sampled countries , with and without CCE amplification ( 1–14 individuals per collection site , S1 Table ) . Sequence alignment revealed several SNPs throughout the amplified region , both in introns and exons , plus some insertions and deletions in the intronic sequence ( S2 File ) . The diversity was higher in the intron sequences ( mean distance = 0 . 041 substitution/site ) than in the exons ( 0 . 012 and 0 . 014 for exon1 and exon2 , respectively ) [44] . A total of 45 different haplotypes , differing by at least one mutation , were identified from the 49 individuals tested ( S2 File ) . Haplotypes did not cluster based on geographic proximity . While most haplotypes were found in a single individual , some sequences from individuals collected in distant geographic locations were indeed identical: for example H3 was present in individuals from Atlanta , Italy , Greece ( Agios Stefanos ) , Florida and Mexico ( Apocada ) , H7 in individuals from Belize , Italy , Florida and Switzerland , H9 in individuals from Belize , Florida and Mexico ( Tapachula ) , H12 in individuals from China and Lebanon , and H13 in individuals from China , Lebanon and Mexico ( Reynosa ) . Moreover , sequences obtained from individuals collected in the same area often showed a great variability and were found in different clades in the tree , to the point that two sequences found in a single heterozygous individual could be quite distant ( e . g . Bangladesh4A and B or Brazil14A and B ) . In particular , haplotypes obtained from individuals with no esterase amplification originating from Florida and Greece were dispersed throughout the phylogenetic tree ( Fig 4 ) , clustering with haplotypes from distant areas . In contrast , all individuals displaying amplified esterases clustered in only two highly supported clades . In the first , all the individuals from Florida ( U . S . A ) showing amplification of only CCEae3a shared a common haplotype ( H29 ) , which was found in no other individual in the dataset . The second clade clustered all the individuals showing amplification of both CCEae3a and CCEae6a , whether from Florida ( U . S . A ) or from Greece ( Agios Stephanos and Koronida ) , including the reference strain TemGR . They also shared a common haplotype ( H30 ) , which again was found in no other individual in the dataset . In addition , this second clade was closer to haplotypes obtained from individuals without esterase amplification ( e . g Brazil 14B and Taiwan 1B ) than to the first clade ( i . e . individuals with amplification of CCEae3a only ) .
Two carboxylesterase genes ( CCEs ) , CCEae3a and CCEae6a have recently been shown in Ae . albopictus to be implicated in OP resistance through gene amplification [20] . To understand the origin and spread of these resistance alleles , we assessed the haplotypic diversity at the CCEae3a locus . Analysis revealed that this gene is polymorphic in non-amplified alleles ( S2 File ) : 45 different haplotypes were identified in 49 individuals collected around the globe , with only five ( H3 , H7 , H9 , H12 and H13 ) shared between two or more individuals . Moreover , for these five haplotypes , the individuals came from distant areas , on different continents . The phylogenetic tree further confirmed the absence of correlation between haplotype similarity and geographic origin , as individuals from the same collection area clustered mostly in different clades of the tree ( this was even observed for the two haplotypes of a single heterozygote ) . This observation echoes previous studies on mitochondrial genes , microsatellites and other nuclear genetic markers [45–47] that showed that Ae . albopictus populations have been repeatedly transported from their original range ( South-East Asia ) to different areas around the globe; progenies of mosquitoes originally from the same locality can thus be found in different continents . The frequent exchange of goods at an international level and travelling of people around the world indeed facilitate the passive transportation of mosquitoes , which are often found in aircrafts and ships [48] . These studies also showed that the non-Asiatic populations often result from a mix of several independent invasions . This mechanism , promoting genetic diversity , is often proposed as a key factor contributing to successful establishment of a species in new areas [45 , 49] . We then addressed the worldwide occurrence , distribution and diversity of esterase amplification in Ae . albopictus populations in response to OP selection: our study evidenced that at least two types of esterase amplifications are currently segregating . The first one , resulting in amplification of both CCEae3a and CCEae6a , had been previously described [20] , while the second; resulting in amplification of CCEae3a only , is being described for the first time . In contrast to the high CCEae3a haplotypic diversity in non-amplified alleles , the two amplified genotypes associated with OP resistance displayed no diversity . It also revealed that the two amplification events were independent: individuals with CCEae3a and CCEae6a co-amplification displayed identical sequences and thus clustered together , far apart from individuals with CCEae3a-only amplification , which also displayed identical sequences . The appearance of independent amplification events at the same genomic region suggests the presence of favoring features , which promote unequal crossing-overs and/or transposition [27] . For example , a repetitive element Juan ( possibly related to transposable elements ) was found close to the amplified esterase locus in Cx . quinquefasciatus [50] . The Ae . albopictus genome is also known to carry many transposable elements , and 68% of its genome is occupied by repetitive sequences [21] . If one of those is close to the CCEae3a locus , it could facilitate its repeated and independent duplications . In addition , these mechanisms might also act after the first amplification event resulting in further variation in copy numbers . This has been hypothesized in Cx . pipiens [51] and might also explain the differences observed in the relative gene copy numbers among individuals from Florida and Greece . The fact that all individuals carrying co-amplifications of both CCEae3a and CCEae6a , from Greece and Florida U . S . A ( two regions hosting many millions of tourists every year ) , share a common haplotype reveals that a single amplification event took place , and then spread between the two countries . This again outlines the importance of passive transportation of disease vectors , including individuals carrying resistance mechanisms , which unfortunately promotes resistance spread at the world scale , as has been described in Cx . pipiens mosquitoes with OP-resistant amplified esterases [30 , 52] . However , the establishment of a resistance mechanism in a new area largely depends on the local selective advantage it offers in these new environmental conditions [33] . The repetitive use of temephos [53] or other organophosphate insecticides , which could show cross resistance , like the adulticide naled commonly used in Florida [54] probably facilitated the establishment of CCEae3a amplified haplotypes in Ae . albopictus populations from both countries . It also ensured that these haplotypes were selected and reached high frequencies . It is actually surprising that the resistance allele found in individuals from the TemGR strain , originally collected in 2010 , is still present in Greece 2016 collection: temephos has been officially banned in Europe in 2007 , and resistance mechanisms are usually ( but not always ) associated with fitness costs [55 , 56] . The persistence of the resistant allele throughout these years suggests either the lack of a significant fitness cost for individuals carrying them or the presence of a current selection source , either from non officially approved vector control activities , or from other substances ( e . g . originating for example from agriculture ) [57] . In any case , the presence of two independently amplified OP-resistant alleles in Ae . albopictus already segregating in distant places in the world ( i . e . Athens—Greece and Florida–U . S . A ) raises concerns about the future control of this species . Although the levels of insecticide resistance in Ae . albopictus are low at present ( i . e . resistance mechanisms/alleles are less frequent ) compared to Ae . aegypti [18 , 58] , our study shows that resistance can be selected and spread rapidly around the globe also in this species and possibly compromise control activities . As only a limited number of mosquito larvicides are available on the market , temephos resistance is an important consideration for many countries , where this active ingredient is still in use . Moreover , striking resistance mutations have been found in other insect species ( Plutella xylostella ) that completely inactivate the IGR diflubenzuron , now one of the most important mosquito larvicides in Europe and other regions [59] . This further raises serious concerns for regions that have banned the use of temephos , such as Europe , as it is certainly a potential reliable resource for emergency epidemics , new invasion cases , or lack of alternative efficient mosquito control solutions .
|
Control of mosquito borne diseases is being seriously challenged by the ongoing development of insecticide resistance . Resistance of Aedes albopictus , a major arbovirus vector , to the organophosphate larvicide temephos was recently associated with the up-regulation , through gene amplification , of two carboxylesterases; CCEae3a and CCEae6a . Here we investigated the worldwide distribution and origin of the amplified esterases , which is of great value for designing and implementing efficient vector control programs . Individuals with amplification of both esterases were found in Greece and Florida ( U . S . A ) , representing a single amplification event that spread between the two countries , highlighting the importance of passive transportation of disease vectors carrying resistance mechanisms , which is mainly facilitated by human activities . In addition , individuals with amplification of the CCEae3a only , but not the CCEae6a , representing a second and independent amplification event were found in Florida . The worldwide haplotypic diversity obtained for CCEae3a is consistent with the highly invasive nature of the Aedes albopictus .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"taxonomy",
"invertebrates",
"united",
"states",
"medicine",
"and",
"health",
"sciences",
"enzymes",
"geographical",
"locations",
"enzymology",
"animals",
"genetic",
"mapping",
"north",
"america",
"phylogenetics",
"data",
"management",
"phylogenetic",
"analysis",
"molecular",
"biology",
"techniques",
"insect",
"vectors",
"research",
"and",
"analysis",
"methods",
"florida",
"infectious",
"diseases",
"computer",
"and",
"information",
"sciences",
"agrochemicals",
"artificial",
"gene",
"amplification",
"and",
"extension",
"proteins",
"evolutionary",
"systematics",
"molecular",
"biology",
"disease",
"vectors",
"insects",
"agriculture",
"arthropoda",
"insecticides",
"people",
"and",
"places",
"biochemistry",
"mosquitoes",
"haplotypes",
"hydrolases",
"esterases",
"polymerase",
"chain",
"reaction",
"heredity",
"genetics",
"biology",
"and",
"life",
"sciences",
"species",
"interactions",
"evolutionary",
"biology",
"gene",
"amplification",
"organisms"
] |
2017
|
Carboxylesterase gene amplifications associated with insecticide resistance in Aedes albopictus: Geographical distribution and evolutionary origin
|
During replication , mismatch repair proteins recognize and repair mispaired bases that escape the proofreading activity of DNA polymerase . In this work , we tested the model that the eukaryotic mismatch recognition complex tracks with the advancing replisome . Using yeast , we examined the dynamics during replication of the leading strand polymerase Polε using Pol2 and the eukaryotic mismatch recognition complex using Msh2 , the invariant protein involved in mismatch recognition . Specifically , we synchronized cells and processed samples using chromatin immunoprecipitation combined with custom DNA tiling arrays ( ChIP-chip ) . The Polε signal was not detectable in G1 , but was observed at active origins and replicating DNA throughout S-phase . The Polε signal provided the resolution to track origin firing timing and efficiencies as well as replisome progression rates . By detecting Polε and Msh2 dynamics within the same strain , we established that the mismatch recognition complex binds origins and spreads to adjacent regions with the replisome . In mismatch repair defective PCNA mutants , we observed that Msh2 binds to regions of replicating DNA , but the distribution and dynamics are altered , suggesting that PCNA is not the sole determinant for the mismatch recognition complex association with replicating regions , but may influence the dynamics of movement . Using biochemical and genomic methods , we provide evidence that both MutS complexes are in the vicinity of the replisome to efficiently repair the entire spectrum of mutations during replication . Our data supports the model that the proximity of MutSα/β to the replisome for the efficient repair of the newly synthesized strand before chromatin reassembles .
During cell division , accurate DNA replication is essential to preserve the integrity of the genome and defects in this process result in diseases including hereditary and sporadic cancers [1] . In eukaryotes , the replicative DNA polymerases , Polε and Polδ , perform leading and lagging strand synthesis respectively [2–5] . The proofreading function of the polymerases combined with the recognition and repair of mismatches ensures faithful transmission of genetic information during each round of replication . The errors generated during replication include single base mismatches , single nucleotide insertion/deletion loops ( indels ) at microsatellites ( MS ) [reviewed in 6] . Microsatellites are repeat regions of 1–10 bp repeat units , which frequently undergo expansion and contraction due to slippage of the polymerases during replication [7] . In prokaryotes , homodimeric MutS binds the full range of mismatches [reviewed in 6] . In eukaryotes , MutS complexes are heterodimers with differing mismatch recognition capabilities . MutSα ( Msh2/Msh6 ) recognizes single base mismatches and single nucleotide indels at homopolymeric runs , and MutSβ ( Msh2/Msh3 ) complex recognizes single nucleotide and larger indels [reviewed in 6] . MutSβ is also able to recognize certain base-base mismatches [8] . The ability of the mismatch repair ( MMR ) machinery to recognize the range of mismatches and target the newly synthesized , error-containing strand for repair is critical for maintaining fidelity during DNA replication . The method of strand discrimination during mismatch repair in most prokaryotes and all eukaryotes appears to require discontinuities in the DNA backbone ( nicks ) and the replication sliding clamp , known as β clamp in prokaryotes or Proliferating Cell Nuclear Antigen , PCNA , in eukaryotes . In vitro experiments using cell extracts demonstrated that a nick is sufficient to direct repair to the strand containing the discontinuity [9 , 10] . During DNA replication , the lagging strand has nicks ~200 bp apart [reviewed in 5]; whereas , the continuously synthesized leading strand may have long stretches without replication generated nicks [4] . However , during the replication process ribonucleotides ( rNMP ) are occasionally incorporated into the DNA molecule and are then cleaved by RNAase H2 [11–13] , thereby increasing the density of nicks during synthesis [14 , 15] . Because removal of RNAase H2 only causes a modest increase in mutation rates [14] , it remains a possibility that the 3’-OH of the leading strand is the primary strand specificity signal . In addition to nicks , the replication sliding clamp has been implicated in strand discrimination . In eukaryotes , PCNA was shown to interact with MutSα/β mismatch recognition complexes [16–18] . It is postulated that the orientation specific association of PCNA with the DNA helix positions mismatch repair proteins to cleave the newly synthesized nicked strand rather than the template strand [19–22] . Taking into consideration the complex nature of the in vivo DNA environment during replication , it is important to note that the newly replicated DNA is thought to quickly re-assemble into nucleosomes behind the replisome [23] after which , a mismatch and the nicks are presumably less accessible to the MMR proteins . This potential for diminished accessibility is based on the fact that nucleosomes without replication/repair associated histone modifications [24] and other DNA bound proteins can block movement of MutS complexes along DNA [25 , 26] . Taken together , the most efficient mechanism for detecting mismatches and for accessing the strand specificity signal would involve a close association between the mismatch recognition complexes and the replisome within the region where chromatin has been cleared . Current data are consistent with the mismatch recognition complexes localizing to the replisome . Mass spectrometry analyses of human proteins at active replication forks , have identified MutS homologues [27] . In yeast , live cell-imaging demonstrated co-localization of MMR complexes and replisome components during S phase [28] . Additionally , a temporal coupling of MMR expression during S-phase and MMR efficiency has also been demonstrated [29] . Finally , as mentioned above , the eukaryotic and prokaryotic mismatch recognition proteins associate with the replication sliding clamps [30–32] . Taken together , the data support the model that the mismatch recognition proteins are associated with the replication machinery during S phase; however , whether the MMR recognition complexes track with the advancing replisome had not yet been demonstrated . The data presented in this work are consistent with the model that both MutSα and MutSβ track with the replisome during replication to efficiently scan protein-free DNA for the entire spectrum of errors and readily access the strand specificity signals in the form of proximal nicks in the DNA generated during replication .
To determine if the mismatch recognition complexes track with the replisome , we first needed suitable controls to define the replication origins and to indicate the position of the advancing replisome during DNA replication . The minichromosome maintenance ( Mcm ) 2–7 helicase is a well-established predictor of potential origins of replication [33] . The Mcm 2–7 helicase is a component of pre-replication complexes that associate with origins during the G1 phase of the cell cycle [34] . We employed a hemagglutinin ( HA ) tagged Mcm4 , a subunit of the replicative helicase , to indicate potential replication origins . Additionally , in a separate strain the leading strand polymerase served as the control for replisome progression . Specifically , we used a HA tagged version of Pol2 , the catalytic subunit of Polε . We performed chromatin immunoprecipitations ( ChIP ) to detect what portions of the genome were associated with the replication proteins in G1 and throughout S-phase . All ChIP experiments included an untagged control for non-specific precipitation of certain DNA regions . This allowed for exclusion of regions of the genome that generate high background signal; for example , highly transcribed regions have a tendency to give a false positive signal in ChIP experiments [35] . Experiments were performed at 18°C to slow the replication process and improve the resolution of the signal for the advancing replisome . Initial arrest and progression through the cell cycle were monitored by determining the DNA content per cell using flow cytometry . One time point in G1 and six time points in S phase were subsequently processed for ChIP ( S1 and S2 Figs ) . ChIP samples were labeled and hybridized to custom DNA tiling arrays . The arrays included 65 of the ~500 origins of replication in the yeast genome ( S1 Table ) [36 , 37] . By measuring the peak corresponding to Mcm4 binding , we were able to mark the specific coordinates of the origins ( S4 Fig ) . We found that Mcm4 binding to potential origins is in agreement with previous studies and annotations ( S1 Table ) . In addition , we showed that the highly reproducible Polε signal throughout S-phase functions as a good metric for replisome progression ( Fig 1 ) . Supplement S6 illustrates the reproducibility of the Mcm4 and Polε signals across multiple experimental trials . These data establish that chromatin immunoprecipitation in combination with DNA tiling arrays is an effective method for tracking the leading strand polymerase at origins of replication during DNA synthesis . A more detailed analysis of Polε progression during S-phase is presented in the following sections . In G1 synchronized cells , all potential origins were detected by Mcm4 binding ( Fig 1; S1 Table ) , consistent with previous studies [33] . We determined that the Polε signal appears at active origins only after release from G1 . For example , ARS301 , ARS303 and ARS320 are inactive origins [38] and are bound by Mcm4 , but not by Polε ( Fig 1A ) . In contrast , ARS305 and ARS306 are known active origins [39–41] and exhibit Mcm4 as well as Polε binding ( Fig 1A ) ; however , the Polε is evident only after release from G1 . This finding is consistent previous studies using ChIP-PCR that detected Polε at origins during S-phase , but not at G1 of the cell cycle [34] . We find that the Polε signal appears at origins consistent with known firing times . Active origins of replication are known to fire at different times ( early , middle and late ) during S-phase [42] . A representative example of the differential timing is depicted in Fig 1C , where Polε binds the known late firing ARS609 at a later time than it does the known early firing , adjacent origins ARS607 and ARS608 [36] . The Polε signal typically diminishes at the origins after fork progression ( Fig 1A and 1E ) ; however , in some cases signal is observed at the origin at later time points ( Fig 1B and 1D ) . One explanation for this signal is lack of synchrony . Alternatively , this may be a consequence of some replication origins firing with less precision during the cell cycle [reviewed in 42 , 43 , 44] . We favor the second explanation because there are examples from the same experiment where at certain origins the signal diminishes ( e . g . at ARS305 and ARS306 in Fig 1A ) , suggesting synchrony , while at other origins the signal persists ( e . g . ARS315 , Fig 1B ) , consistent with less precision of firing of ARS315 during the cell cycle . The Polε signal observed is also in agreement with the known differences in firing efficiency of each origin [36] . ARS315 is highly efficient and fires in ~90% of each S-phase of the cell cycle [45] . In this study , ARS315 exhibits a robust Polε signal initially localized that then migrates away from the origin over time ( Fig 1B ) . Additionally , efficiently firing ARS607 ( fires >85% of the cell cycles ) [46] displays a particularly robust Polε signal at the origin; whereas the adjacent , less efficient ARS608 ( fires in <10% of the cell cycles ) [46] has a reduced signal ( Fig 1C ) . Finally , highly efficient origins such as ARS1207 and ARS1209 , exhibit a strong Polε signal ( Fig 1D ) . We observed that the Polε signal throughout S-phase is consistent with the advancing replisome kinetics . Specifically , Polε signal first appears at the origins and advances bi-directionally to adjacent regions as the cells progress through S-phase of the cell cycle ( Fig 1A–1E ) . The ~100 kb region on Chromosome IV represents a good example of Polε initially binding origins followed by the signal migrating to flanking regions up and downstream in subsequent time points ( Fig 1E ) . By measuring the leading edges of the replisome signal , the average rate of replication fork progression was calculated as ~430 base pairs per minute ( S2 Table ) . Previous studies showed that replication fork rate may vary with temperature , nutrient availability or drug treatment ( summarized in Table 1 ) . For example , experiments were performed at room temperature and replication fork rate was determined to be 1 . 6 kb/min , consistent with a faster doubling time [47] . In summary , Polε signal is detected at active origins with expected timing and firing efficiencies . Additionally , the movement of the Polε signal is consistent with the advancing replisome during S phase . These experiments established the foundation for a comparative analysis of the eukaryotic mismatch recognition complexes during S phase . With the appropriate controls for origin position and for replication fork migration established , we next aimed to determine the dynamics of the mismatch recognition complexes during S phase . MutSα ( Msh2/Msh6 ) and MutSβ ( Msh2/Msh3 ) are the two mismatch recognition complexes in eukaryotes that function in post-replicative mismatch repair [6 , 48] . Since Msh2 is the invariable component of both complexes , we tagged Msh2 with the myc epitope , facilitating the detection of both MutSα and MutSβ complexes within the cell . We employed the methods described above to determine the mismatch recognition complex dynamics during S-phase and observed that the Msh2 signal is remarkably similar to Polε ( compare Figs 1 and 2 ) . Msh2 is observed at origins in S-phase , but not G1 , and the signal progresses away from the origins bi-directionally ( Fig 2A–2E ) . For example , the Msh2 signal originates at ARS305 and ARS306 and migrates bi-directionally from each origin ( Fig 2A ) . Additionally , while Mcm4 signal is at ARS301 , ARS303 , ARS304 and ARS320 these inactive origins do not exhibit Msh2 signal as was observed for the Polε signal ( Figs 1 and 2 ) . Taken together , the data show that Msh2 binds and moves bi-directionally away from active origins ( Mcm4 and Polε both bound ) , but does not bind inactive origins ( Mcm4 only ) . It is important to note that we would not expect to detect significant signal of MutS complexes at mismatches because mismatches are rare within any given population of replicating cells [49–53] . The Msh2 signal in these experiments represents DNA surveillance during replication and not mismatch binding . Similar to Polε ( Fig 1 ) , the time of appearance and intensity of the Msh2 signal is consistent with the expected timing and firing efficiencies of the origins ( Fig 2 ) . For example , the intensity and distribution of Msh2 signal at ARS315 is consistent with the efficient , early firing of this origin ( Fig 2B ) . In addition , for the early-efficient ARS607 origin , the Msh2 signal is robust compared to the early-inefficient ARS608 origin ( Fig 2C ) . Correspondingly , the early-efficient ARS1207 and ARS1209 origins exhibit an early Msh2 signal that progressively migrates bi-directionally away from both origins while the ARS1208 remains inactive ( Fig 2D ) as was observed for Polε ( Fig 1 ) . After fork progresses bi-directionally from the origins , some signal can be observed in the position of the origin . The absence of bi-directional movement from ARS1208 is consistent with a previous study showing no Orc or Mcm4 binding at this origin [33] . We observed one potential difference between the Msh2 and Polε signal in regions behind the advancing replisome . While the Polε signal clears as the replisome advances ( Fig 1 ) , there is still some Msh2 signal in the region behind the replication fork ( Fig 2 ) . The persistent Msh2 signal is significantly higher than is observed in the “no tag” control . In the following section we discuss the significance of this persistent signal . Because the initial analyses showed similarities and differences in the dynamics of the Polε and Msh2 during S phase , we wanted to determine how precisely the signals coincided by using a strain tagged for both proteins . The G1 and S-phase samples from a doubly-tagged strain were processed as described above except that half the sample was processed for Msh2 ChIP and the other half for Polε ChIP . We examined the occupancy of both Polε and Msh2 at all of the 65 origins on the tiling arrays . Representative images of ARS1012 , ARS1013 and ARS1407 are shown in Fig 3 . The Msh2 signal is in regions occupied by Polε at each corresponding time point ( Fig 3 ) . These data are consistent with the mismatch recognition complexes loading at origins with a timing similar to the leading strand polymerase and associating with the replisome throughout DNA replication . Interestingly , there is a persistent Msh2 signal localized in the region behind the advancing replisome at ARS315 ( Fig 4 ) . When examining 12 early firing origins , the Msh2 signals persist after Polε signal diminishes for 9 origins . The no tag control does not exhibit the signal and statistical analyses of replicates discussed below confirm that this persistent signal is significant . To determine the statistical significance of the co-incident signals of Msh2 and Polε , we employed Chipper Software [54] to assign p-values to the ChIP signals from the tiling arrays . We averaged three replicates for Mcm4 , Polε , Msh2 and the no tag control . The data are visualized as the negative of the log10 of the calculated p-values ( Fig 5A–5C ) . We examined the significance of the ChIP signals for all potential origins represented on the tiling arrays . Of the 65 putative origins on the arrays , 60 exhibited Mcm4 signal ( p values ranged from 10−5 to 10−40 ) . These 60 potential origins were used to calculate the occupancy by Polε and Msh2 . A total of 55 origins ( ~91% ) showed Polε and Msh2 signal during S-phase . Origins where no Polε or Msh2 signal is detected are origins that are not bound by Mcm4 or have previously been established as inefficient and firing only in a small percentage of each round of replication [39 , 40 , 46] . Importantly , the co-occupancy signal of Polε and Msh2 at origins and adjacent regions during replication is highly significant . The Polε p values ranged from 10−3 to 10−25 and Msh2 p values were from 10−1 to 10−10 . Fig 5A illustrates the significance of the signal observed for both Polε and Msh2 at ARS1207 and ARS1209 , whereas no significant signal is seen at the adjacent inactive ARS1208 . Additionally , ARS1213 and ARS416 display overlapping , highly significant signal from Polε and Msh2 ( Fig 5B and 5C ) . In summary , the Mcm4 signal is highly significant and is observed at each potential origin of replication . During S phase , Msh2 and Polε signal are co-incident in the majority of the active origins and flanking regions with high significance . These data confirm that Msh2 , the invariant member of the mismatch recognition complex , remains closely associated with the replisome throughout S-phase of the cell cycle . Having determined that the mismatch recognition complex is closely associated with the replisome throughout S phase , we wanted to explore what factors might influence MMR protein loading at origins as well as efficient scanning of the genome during replication . PCNA plays a critical role in MMR at multiple stages [16 , 17 , 55 , 56] . To determine whether PCNA mutants implicated in MMR alter the binding and movement of the MMR recognition complexes during S phase , we examined Msh2 and Polε dynamics in PCNA mismatch repair defective strains . Two missense variants of yeast PCNA ( Pol30 ) were previously reported to disrupt MMR , but not to alter replication significantly [56] . We reasoned that the “separation of function” variants would be good candidates for determining the role of PCNA in mismatch recognition complex dynamics during replication . We first utilized the pol30-201 separation of function mutant coding for Pol30C22Y in ChIP-chip experiments . Pol30C22Y confers a partial MMR defect; however , the Pol30C22Y variant still interacts with MutSα ( Msh2/Msh6 ) in vitro [56] . In this strain , the Polε signal appears normally distributed ( relative to strains expressing wild-type PCNA/Pol30 ) , suggesting that there is no effect on the processivity of the polymerase ( Fig 6A and 6B ) . Cell cycle progression is also unaffected in this mutant , further supporting the absence of a replication defect [56] . The Msh2 signal coincided with Polε signal in the presence of the MMR defective Pol30C22Y variant ( Fig 6 ) . This is in agreement with in vitro studies that show interaction is not fully disrupted between MutSα complexes and Pol30C22Y protein [56] . However , the Msh2 signal does seem reduced in some regions adjacent to the origins ( right side of ARS315 , Fig 6A ) , suggesting a potential defect in association . Previous work showed that the partial MMR defects caused by the separation of function variants Pol30C22Y and Pol30C81R are exacerbated by converting two conserved phenylalanine residues to alanines in the PCNA interacting region ( PIP box ) of Msh6 [56] . Additionally , strains expressing Pol30C81R have a partial MMR defect that is more severe than is seen in strains expressing Pol30C22Y [56] . Finally , in strains expressing Pol30C81R in combination with the Msh6 PIP box variant ( Msh6PIP ) , MutSα no longer associates with replication foci [28] . We engineered strains containing pol30-204 ( coding for Pol30C81R ) and the msh6-F33A , F34A PIP box mutation ( expressing Msh6PIP ) to analyze mismatch recognition complex dynamics during replication . Consistent with the finding that Pol30C81R does not affect replication , the engineered strain exhibited normal cell cycle progression ( S7 Fig ) . The ChIP-chip data shows some Msh2 signal in the vicinity of the replisome in the strain expressing Pol30C81R and Msh6PIP ( Fig 7A ) ; however the signal is not highly correlated with the Polε signal . Fig 7B and 7C show a comparative example of the typical signal for each protein in wild-type cells . Two explanations could account for the presence of Msh2 signal in a strain in which the interaction between PCNA and MutSα should be diminished . First , the mismatch recognition complex might have alternative mechanisms for loading at origins as has been described previously [28 , 57] . Second , the signal may be from MutSβ , which is known to be partially redundant with MutSα . The second explanation is in contrast to the studies showing that MutSβ does not co-localize with the leading strand polymerase during replication [28] . However , studies using human cell lines have also identified MutSβ at sites of active replication [27] . In the following sections , we address the issue as to whether both MutS complexes are needed for mismatch recognition and whether they are both found at origins during replication . To confirm on a genome-wide level that both MutS complexes are required for the full spectrum of mutations generated during replication , we performed mutation accumulation experiments followed by whole genome sequencing in strains lacking one of the components of the two mismatch recognition complexes ( Table 2 ) . Wild-type , msh2 , msh6 and msh3 knockout strains were propagated in rich medium ( YEPD ) for ~210 generations with bottlenecks every ~21 generations . Single isolates from each strain were propagated . We have included previously published [49] msh2 null ( msh2Δ ) and wild-type data normalized to 210 generations for comparison . In this previous analysis , we determined that mutation rate for DNA mismatch repair null strains was ~1 mutation per genome per generation , 225-fold higher than the wild-type rate and that the mutation spectra for mismatch repair defective cells included insertions/deletions at homopolymeric runs ( HPRs ) ( ~87% ) and at larger microsatellites ( ~6% ) , as well as transitions ( ~5% ) and transversions ( ~2% ) [49] . As was expected from the use of reporter constructs [58] mutation accumulation analysis of the msh3Δ strain revealed an increase in mutations at larger microsatellites at a rate comparable to a complete MMR knockout ( msh2Δ ) . No single base substitutions and only a few single base insertion/deletions at homopolymeric runs were observed ( Table 2 ) . These data are consistent with having a fully functional MutSα ( Msh2/Msh6 ) , because MutSα is capable of repairing single base substitutions and single nucleotide indels in the absence of Msh3 [48] . The msh6Δ strain , acquired 14 single base substitutions . This observed number is also comparable to the single base substitutions observed in the MMR knockout ( msh2Δ ) strain . Of the 14 mutations observed 12 were transitions while 2 were tranversions , similar to the ratio of transitions to transversions observed for MMR defective cells [49 and references therein] . The msh6Δ strain accumulated 3 insertion/deletions at homopolymeric runs , whereas mutations were not observed at larger microsatellites , consistent with the repair of larger indels being MutSβ specific . After ~210 generations , msh2Δ accumulated a large number of insertion/deletions at HPRs ( 177 ) relative to the single deletion of the binding partners . We observe only 5 and 3 insertion/deletions at HPRs in msh3Δ and msh6Δ respectively ( Table 2 ) . This underscores the functional redundancy of MutSα and MutSβ for repair at HPRs and is in agreement with previous genetic analyses showing that MutSα and MutSβ are redundant for the repair of homopolymeric runs [48] . Additionally , analyses in mammalian systems also demonstrate MutSα/MutSβ redundancy in repair of indels at homopolymeric runs [59–64] . In summary , using mutation accumulation assays we showed on a genome-wide level that Msh6 and Msh3 are fully redundant for repair of single-base indels at homopolymers and that each MutS complex is needed to repair the entire spectrum of mismatches generated during replication . Given this data , it is reasonable to conclude that both MutSα and MutSβ are needed at the replisome to capture all of the types of mismatches as they emerge . Although both MutS complexes are needed for the full spectrum of mismatches generated during replication , it is possible that MutSα was previously found to be the replisome associated mismatch recognition complex [28] because MutSα is more abundant . Previous studies have examined the levels of the MutS complexes in human [59 , 65] and yeast cells [66] . In yeast , high throughput abundance studies reported ~ 1 , 000 copies of Msh2 , ~5 , 000 of Msh6 and ~700 of Msh3 per cell [66] . The data suggest that MutSα accounts for a greater percentage of the MutS complexes present in the cell . However , the high throughput experiments in yeast required validation . We examined the relative abundance of the individual components of both mismatch recognition complexes , using western blot analysis . We engineered a strain in which all three proteins were tagged with an identical myc epitope . The fusions were engineered at the endogenous chromosomal positions using the native promoters . The tagged proteins were shown to be functional for mismatch repair in vivo . The molecular weights of Msh2 , Msh3 , and Msh6 are sufficiently distinct to resolve the proteins on a 7% acrylamide gel . Although there may be differences in the accessibility of the multiple epitopes in each protein , this method provides a more direct comparison of the relative abundance of singly tagged strains . The data reproducibly show that the relative abundance of Msh2:Msh6:Msh3 is 2:1:1 ( Fig 8A ) . This ratio is visualized by examining the protein extract from the strain in which all three proteins are identically tagged ( lane 2 , Fig 8A ) . The discrepancy with our results and the high throughput method may be due to differences in the method of visualizing protein levels , the epitope tag used , or because the high throughput measured the levels from singly tagged strains . In summary , we find that Msh2 is in a 2-fold excess of the Msh6 and Msh3 binding partners such that there could be equal levels of MutSα and MutSβ in the cell; however , the experiments do not prove that the complexes are actually formed , that they are functional or that they are properly localized . The data presented here and previously suggest that MutSα and MutSβ are needed to cover the full spectrum of mutations and the levels of the protein subunits suggest that the stoichiometry of the MutS complexes may be in balance; however it is still possible that MutSα is the major complex found at the replisome and that MutSβ only binds when larger indels form at the advancing fork . Additionally , some of the MutSβ complexes could be partitioned to function in other processes such recombination [67] . We therefore aimed to determine whether both complexes associate with the replisome during replication . The tracking expreriments presented above examined Msh2 dynamics and therefore do not allow for the determination of whether one or both mismatch recognition complexes are co-incident with the polymerase throughout S-phase . To determine if MutSα ( Msh6/Msh2 ) and MutSβ ( Msh3/Msh2 ) are both in the vicinity of the replisome throughout S-phase , we performed ChIP-chip time course experiments using strains with Pol2-HA tagged ( Polε ) and either Msh6-myc tagged ( MutSα ) or Msh3-myc tagged ( MutSβ ) . A simplified time course time course experiment was performed to examine MutSα or MutSβ binding during S-phase . Briefly , samples were taken at the time of arrest and two additional times during S phase of the cell cycle and processed for ChIP . The samples were analyzed with the custom tiling arrays for Msh6 and using quantitative PCR ( ChIP-PCR ) for Msh3 . Fig 8B shows an example of the data for MutSα during replication . As was observed for Msh2 , we observed binding of Msh6 in regions corresponding to Polε binding ( ~90% co-occupancy for 55 origins ) . Using qPCR to detect binding to ARS305 , an early firing origin with a robust signal ( Figs 1 and 2 ) . We observed that Msh3 is enriched at the origin with a signal similar to Msh2 ( Fig 8D ) . The Msh3 signal is significantly higher than is observed in the no-tag control . Additionally , as is seen routinely with Msh2 ChIP-chip ( Fig 2A ) and ChIP-qPCR ( Fig 8D ) , the Msh3 signal is not as strong as the Polε signal ( Fig 8C and 8D ) . While this ChIP-qPCR approach does not show what occurs across the genome , it does provide an example of a highly efficient , early origin with Msh3-myc signal . In summary , we provide data consistent with a hypothesis positing that both MutSα and MutSβ associate with the replisome to capture the entire spectrum of mismatches that escape DNA polymerase proofreading .
Our findings show that both MutSα and MutSβ are detected at origins during DNA replication . This finding contradicts the studies showing that MutSβ does not co-localize with the leading strand polymerase using fluorescence microscopy as the method of detecting associations [28] . Many reasons could account for the difference , including differences in detection methods . We are not able to detect the relative amounts of the complexes at the origins and it is possible that MutSα is more abundant and easily detected by both methods . Our finding that both complexes are present is consistent with studies using human cell lines where hMSH2 , hMSH3 and hMSH6 are found at sites of active replication [27] . Additionally , using mass spectrometry , MutSβ was shown to interact with the replisome in Schizosaccharomyces pombe ( Karin McDonald and Virginia Zakian , personal communications ) . Taken together , we favor a model where both complexes track with the replisome to cover the full spectrum of mismatches generated during replication . In this work we showed that the mismatch repair complex loads at origins of replication with kinetics similar to the DNA polymerase during S phase . The precise mechanism of loading at origin is not known . One hypothesis we explored was that PCNA was responsible for the loading and potentially for aiding in the scanning efficiency . We found that the mismatch recognition complex signal is still observed in the presence of the PCNA variants that perturb the interaction with MutSα however , the signal was aberrant in the mutant strains . These data are consistent with a model in which there is a PCNA independent association of the mismatch recognition complexes with the replisome . In this model , PCNA plays an important role in MutSα/β dynamics during replication , but it is not the sole determinant controlling MutSα/β loading at origins . This model is supported by findings showing a PCNA dependent and independent mechanism for mismatch repair [28 , 57] . Using chromatin immunoprecipitation and DNA tiling arrays , we are able to visualize the dynamics of MutSα/β binding during S phase; however , two models for movement along the DNA are consistent with the data: ( 1 ) MutSα/β loads at origins and scans immediately behind the advancing replisome facilitated by direct interactions with replisome components , or ( 2 ) MutSα/β loads at origins , but scans independently of the replisome . Because the two models involve loading of MutSα/β at active origins where the chromatin has been cleared , they both address the protein blockage problems discussed earlier . The first model is dependent upon a physical connection between MutSα/β and the replisome . Live-cell imaging during S-phase of S . cerevisiae cells show that Msh6 co-localizes with Pol2 [28] . Additionally , as mentioned above , MutSα/β has been shown to interact with PCNA [17 , 18 , 30 , 55 , 68] and PCNA is associated with the replisome [69] . Finally , in Bacillus subtilis MutS and MutL have been shown to interact with the catalytic subunit of the DNA polymerase III ( DnaE ) in vitro [70] . In vivo experiments in B . subtilis using GFP-tagged DnaE showed that mismatch detection causes the polymerase to disengage from the DNA during replication [70] . These experiments support the model that MutS and MutSα/β are directly associated with the replisome . Thus , we favor the first model based on the previous studies and our observations that the distribution of Msh2 signal is very similar to the distribution of the leading strand DNA polymerase as the replisome advances during S phase . The first model is also appealing because tracking directly behind the replisome ensures that the MutSα/β complexes are always in close proximity to a strand specificity signal: the 3’-OH of the newly synthesized strand . A further refinement of the model includes the following: MutSα/β loads at origins and scans immediately behind the advancing replisome as well as in the regions behind the replication fork . The addition to the model is based on the fact that the MutSα/β signal persists in the newly replicated region even when the leading strand polymerase appears to have cleared the region . The persistence of signal could be explained by the interaction of MutSα/β and PCNA . In eukaryotes , PCNA is known to accumulate behind the replisome [71] and in B . subtilis , DnaN ( PCNA ) clamp zones have been shown to remain behind the replication zone [72] . This DnaN-mediated recruitment of MutS is responsible for 90% of repair in B . subtilis with the remaining mismatch repair being DnaN-independent [32] . Taken together , we favor a model in which the persistent MutSα/β signal after fork passage is explained , in part , by interactions with PCNA molecules that remain behind the replisome . Fig 9 illustrates a model for MutSα/β signal distribution during replication . In the model , MutSα/β complexes bind to activated origins during S phase with a timing similar to DNA polymerase . The MutSα/β loading may be facilitated by direct PCNA interactions or modified histones may function to recruit MutSα/β to active origins . Once MutSα/β is loaded , a close association with the advancing replisome ensures that mismatches are rapidly detected and that the MMR machinery always has a proximal replication-specific nick to direct repair to the newly synthesizes strand . In the model , upon detection of a mismatch , the most proximal signal is the 3’-OH of the newly synthesized strand . The MutSα/β signal persisting behind the advancing replisome may be a consequence of PCNA interactions . PCNA is bound to nicks behind the replisome created during lagging strand synthesis and caused by rNMP excision .
Yeast strains used in this work are listed in S3 Table . Microbial and molecular manipulations were conducted according to previously published procedures [73 , 74] . Plasmid DNA extractions were performed using the Qiagen procedure ( Qiagen Inc . , Valencia , CA ) . Primers were synthesized by Integrated DNA Technologies Inc . ( Coralville , IA ) . Restriction endonuclease digestions and polymerase chain reactions ( PCR ) were performed using the enzyme manufacturer recommended reaction conditions ( New England Biolabs , Beverly , MA ) . The MSH2-myc POL2-HA strain was constructed by a genetic cross using strains from the Gammie laboratory ( MSH2-myc ) and the Bell laboratory ( POL2-HA ) . The pol30 mutant strains were constructed by creating the mutations on a centromere-based plasmid by recombination [75] and cloning the mutated pol30 gene into a URA3-based integrative plasmid vector backbone [76] . The mutations were introduced into the chromosome by a two-step integration method [77] . To produce the Msh6 PIP box mutant ( msh6-F33A , F34A ) we employed in vivo site directed mutagenesis [78] . We sequenced the MSH6 locus and confirmed the change resulting in replacement of the two conserved phenylalanines at codons 33 and 34 with alanines in the PIP box of MSH6 . To chromosomally tag MSH6 and MSH3 at the C-terminal coding regions , the myc or HA epitope tag and the kanamycin gene was amplified from the pFA6-x13myc or pFA6x3HA plasmids as described previously [79] . PCR amplified products were transformed into wild-type W303 . Integration was confirmed by PCR amplification of the epitope tag and sequencing . Western blot analysis was employed to confirm expression of the tag . Finally , the functionality of the fusions were confirmed by performing mismatch repair assays [80] . To achieve synchrony , cultures are grown to mid-exponential phase ( ~0 . 5 OD600 ) in SC medium at 30°C . The cells were then shifted to 18°C to slow the growth rate and arrested in the G1 phase of the cell cycle with 10 μg/ml α–factor . The cells were released from G1 arrest by washing the cells and resuspending in fresh medium . Samples were taken initially at 6 or 10-minute intervals for ChIP-chip , followed by 30 minute intervals for continued analysis of DNA content . Samples were collected for each time-point; the cells were cross-linked with freshly made 4% para-formaldehyde ( final concentration ~1% ) and flash frozen in liquid nitrogen . Aliquots from each time point were processed and analyzed by flow cytometry to determine which samples correspond to the cells in S phase . An aliquot of the fixed samples were processed for flow cytometry as previously described with modifications [81] . Briefly , cells were incubated with RNAase and SYTOX Green and the DNA content per cell was measured using the Becton-Dickinson LSRII Multi laser analyzer . Samples corresponding to S-phase of the cell cycle were then processed for ChIP-chip . The samples were processed for ChIP by mechanically disrupting the cell walls using a Fastprep -24 instrument ( MP Biomedicals LLC ) followed by sonication to generate DNA fragments averaging ~500 bp ( Covaris S220 Focused-ultrasonicator ) . A portion of each sample was retained as the input DNA . The remaining sample was split into two equal fractions and the cross-linked protein/DNA complexes were immunoprecipitated with antibodies conjugated to agarose beads ( one fraction with α-HA for Polε and the other with α-myc for MutSα/β complexes ) . To obtain signal corresponding to both MutSα/β and Polε , fixation conditions required optimization . Fixation for 45 minutes with freshly prepared para-formaldehyde facilitated the immunoprecipitation of both proteins . To confirm that the Msh2-myc tagged mismatch repair protein was immunoprecipitated , a time point in S phase was collected in duplicate . The sample was processed using identical conditions and the crosslinks were reversed . Western blot analysis was performed as described previously [82] to verify the immunoprecipitations . For the time course experiments , the cross-links from the ChIPs and inputs were reversed and the DNA was purified . Routinely , a portion of the samples was quantified by PCR to ensure that a ChIP-specific signal was detectable using Power SYBR Green PCR master mix ( Applied biosystems ) . Three technical replicates were performed for each time point . Samples were amplified for and the threshold cycles ( Ct ) were determined using the Sequence Detection System , SDS version 2 . 3 software ( Applied Biosystems ) . Both the input and ChIP DNA were amplified using ligation-mediated PCR [83] and labeled with fluorescent dyes Cy3 and Cy5 respectively ( reverse dye labeling controls were also performed ) . Labeled samples were hybridized to custom DNA tiling arrays with 15 , 000 probes ( Agilent technologies ) . The 24 regions represented include early , middle , and late firing origins and at least 20 kb of flanking DNA . A total of 65 origins were represented on the arrays: 53 are confirmed origins , 3 have previously been identified as likely origins , 6 proposed origins and 3 as dubious origins [36] . Additional features included telomere sequences , silenced loci , tRNA genes , highly transcribed genes and long terminal repeats , which are known replication pause sites [84] . Additionally , mono- , di- and tri-nucleotide repeats were included because these regions are associated with insertion/deletion loops requiring mismatch repair [85] . An Agilent DNA microarray scanner was used to detect the fluorescence intensities for Cy3 and Cy5 . The data was subsequently processed using Agilent Feature Extraction Software . Algorithims were used to correct for background and normalize the data and the log2 ratios of ChIP/input were calculated for the adjusted data . The data were then uploaded into the Princeton University microarray database ( PUMAdb ) . PUMAdb has features that facilitate data visualization and processing for a variety of programs . The data files processed in PUMAdb were converted to files compatible with Integrated Genome Browser ( IGB ) ( Affymetrix ) . IGB was used to represent the data as log2 ratios for individual experiments ( Version 8 . 0 . 1 ) [86] . For replicate experiments we used the Chipper software [54] with minor modifications . Chipper analysis generates the significance ( p-values ) of enrichment obtained from individual experiments using variance stabilization and not log2 ratios . The averaged data were visualized using IGB . The data are provided in an attached supplement ( S1 data ) . To determine the rate of fork progression the leading edge of the peak for each time point was measured . The difference the time points ( tp ) were then taken and divided by the time interval ( Rate of fork progression = ( tp1- tp2 ) / time interval ) . For two experiment sets , the average rate of fork progression of ~423 bp/min and ~ 438 kb/min respectively was determined for all origins by analyzing the bi-directional movement . In a few instances the leading edge was not discernable due to background signal . For the mutation accumulation and whole genome sequencing , the wild-type , msh2 , msh6 and msh3 knockout strains were propagated in rich medium ( YEPD ) for ~210 generations with bottlenecks every ~21 generations . Genomic DNA preparations , whole genome sequencing , and data analyses were as described previously [49] . Quantification of DNA enrichment in the ChIP and the input was performed using Q-PCR ( Power SYBR Green PCR master mix , Applied biosystems ) . Three technical replicates were performed for each time point . Samples were amplified for and the threshold cycles ( Ct ) were determined using the Sequence Detection System , SDS version 2 . 3 software ( Applied Biosystems ) . For the amplification of ARS305 , the forward and reverse primers used were 5’- GATTGAGGCCACAGCAAGAC-3’ and 5’- TCACACCGGACAGTACATGA-3’ respectively .
|
During replication , errors that escape the replication machinery are identified and repaired by DNA mismatch repair proteins . A mismatch in the helix is recognized by MutS homologs and subsequent events include excision of the error-containing strand followed by re-synthesis . A critical step in this process is directing repair to the newly synthesized strand . Current data suggest that transient discontinuities in the DNA backbone , known as nicks , generated during replication serve as the strand discrimination signals . Additionally , proteins that package DNA have the capacity to block mismatch recognition and are known to rapidly assemble behind the replication fork . Thus , there must be a short window of opportunity for the mismatch recognition complexes to scan for mismatches and access the strand discrimination signals . To address these issues , we tested the model that the mismatch recognition complexes track with the replisome . We employed high resolution genomic methods to determine that during replication , the mismatch recognition complexes bind origins of replication and advances with the replisome . The findings support the hypothesis that the mismatch recognition proteins track with the DNA replication machinery to accurately survey and repair the newly synthesized strands while the DNA is unpackaged and strand specificity signals are accessible .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
The Eukaryotic Mismatch Recognition Complexes Track with the Replisome during DNA Synthesis
|
The retinoblastoma ( Rb ) tumor suppressor acts with a number of chromatin cofactors in a wide range of species to suppress cell proliferation . The Caenorhabditis elegans retinoblastoma gene and many of these cofactors , called synMuv B genes , were identified in genetic screens for cell lineage defects caused by growth factor misexpression . Mutations in many synMuv B genes , including lin-35/Rb , also cause somatic misexpression of the germline RNA processing P granules and enhanced RNAi . We show here that multiple small RNA components , including a set of germline-specific Argonaute genes , are misexpressed in the soma of many synMuv B mutant animals , revealing one node for enhanced RNAi . Distinct classes of synMuv B mutants differ in the subcellular architecture of their misexpressed P granules , their profile of misexpressed small RNA and P granule genes , as well as their enhancement of RNAi and the related silencing of transgenes . These differences define three classes of synMuv B genes , representing three chromatin complexes: a LIN-35/Rb-containing DRM core complex , a SUMO-recruited Mec complex , and a synMuv B heterochromatin complex , suggesting that intersecting chromatin pathways regulate the repression of small RNA and P granule genes in the soma and the potency of RNAi . Consistent with this , the DRM complex and the synMuv B heterochromatin complex were genetically additive and displayed distinct antagonistic interactions with the MES-4 histone methyltransferase and the MRG-1 chromodomain protein , two germline chromatin regulators required for the synMuv phenotype and the somatic misexpression of P granule components . Thus intersecting synMuv B chromatin pathways conspire with synMuv B suppressor chromatin factors to regulate the expression of small RNA pathway genes , which enables heightened RNAi response . Regulation of small RNA pathway genes by human retinoblastoma may also underlie its role as a tumor suppressor gene .
The tumor suppressor protein Rb ( retinoblastoma ) is a chromatin factor that functions in transcriptional repression of cell cycle regulatory genes as well as other genes . As a transcriptional repressor , Rb functions in a core complex ( dREAM/Muv B ) that binds to specific promoters and recruits a crew of repressive chromatin cofactors to inhibit the expression of target genes [1] , [2] . Rb-recruited factors include repressive histone methyltransferases ( Suv39 , Suv420 ) , repressive heterochromatin proteins that bind to methylated histones ( HP-1 , L3MBT ) , and the Nucleosome Remodeling and Deacetylase complex ( NuRD complex ) [3]–[5] . Beyond transcription , Rb also interacts with other chromatin factors ( e . g . , condensin II ) and participates in other chromatin functions such as chromosome condensation and maintaining genome stability [6] , [7] . Even though it has been studied as a cell cycle regulator for two decades , the functions of Rb are clearly much broader . Rb pathway genes have been studied in the nematode C . elegans because Rb and many of its interacting proteins identified biochemically in flies and mammals are conserved in C . elegans and null alleles in the corresponding C . elegans genes cause similar developmental phenotypes [2] . Genes encoding Rb ( lin-35 ) , the rest of the core DRM/dREAM complex , and Rb-recruited repressive chromatin factors all belong to the class of synMuv B ( synthetic multivulva B ) genes , mutations in which cause a Muv ( Multivulva ) phenotype when combined with a mutation in a synMuv A gene . synMuv B genes , along with synMuv A genes , repress the expression of the growth factor gene lin-3/EGF in the developing hypodermis [8] . Excess EGF ( Epidermal Growth Factor ) signaling from the hypodermis in synMuv AB double mutant animals induces multiple vulva precursor cells to adopt the cell division patterns normally specified for only one vulval precursor cell , causing the Muv phenotype . Rb pathway chromatin factors comprise the bulk of the synMuv B genes as revealed by saturation genetic analysis and whole genome RNAi screens in synMuv A mutant strains , from which a few additional synMuv B genes have been identified , some of which also encode probable chromatin factors [9] , [10] . The genetic pathways necessary for the Muv phenotype in synMuv AB double mutant worms have been revealed by whole genome RNAi screens for gene inactivations that suppress the Muv phenotype of synMuv AB mutants , which also identified distinct chromatin factors [11] . Therefore , Rb pathway proteins function with particular chromatin factors while antagonizing others to specify the production of the LIN-3/EGF signal from particular cells during vulval development . Mutations in several synMuv B genes , especially those that encode the Rb core complex also cause a dramatic enhancement of response to dsRNA , that is enhanced RNAi ( Eri ) [12] , [13] . Inactivation of lin-35/Rb also causes enhanced transgene silencing , a process that depends on some RNAi factors [12] , [13] . Eri and enhanced transgene silencing are also caused by mutations in distinct RNAi regulatory factors , for example , the genes eri-1 to eri-9 , that do not interact with synMuv A mutations to induce a Muv phenotype [14] , [15] . lin-35/Rb likely inhibits RNAi using a distinct mechanism from these other eri genes because null alleles in lin-35 and eri genes are genetically additive , have different genetic requirements for canonical RNAi factors , and display specificity in gene inactivation tests involving distinct dsRNAs [13] . One potential mechanism of the enhanced RNAi response in synMuv B mutants is the somatic misexpression of germline-specific genes observed in these animals , given that many RNAi factors are preferentially expressed in the C elegans germline which is also more proficient at RNAi [13] . Many synMuv B mutants misexpress germline-specific P granules in their somatic tissue [13] , [16] , [17] . Homologous to nuage and polar granules of insects and mammals , P granules mark the germline of C . elegans from the very first cell divisions and are essential to the function and maintenance of the germline [18] , [19] . These perinuclear RNP granules harbor processing and binding proteins for mRNAs as well as endogenous small RNAs , and are thought to be the site of nascent mRNA export and endogenous small RNA biogenesis and storage [20]–[22] . The somatic misexpression of P granule components was first observed in mep-1 and let-418 mutants which were found to also function in the synMuv B pathway [17] . Unlike null mutants of most synMuv B genes , which are viable , null mutations of mep-1 or let-418 cause L1 arrest or sterility , suggesting that the expression of germline P granules in somatic cells can have severe developmental consequences or that these genes also regulate other essential functions in contrast to lin-35/Rb . The somatic misexpression of germline P granules as well as the developmental arrest phenotypes of mep-1 or let-418 are fully suppressed by inactivation of three synMuv B suppressor genes , which encode germline chromatin factors ( mes-4 , mrg-1 , isw-1 ) , pointing to antagonistic chromatin factors in regulating the expression of germline genes in the soma [11] , [23] , [24] . It was proposed that the NuRD complex , containing LET-418 , prevents somatic misexpression through chromatin modification or remodeling and antagonizing the function of germline specific chromatin factors [17] . Later , the misexpression of P granules in somatic cells was shown in several other synMuv B mutants , including lin-35/Rb , indicating a broader involvement of synMuv B genes in this process [13] . While these synMuv B mutants are viable , all of them display high temperature larval arrest , again suggesting severe developmental consequences from such misexpression of normally germline-specific genes [16] . Loss of the germline chromatin factors mes-4 , mrg-1 and isw-1 , which suppresses the somatic misexpression and high temperature larval arrest in these other synMuv B mutants , was also shown to suppress the enhanced transgene silencing and vulval specification phenotypes , suggesting that the somatic misexpression of germline components contributes to the enhanced RNAi and the vulval cell fate change in these animals [13] , [16] , [23]–[25] . The inappropriate expression of germline genes including small RNA pathway components could also contribute to the oncogenicity of loss of Rb in mammalian cells . Both the enhanced RNAi and the expression of germline-specific genes in soma uniquely reflect the function of the synMuv B pathway , as they are not present in mutants of other synMuv classes . However , not all synMuv B mutants are Eri or show PGL-1 misexpression , suggesting functional specialization . Moreover , only some of the Muv suppressing chromatin factors affect RNAi and only three ( mes-4 , mrg-1 , isw-1 ) suppress PGL-1 misexpression [11] . Thus , the genetic pathways underlying Eri and the expression of germline genes in somatic cells remain to be further investigated . Here we use response to RNAi , subcellular analysis of somatic P granules , and characterization of somatic misexpression of germline-specific genes to reveal that synMuv B genes constitute distinct intersecting axes for the repression of germline genes in somatic cells . We find that these chromatin factors regulate the expression of partially overlapping sets of germline target genes , including genes that encode annotated small RNA cofactors such as Argonaute proteins , and P granule components such as helicases and RNA binding proteins . The three synMuv B chromatin complexes we find are the LIN-35/Rb-containing core complex ( DRM ) , the SUMO-mediated Mec complex ( rather than the previously suspected NuRD complex ) , and a synMuv B heterochromatin complex , representing distinct classes of synMuv B mutants . We show that the DRM and synMuv B heterochromatin complexes each have distinct requirements for MES-4 function , suggesting different placements in the genetic pathway . We present a model of how synMuv B complexes collectively inhibit RNAi and prevent germline gene expression in the soma .
The majority of synMuv B genes revealed by saturation genetic and RNAi screens encode chromatin factors that belong to three functional classes: the DRM complex , a predicted NuRD complex , and several heterochromatin proteins ( Figure S1 ) . The worm DRM complex ( dREAM/MMB in flies and DREAM in mammals ) consists of LIN-35/Rb , the Rb binding protein LIN-53/RbBP4 , the transcription factor dimer EFL-1/E2F4-DPL-1/DP1 , and homologues of Myb-interacting proteins LIN-9/Mip130 , LIN-54/Mip120 and LIN-37/Mip40 [3] , [26]–[28] . LIN-53/RbBP4 is a shared component with the mammalian NuRD complex . The C . elegans NuRD complex has not been biochemically defined , but homologues of mammalian NuRD have emerged from genetic analyses . For example , genes encoding the histone deacetylase HDA-1/Rpd3 and the ATP-dependent chromatin remodeling enzyme LET-418/Mi-2 act in synMuv B pathways [17] , [29] , [30] . Although MEP-1 is not part of the mammalian NuRD due to the lack of an obvious homologue , it physically and functionally interacts with LET-418 in worms and flies [17] . synMuv B heterochromatin proteins include histone methyltransferases MET-1 and MET-2 , methyl histone binding proteins HPL-2/HP1 and LIN-61/L3MBT , and the multiple zinc finger protein LIN-13 , which mediates HPL-2 localization via physical interaction [31]–[34] . The molecular functions of the remaining synMuv B proteins are less known . These include LIN-15B , LIN-36 , TAM-1 , RPB-11 , E01A2 . 4 , GEI-4 and LIN-65 [9] , [10] , [35]–[38] . Many synMuv B mutants show an enhanced RNAi ( Eri ) phenotype and PGL-1 misexpression in somatic cells , but some do not ( Figure S1 and [11]–[13] , [16] , [17] , [31] , [33] , [39] , [40] ) . Even among the mutants that show enhanced RNAi and misexpress P granules , we noticed significant phenotypic differences , including strongly vs . weakly enhanced RNAi ( Figure 1A ) , enhanced transgene silencing vs . transgene desilencing ( Figure 1B , 1C ) , and large sparsely distributed vs . small densely clustered PGL-1 granules that are misexpressed in the intestine ( Figure 1D ) . These differences suggest functional specializations among synMuv B proteins , and may reflect distinct activities from individual functional classes . To test this , we systematically surveyed and classified all synMuv B genes based on these phenotypes . As shown in Figure S2 , the phenotypic differences segregated coherently and could be used to classify synMuv B mutants into three distinct classes . Null or strong mutations in all but one of the seven DRM components strongly enhanced response to dsRNA causing a strong enhanced RNAi phenotype ( close to 100% RNAi phenotype , Figure 1A ) , enhanced transgene silencing ( Figure 1B ) and caused the somatic misexpression of PGL-1 granules that were large and sparsely distributed around the nucleus of intestinal cells , as revealed by immunofluorescence imaging ( Figure 1D ) . The only exception was efl-1/E2F , which was previously reported not to cause enhanced RNAi and PGL-1 misexpression [12] , [16] . We also did not observe any phenotype upon inactivation of efl-1 , suggesting that EFL-1 is redundant or may not participate in all the activities of the DRM complex . In sharp contrast , null or strong mutations in the synMuv B heterochromatin class genes induced a noticeably weaker enhanced RNAi phenotype ( 60% or barely detectable RNAi phenotypes , Figure 1A ) , transgene desilencing , which is exactly the opposite from the transgene silencing phenotype of the DRM complex factors ( Figure 1C ) , and somatic misexpression of PGL-1 granules that were much smaller and densely clustered around the nucleus in intestinal cells ( Figure 1D ) . The only exception is met-1 , which had no detectable phenotype , suggesting a specific involvement of H3K9 methylation ( catalyzed by MET-2 ) but not H3K36 methylation ( catalyzed by MET-1 ) [31] , [41] , [42] . One interesting feature of this class is the decoupling between feeding RNAi efficiency and transgene silencing ability . It suggests that either these proteins inhibit a unique aspect of feeding RNAi , or they have additional roles in transgene silencing ( see below ) . Results for the synMuv B NuRD components were distinct from the other gene classes . Only transgene silencing and PGL-1 misexpression in the soma were assayed due to sterility/lethality associated with these mutants . RNAi inactivation of mep-1 caused enhanced transgene silencing , whereas inactivation of hda-1 or let-418 had no effect , suggesting that MEP-1 may function separately in transgene silencing . As reported [17] , inactivation of let-418 and mep-1 caused somatic PGL-1 misexpression . However , the misexpressed PGL-1 granules were small and densely clustered around the nucleus , as in the case of the heterochromatin class of synMuv B mutants but distinct from the DRM class mutants . Intriguingly , neither hda-1 mutant nor hda-1 RNAi treated worms showed detectable misexpression of PGL-1 in the soma ( Figure 1D ) , suggesting that either the histone deacetylase activity of NuRD is not required , or MEP-1 and LET-418 function outside the context of NuRD in preventing germline gene expression in the soma ( see below ) . Among the less studied synMuv B genes , we found that lin-15b mutant worms showed strongly enhanced RNAi ( Figure 1A ) , enhanced transgene silencing ( Figure 1B ) , and misexpressed large PGL-1 granules that are sparsely distributed around the nucleus ( Figure 1D ) , resembling mutants of the DRM complex ( Figure 1E ) . On the other hand , inactivating lin-65 led to weakly enhanced RNAi ( Figure 1A ) , transgene desilencing ( Figure 1C ) , and misexpressed small PGL-1 granules that are densely clustered ( Figure 1D ) , resembling the synMuv B heterochromatin class ( Figure 1E ) . Inactivation of tam-1 caused weakly enhanced RNAi , transgene silencing , but no PGL-1 misexpression , and thus may function independently in inhibiting only RNAi ( Figure 1E ) . No phenotype was observed upon inactivating lin-36 , rpb-11 , E01A2 . 4 or gei-4 , suggesting that they may not be involved in inhibiting RNAi or repressing germline genes in soma but rather may be involved in other aspects of vulva precursor cell specification ( Figure 1E ) . To test whether each synMuv B gene class differentially represses the expression of germline genes in somatic cells in general , we sought to identify molecular targets of each class . Given the enhanced RNAi and pgl-1 misexpression phenotypes , we first inspected the microarray studies of lin-35 mutant animals for somatic misexpression of known P-granule and RNAi genes [43] , [44] . We focused on the larval stage one ( L1 stage ) microarray comparisons between wild type and lin-35 null mutant animals because at that stage the germline has two quiescent cells and the soma has 550 cells . As verified in our analyses below , any misexpression of germline genes in somatic tissues at this stage would be more readily observed than at later stages , where a genetic ablation of the germline would be necessary . Because of the distinct features of enhanced RNAi and somatic misexpression of P granules in the various classes of synMuv B mutants , we expected that particular suites of germline components might be misexpressed in each class of synMuv B mutants . pgl-1 was dramatically upregulated at the L1 stage in two microarray studies of lin-35/Rb mutant larvae . In addition , three other genes known to encode P granule components , glh-1 , pgl-3 and spn-4 , were also upregulated ( Figure S3B ) , suggesting that the somatic misexpression of P granule components stems from transcriptional misregulation of germline genes in the somatic cells of the lin-35/Rb mutant . In fact , as shown in Figure S3A , out of the 307 genes that are upregulated in both L1 stage arrays , 28% ( 86 genes ) were identified as more than 2-fold germline-enriched by microarray [45] , representing a 4 . 6-fold enrichment over the entire genome ( 1584 genes , 8 . 3% ) . Similarly , 18 . 6% ( 57 genes ) of the lin-35 L1 up genes were identified as germline-specific or germline-enriched by SAGE [46] , a 3 . 3-fold enrichment over the entire genome ( 1063 genes , 5 . 6% ) . These analyses point to a striking preference of germline-enriched genes among the upregulated genes in these synMuv B mutants , which highlights the targeting of synMuv B genes to the repression of germline genes in somatic cells . We next looked for RNAi factors that were upregulated , which may contribute to the enhanced RNAi phenotypes . None of the genes encoding canonical RNAi factors such as rde-1 or dcr-1 were upregulated in any of the arrays , consistent with previous studies [11] . However , we found several genes encoding less well studied RNAi factors to be upregulated ( Figure S3B ) . Among them , C16C10 . 3/wago-9 , mut-2 , drh-3 , rde-4 and csr-1 are normally germline enriched [45] , [46] , and thus likely represent germline genes that are misexpressed in the soma . Four genes C04F12 . 1 , sago-2 , rrf-2 and drh-1 are not germline-enriched [45] , [46] , and thus represent ubiquitous RNAi factors that are upregulated . We carried out real-time RT-PCR experiments to confirm the upregulation of these genes in lin-35/Rb mutants , and to test whether the same upregulations occur in mutants of other synMuv B classes . To specifically detect upregulation in the soma , the glp-4 ( bn2 ) temperature sensitive mutant was used , which ablates germline development at 25C . We were able to verify more than two-fold upregulation for 11 out of the 13 candidate somatically expressed germline genes in glp-4 lin-35 mutants . Two genes , csr-1 and drh-1 , showed only marginal upregulation ( less than two fold ) , and were not analyzed further . We tested whether the same upregulations occur when inactivating other synMuv B classes . Specifically , we individually inactivated known components of each class , either by genetic mutation or by RNAi knockdown , and measured the expression of these 11 genes in the soma . Of these eleven genes , seven were upregulated upon the inactivation of most components of all three synMuv B classes . These include the P granule genes pgl-1 , pgl-3 and glh-1 , the germline enriched RNAi gene C16C10 . 3/wago-9 ( common germline targets ) , and all three ubiquitously expressed RNAi genes C04F12 . 1 , sago-2 and rrf-2 ( common ubiquitous targets ) ( Figure 2A , 2B and Figure S4 ) . Somatic upregulations ranged from several fold to several hundred fold , but for a given gene , the extent of upregulation was largely similar among different synMuv B mutants . Interestingly , despite no enhanced RNAi or PGL-1 misexpression phenotypes , inactivation of efl-1 induced upregulation of glh-1 , pgl-3 and C16C10 . 3/wago-9 ( Figure S4 ) , suggesting that EFL-1 participates in the repression of some but not all DRM target genes . Four germline-enriched genes , spn-4 , mut-2 , drh-3 and rde-4 , showed DRM-specific upregulation ( DRM-specific targets ) ( Figure 2C and Figure S4 ) . Compared to glp-4 control , these genes were upregulated 4–16 fold in lin-35 glp-4 mutants and upon RNAi knockdown of lin-9 , lin-54 or lin-52 in eri-1; glp-4 mutants . spn-4 was also upregulated in glp-4 worms treated with efl-1 ( RNAi ) . Consistent with its classification as a DRM class mutant based on P granule morphology and enhanced RNAi , inactivation of lin-15b caused upregulation of all seven common targets , and a DRM-specific target spn-4 ( Figure 2C ) . None of these DRM-specific target genes were upregulated in mep-1 or let-418 inactivations ( by RNAi in eri-1; glp-4 mutants ) , nor in any of the synMuv B heterochromatin inactivations ( in glp-4; hpl-2 or glp-4 lin-61 mutants , or upon RNAi knockdown of lin-13 or met-2 in eri-1; glp-4 mutants ) ( Figure 2C and Figure S4 ) . These results suggest that somatic repression of these genes solely depends on the DRM complex but not the synMuv B heterochromatin factors or MEP-1-LET-418 proteins . The DRM-specific upregulations prompted us to look for genes that might be specifically upregulated in synMuv B heterochromatin or mep-1-let-418 mutants . We focused on germline-enriched Argonaute genes that were not upregulated in lin-35 microarrays because of the central position of Argonaute genes in small RNA pathways . A limited survey of these genes showed that R06C7 . 1/wago-1 and F55A12 . 1/wago-10 were significantly upregulated in almost all inactivations of the synMuv B heterochromatin class genes and mep-1 & let-418 ( Figure 2D and Figure S4 ) . In contrast , no significant upregulation of these Argonaute genes was detected in any of the DRM class synMuv B gene inactivations ( Figure 2D and Figure S4 ) . Therefore , the synMuv B heterochromatin class and MEP-1-LET-418 proteins are uniquely required to repress a specific set of germline-enriched RNAi factors in somatic cells . We also extended the gene expression analyses to the remaining synMuv B genes . Inactivation of either lin-15b or lin-65 led to the misexpression of common germline and common ubiquitous targets ( Figure 2A , 2B ) . However , as shown in Figure 2C and 2D , inactivating lin-15b caused upregulation of spn-4 ( DRM -specific target ) , but not R06C7 . 1/wago-1 or F55A12 . 1/wago-10 ( heterochromatin- and MEP-1&LET-418-specific targets ) . In contrast , inactivation of lin-65 led to the upregulation of R06C7 . 1/wago-1 and F55A12 . 1/wago-10 , but only very marginal upregulation of spn-4 and no upregulation of the other three DRM-specific targets . These gene expression patterns again place lin-15b into the DRM class genes while lin-65 into the heterochromatin class , consistent with the previous phenotypic classification . For synMuv B genes that did not show PGL-1 misexpression or enhanced RNAi phenotypes in the phenotypic classification , most of the inactivations led to no or very modest upregulations of these P granule or RNAi factors ( Figure S4 ) . The only exception is gei-4 , whose inactivation led to the upregulation of several target genes . However , the most impressive upregulations observed were for the ubiquitous common targets , suggesting that gei-4 may be more specifically involved in the repression of ubiquitous targets . In summary , the three classes of synMuv B proteins function to repress overlapping sets of P granule and RNAi genes . The different spectra of misexpression suggest overlapping as well as non-overlapping functions among individual classes , and may underlie the distinct enhanced RNAi and transgene silencing phenotypes , and the distinct somatic P granule architectures observed in these mutants . The observation that loss of hda-1 , one of the catalytic subunits of the NuRD complex , does not induce somatic PGL-1 misexpression suggests that the entire complex is not required . To discern which NuRD subunits might be required , we surveyed the annotated C . elegans homologues of the NuRD complex for somatic P granule misexpression . We tested loss of function mutations and gene inactivations by RNAi of egr-1 and egl-27 ( MTA homologues ) , dcp-66 ( p66 homologue ) , chd-3 ( let-418 Mi2 paralogue ) and rba-1 ( lin-53 RBBP4 paralogue ) . We detected no somatic PGL-1 misexpression in any of the inactivations ( Figure 3A ) , suggesting that NuRD is not involved . Consistent with this , loss of lin-53 , a shared component between the DRM complex and the presumed NuRD complex , caused PGL-1 misexpression with a pattern that resembles that in the DRM mutants and differs from that in mep-1 or let-418 mutants . Thus , MEP-1 and LET-418 appear to function independently of NuRD in preventing somatic PGL-1 expression . Recent work in flies has revealed that dMEP-1 and dLET-418 form a stable complex named dMec , which recognizes SUMO modifications on transcription factors and carries out transcriptional repression [47] , [48] . C . elegans MEP-1 was reported to recognize and bind SUMO modified transcription factor LIN-1 and help repress LIN-1 target genes [49] . To test whether repression of PGL-1 misexpression required SUMO and a homologous Mec complex , we tested whether the SUMO pathway is required . Indeed , inactivation of the genes encoding the SUMO-activating enzyme ( uba-2 ) , the SUMO conjugation enzyme ( ubc-9 ) , and SUMO itself ( smo-1 ) all caused somatic PGL-1 misexpression ( Figure 3A ) . More importantly , the misexpressed PGL-1 form small granules that are densely clustered around the nucleus , identical to those in mep-1 and let-418 mutant worms ( Figure 3A ) . This result strongly supports the idea that in worms , MEP-1 and LET-418 prevent somatic PGL-1 misexpression as subunits of the Mec complex that requires SUMO modification of target transcription factors or chromatin factors , rather than as components in a NuRD complex . We next examined whether the SUMOylation machinery also regulates the other germline RNAi components that are responsive to mep-1 and let-418 . Using real-time RT-PCR , we tested the effect of hda-1 and smo-1 on the repression of other target genes . As shown in Figure 3B and Figure S4 , loss of smo-1 , but not hda-1 , caused somatic misexpression of glh-1 , pgl-3 and C16C10 . 3/wago-9 ( common germline targets ) as well as R06C7 . 1/wago-1 and F55A12 . 1/wago-10 ( synMuv B heterochromatin and MEP-1 and LET-418-specific targets ) . The same treatment did not lead to the upregulation of spn-4 , rde-4 , mut-2 , and drh-3 ( DRM-specific targets ) , indicating that smo-1 is specifically required for the repression of MEP-1/LET-418 target genes . Interestingly , smo-1 ( RNAi ) failed to induce the upregulation of C04F12 . 1 and sago-2 , and only weakly upregulated rrf-2 ( common ubiquitous targets ) , suggesting that the SUMO pathway may be specifically required to repress germline genes . Inactivation of hda-1 by RNAi did not lead to significant upregulation of any of the germline genes tested ( Figure 3B and Figure S4 ) , again supporting that the NuRD complex is not involved in repressing germline targets in the soma . On the other hand , hda-1 ( RNAi ) did lead to a modest upregulation of ubiquitous targets C04F12 . 1 and rrf-2 ( Figure 3B and Figure S4 ) , suggesting that the NuRD complex may contribute to the repression of ubiquitous target genes . The above results strongly suggest that it was the Mec complex , not the NuRD , that is required for preventing misexpression of germline genes in somatic cells . As in flies , the worm Mec complex may mediate transcriptional repressive effects of SUMO modifications on relevant transcription factors or chromatin factors . Phenotypic and gene expression analyses suggest that the class of synMuv B heterochromatin proteins likely function together , and in the case of transgene expression , act separately from the DRM or the Mec complexes . Consistent with acting separately from the DRM and the Mec complexes , members of the synMuv B heterochromatin class have distinct cytological localizations . GFP fusions to LIN-13 and HPL-2 form subnuclear foci that may correspond to compact heterochromatin [32] , [34] . In contrast , DRM complex components and GFP fusions to MEP-1 or LET-418 showed uniform nuclear localization [26] , [50]–[53] . We tested whether the synMuv B heterochromatin proteins physically function together cytologically and biochemically . We constructed a GFP fusion to the full length LIN-61/L3MBT protein and observed its subcellular localization . LIN-61::GFP rescued the Muv , enhanced RNAi and PGL-1 misexpression phenotypes of lin-61 ( tm2649 ) mutant animals , and thus is a functional fusion protein ( data not shown ) . GFP expression was detected by western analysis in all stages of worm development . Expression was the highest in embryos , where GFP was visible in most cells . LIN-61::GFP showed nuclear localization with concentrated foci , just like LIN-13::GFP and HPL-2::GFP ( Figure 4A ) . To look at the relationship between LIN-61::GFP foci and DNA , we costained LIN-61::GFP embryos with anti-GFP and DAPI . As shown in Figure 4B , anti-GFP antibody revealed foci resembling those formed by live GFP . Moreover , the most intensely stained GFP foci colocalize with areas of low DAPI staining , which is very similar to the reported pattern for HPL-2::GFP in interphase nuclei [32] . LIN-61::GFP was specifically depleted in nuclei with condensed chromosomes that are probably undergoing mitosis ( Figure 4B ) , which was also observed for LIN-13::GFP ( data not shown ) . The similar cytological localizations suggest that LIN-61 may colocalize with LIN-13 and HPL-2 in the nucleus . We thus generated a rescuing LIN-61::3xFLAG fusion gene ( data not shown ) and simultaneously visualized both LIN-61::3xFLAG and LIN-13::GFP fusion proteins in embryos by double immunostaining . As shown in Figure 4C , both LIN-61::3xFLAG and LIN-13::GFP localized to subnuclear regions that ranged from concentrated localization to discrete foci , and mostly coincided with poor DAPI staining ( data not shown ) . Most importantly , in cells coexpressing both fusion proteins , FLAG and GFP signals largely overlapped , consistent with a colocalization . However , unlike HPL-2::GFP foci , which depend on LIN-13 and disappear in the absence of lin-13 , LIN-61::GFP foci were not affected by the loss of lin-13 ( data not shown ) , suggesting a different mechanism of recruitment . Indeed , C . elegans LIN-61 was recently shown to directly bind to H3K9 methylation marks , just like HPL-2 [54] , which may serve as the basis of colocalization . To test whether LIN-61 physically interacts with other synMuv B heterochromatin proteins , we identified LIN-61-interacting proteins using large scale immunoprecipitation followed by mass spectrometry analysis . Embryos expressing LIN-61::GFP were used . Mass spectrometry analysis identified significant peptide coverage for LIN-13 and HPL-2 , which was absent from a parallel IP using untagged control embryos ( Figure S5 ) . This indicates that LIN-13 and HPL-2 were present in the immunoprecipitants and thus likely physically interacted with LIN-61 ( Figure 4D ) . Although peptides corresponding to additional proteins were identified ( X . Wu and G . Ruvkun , unpublished ) , none corresponds to components of the DRM or Mec complex . These results suggest that synMuv B heterochromatin proteins LIN-61 , LIN-13 and HPL-2 form a complex that is distinct from the DRM or the Mec complex . The interactions between LIN-61 and LIN-13 as well as between LIN-61 and HPL-2 were confirmed using coimmunoprecipitation-Western analyses . To this end , in addition to the LIN-61::3xFLAG fusion described above , we also generated a rescuing LIN-54::3xFLAG fusion gene ( data not shown ) . LIN-13::GFP or HPL-2::GFP fusion proteins immunoprecipitated the coexpressed LIN-61::3xFLAG protein ( Figure 4E ) . LIN-61::GFP fusion protein also immunoprecipitated LIN-61::3xFLAG , suggesting that there are more than one LIN-61 protein present in the immunoprecipitant , presumably reflecting a higher order packaging potential of heterochromatin proteins either by the self-dimerization potential of LIN-61/L3MBT proteins or of other members of this complex [5] , [55] . By contrast , the LIN-54::3xFLAG protein was not detected in the LIN-13::GFP or LIN-61::GFP precipitates , and only weakly detected in the HPL-2::GFP precipitate , again pointing to the existence of a distinct heterochromatin complex , which may only weakly interact with other synMuv B chromatin complexes such as the DRM complex . It is intriguing that the synMuv B heterochromatin mutants desilence transgenes , despite having enhanced RNAi response . This contradicts the general positive correlation between RNAi efficiency and the ability to silence transgenes , where enhanced RNAi efficiency often leads to enhanced transgene silencing while a deficiency in RNAi response usually leads to desilencing of transgenes [14] , [15] , [56] . Our analysis of double mutants showed that the transgene desilencing phenotype of synMuv B heterochromatin mutants was epistatic to a range of other enhanced RNAi mutants that also silence transgenes . Inactivation of synMuv B heterochromatin class genes reversed the transgene silencing induced by eri-1 mutations , DRM class synMuv B mutations and the natural silencing in the germline of wild type C . elegans worms ( Figure S6 and data not shown ) . Thus , the heterochromatin class of synMuv B proteins is uniquely required for transgene silencing , likely downstream or in parallel to their effects on RNAi efficiency . Similar to repetitive transgenes , silencing of endogenous repetitive gene clusters in the genome may also require both the small RNA-mediated silencing and the downstream or parallel actions of these heterochromatin factors . It is possible that the presence of heterochromatin proteins at the repetitive loci may in turn recruit additional small RNA factors to continue small RNA-mediated silencing , as has been suggested in yeast for the spreading of heterochromatin [57] , [58] . Thus , in addition to the transcriptional upregulation of RNAi factors , the loss of silencing at endogenous repetitive gene clusters and the release of small RNA factors such as Argonaute proteins to the exogenous RNAi pathway may also contribute to the enhanced RNAi effect in these heterochromatin mutants . The phenotypic differences between DRM and heterochromatin classes of synMuv B mutants suggest that the two complexes may affect RNAi and the repression of germline genes in the soma through different pathways . We tested this possibility using a genetic approach , that is , genes that function in separate pathways should be additive to each other when testing null allele combinations . We first looked at feeding RNAi efficiency of DRM; synMuv B heterochromatin double mutants . To allow the detection of further enhanced RNAi , we used conditions where strong Eri mutants only showed a partial phenotype . We found that diluting 2 parts of the cel-1 ( RNAi ) culture with 1 part of vector control RNAi culture dramatically reduced the RNAi response , reducing the nearly 100% L2 arrest phenotype in lin-35 mutants to only ∼5% . Under this condition , lin-35; hpl-2 and lin-35 lin-61 double null mutants showed significantly stronger RNAi response ( ∼40% to ∼60% L2 arrest ) compared to either single null mutant , indicating that the synMuv B heterochromatin mutations and lin-35 are additive ( Figure 5A , left panel ) . The same is true when using myo-2 ( RNAi ) as tester: while 0% to 2% single mutants scored showed L1 arrest , over 14% to 25% of the double mutants showed L1 arrest ( Figure 5A , middle panel ) . Consistent with lin-15b being classified with the DRM complex , additivity was also observed in heterochromatin; lin-15b double mutants using the diluted cel-1 ( RNAi ) assay ( Figure 5A , left panel ) or his-14 ( RNAi ) as tester ( Figure 5A , right panel ) . Since the mutations assayed were all presumed null alleles , the additivity in double mutants suggests that the DRM and synMuv B heterochromatin genes function in separate pathways . To see whether DRM and synMuv B heterochromatin proteins prevent gene misexpression via the same pathway or separate pathways , we assayed the misexpression of target genes that are commonly upregulated in both single mutants ( Figure 5B ) . Compared to treatment with vector control RNAi , glp-4; hpl-2 mutant animals treated with lin-35 ( RNAi ) displayed higher levels of misexpression of glh-1 ( germline targets ) , as well as rrf-2 ( ubiquitous targets ) . More importantly , the observed higher levels of misexpression exceeded those in glp-4 lin-35 mutant animals , thus represent true additivity . Similar treatment with lin-15b ( RNAi ) also led to significant higher levels of misexpression , whereas treatment with hpl-2 ( RNAi ) or lin-61 ( RNAi ) did not . Results for other common target genes were inconclusive; although higher mean levels of misexpression were detected upon lin-35 ( RNAi ) or lin-15b ( RNAi ) treatment in glp-4; hpl-2 mutant animals , the higher levels did not significantly exceed those in glp-4 lin-35 or glp-4; lin-15b animals , respectively ( data not shown ) , possibly due to the much milder upregulation from the hpl-2 mutation than from the lin-35 or lin-15b mutation and the incomplete knockdown of lin-35 or lin-15b activity by RNAi . Conversely , glp-4 lin-35 mutant animals treated with hpl-2 ( RNAi ) or lin-61 ( RNAi ) displayed higher levels of misexpression of glh-1 and rrf-2 , consistent with additivity . Interestingly , treatment with lin-15b ( RNAi ) also led to a significant higher misexpression in glp-4 lin-35 mutants , suggesting that lin-15b likely has additional functions beyond the DRM complex . Overall , the observed additive upregulations suggest that even though the DRM complex and synMuv B heterochromatin complex repress a common set of genes , they do so by providing nonoverlapping functions . Genetic suppressors of synMuv B mutations have emerged from screens for suppression of the Muv phenotype or screens for suppression of the enhanced RNAi phenotype [11] , [13] , [23] . Most of these suppressor mutations correspond to other chromatin factors that may counteract the action of synMuv B chromatin proteins . A few of the chromatin-related suppressors suppressed multiple phenotypes associated with synMuv B mutants [11] , [13] , [23] , suggesting that they may function antagonistically at target loci with synMuv B proteins . Given that the DRM complex and the synMuv B heterochromatin complex likely provide non-overlapping functions , it is important to know whether these broad-spectrum suppressors counteract all synMuv B activities , including those provided by different complexes . We addressed this question in the context of the somatically misexpressed P-granule and RNAi factors . We asked whether RNAi inactivation of two suppressors of both the PGL-1 misexpression and the enhanced RNAi phenotypes ( mes-4 and mrg-1 ) and two Eri-specific suppressors ( zfp-1 and gfl-1 ) can reverse the transcriptional upregulation observed in lin-35 and hpl-2 mutants . Inactivation of zfp-1 or gfl-1 did not suppress any of the transcriptional upregulation ( data not shown ) , suggesting that they do not antagonize the function of synMuv B complexes at the level of target transcription . Instead , they might be directly involved in the RNAi process and thus function downstream of the upregulation of P granule and RNAi factor genes . In contrast , inactivation of mes-4 ( Figure 6A , 6B ) and mrg-1 ( data not shown ) suppressed the upregulation of the P granule and RNAi factor genes that are common to both DRM complex and synMuv B heterochromatin class . However , at least for the common germline targets ( Figure 6A ) , the suppression was more complete in hpl-2 mutants than in lin-35 mutants , suggesting preferential suppression . We explored this possibility by analyzing the effect of mes-4 and mrg-1 inactivations on target genes that are unique to individual synMuv B classes . As shown in Figure 6C , mes-4 ( RNAi ) had no effect on the upregulation of DRM-specific targets spn-4 , mut-2 and rde-4 . But rather , it fully suppressed the upregulation of synMuv B heterochromatin class-specific targets R06C7 . 1/wago-1 and F55A12 . 1/wago-10 ( Figure 6D ) . mrg-1 ( RNAi ) gave identical results ( data not shown ) . The preferential suppression of target gene misexpression in the heterochromatin class synMuv B mutants further support functional specializations between individual synMuv B classes , and strongly suggest that the heterochromatin class synMuv B proteins specifically antagonize the activity of MES-4&MRG-1 . Upregulation of RNAi factors was only observed in synMuv B mutants that exhibit enhanced RNAi , suggesting that the elevated levels of these RNAi factors contribute to the enhanced RNAi phenotype . We explored this possibility by two means . First , RNAi factors that are commonly upregulated in synMuv B enhanced RNAi mutants include three Argonaute proteins and an RNA-dependent RNA polymerase , all of which are predicted to promote siRNA accumulation . We measured the siRNA levels in synMuv B mutants after pos-1 ( RNAi ) . pos-1 siRNAs accumulated to significantly higher levels in synMuv B mutants than in wild type animals ( Figure 7A ) . The increase correlates with the strength of enhanced RNAi response: 3 . 8- and 3 . 0-fold in the strong enhanced RNAi mutants lin-35 and lin-15b vs . 2 . 6- and 1 . 4-fold in the weaker enhanced RNAi mutants hpl-2 and lin-61 . Increased siRNA levels are likely the result of transcriptional upregulation of RNAi factors , since it was not observed in eri-1 mutants ( Figure 7A ) . Second , we asked whether the upregulated RNAi factors are required for the enhanced RNAi phenotype in synMuv B mutants . To this end , we introduced mutations of these RNAi factors into lin-35 or lin-15b mutants , and asked if RNAi efficiency was attenuated in the double mutants . As shown in Figure 7B , mutations in sago-2 and rrf-2 mildly reduced RNAi efficiency in lin-35 and lin-15b mutants . Mutation in C04F12 . 1 also mildly reduced RNAi efficiency in lin-15b mutants . Since mutations of sago-2 and rrf-2 on their own had little or no effect on RNAi efficiency [59] , [60] , the observed effects are specific to synMuv B mutants . The strain carrying a mutation in C16C10 . 3/wago-9 has weakly enhanced RNAi and was additive to the synMuv B mutations in the double mutant , and was thus not suitable for this analysis ( data not shown ) . We also tested whether forced overexpression of these RNAi factors would be sufficient to cause enhanced RNAi . Overexpression of the RNAi factors individually using sur-5 promoter-driven high copy transgenes did not lead to detectable enhanced RNAi ( data not shown ) . When tried to overexpress all four factors , we did not achieve systemic overexpression , and the resulting strain was not Eri ( data not shown ) . Overall , we conclude that transcriptional upregulation of RNAi factors likely contribute to the enhanced RNAi phenotype in synMuv B mutants , and that robustly enhanced RNAi response may require simultaneous upregulation of multiple RNAi factors .
The synMuv B class genes inhibit RNAi using distinct mechanisms from other eri genes . First , synMuv B mutations are phenotypically additive with eri mutations . eri-1; DRM and eri-1; synMuv B heterochromatin double mutants are even more enhanced in RNAi than either single mutant ( [13] and XW , ZS and GR , unpublished ) . Second , synMuv B proteins affect distinct aspects of RNAi compared to proteins in the eri-1 pathway . Unlike eri-1 mutants , synMuv B mutants are not defective for the production of eri-1-dependent endo small RNAs ( XW , ZS and GR , unpublished ) , and are normal for the eri-1-dependent nuclear localization of the NRDE-3/Argonaute protein ( MC and HM , unpublished ) . synMuv B mutants , however , accumulate higher levels of siRNAs , which did not occur in eri mutants . Taken together , synMuv B proteins likely inhibit RNAi at a different step than Eri pathway proteins . Rather than losing endogenous small RNA processes and releasing shared factors to exogenous RNAi [61] , [62] , as proposed for the eri pathway , synMuv B proteins directly affect the expression levels of RNAi factors . synMuv B proteins may repress the expression of yet to be identified RNAi factors in addition to those we have tested . This is particularly likely given that the upregulated RNAi factors studied here only partially mediate the enhanced RNAi phenotype . Indeed , several other lin-35-responsive genes were positives in previous screens for RNAi factors [56] , [63]–[65] . Specifically , cin-4 , spd-5 and lin-5 are required for dsRNA-mediated silencing of an RNAi sensor [56] . rad-51 is required for RNAi-mediated transcriptional gene silencing of a transgene [63] . rnp-8 , kbp-3 , rsa-1 and vha-7 are required for tandem high copy transgene-induced cosuppression of homologous endogenous gene [64] , while ucr-2 . 3 is required for transposon silencing [65] . Intriguingly , six out of these nine genes encode proteins that are important for chromosome biology during cell division ( cin-4 , spd-5 , lin-5 , rad-51 , kbp-3 , and rsa-1 ) , which again may reflect an intersection between chromatin regulators and RNAi machinery . The role of synMuv B proteins in RNAi involves more pathways than the misexpression of RNAi factors , as the effects of chromatin go beyond transcriptional regulation . In support for this , many of the chromatin-related synMuv suppressors are required for efficient RNAi [11] , suggesting that they may directly function in RNAi and their activities may be affected by synMuv B proteins . Among them , some may rescue the gene expression defects ( e . g . , mes-4 and mrg-1 ) , making it complicated to discern their direct effects on RNAi . Others , however , may not rescue target gene misexpression in somatic cells ( e . g . , zfp-1 and gfl-1 ) , and thus are more likely to function directly in RNAi mediated silencing . Future work will be focused on establishing potential physical interactions between synMuv B proteins and their suppressors and possible functional regulations on one another . Among synMuv B proteins , the DRM complex and the synMuv B heterochromatin complex each distinctively inhibit RNAi . This may involve the transcriptional regulation of common and unique RNAi factors ( as shown in this paper ) , or functional regulation of specific chromatin-related RNAi factors . Further understanding requires the identification of RNAi-related transcriptional as well as functional targets that are specifically regulated by individual synMuv B complexes . Our gene expression analysis of retinoblastoma pathway mutants reveals transcriptional upregulation of a suite of germline-specific genes in somatic cells of these mutant animals . More importantly , we provided strong evidence pointing to nonredundant functions by multiple synMuv B complexes , including the distinct subcellular architectures of the misexpressed PGL-1 granules , the partially nonoverlapping spectra of misexpression , and the genetic additivity between them . Their differential interactions with MES-4 and MRG-1 support their assignment to distinct pathways that prevent somatic expression of germline genes ( discussed later ) . Inactivation of the insulin pathway or components of the CCT cytosolic chaperonin complex also cause misexpression of germline-specific genes in somatic cells [66] . Some of the dysregulation in the daf-2 insulin pathway mutant may be through the inhibition of the MEP-1-LET-418 Mec complex as a result of PIE-1 misexpression [17] , [66] . However , the somatic misexpressions in daf-2 or cct mutants are much more modest than those in Mec mutants , and are hypothesized to involve other factors in addition to MES-4 and MRG-1 [66] . Thus , the insulin and cytosolic chaperonin pathways may represent a parallel mechanism for preventing germline gene expression in somatic cells . The relationship between daf-2 and the DRM complex awaits more investigation . The misexpression of germline-specific genes in somatic cells of animals lacking DRM function is extensive . Analogous microarray experiments in the soma of Mec , synMuv B heterochromatin , insulin and cct mutants will be necessary to disclose the full range of misexpression upon losing these other pathways . Ultimately , these misexpression profiles may help us understand the developmental and physiological phenotypes unique to each mutant . During C . elegans embryogenesis , primordial germline cells ( PGCs ) and somatic cells are derived from the same mother cells and both inherit MES-4/histone methyltransferase , whose activity sets up a germline competent chromatin [41] , [67] . MES-4 binds to genes that are expressed in the maternal germline , deposits RNA pol II-independent histone H3K36 trimethylation and marks them for germline expression in the current generation [41] . To prevent germline genes from expression in the soma , somatic cells require a mechanism to counteract the action of MES-4 until it gradually disappears at about the 100-cell stage . We propose that the Mec and synMuv B heterochromatin complexes counteract MES-4 activity in somatic cells . This is supported by the fact that for those germline targets affected by these two complexes , MES-4 binds to all of them ( Figure S3 ) [41] , [67] , [68] and mediates all their misexpressions . These two complexes may prevent the binding of MES-4 so that no new H3K36me3 marks can be deposited , or , they may prevent the recognition of MES-4-generated H3K36me3 marks . Our results that mes-4 is also required for the misexpression of somatic targets , which are normally not bound by MES-4 [68] , suggests that antagonizing MES-4 binding may be more likely . Thus , in the absence of Mec and synMuv B heterochromatin complexes , MES-4 may spread to these somatic genes and erroneously mark them as germline expressed . Directly analyzing the effects of Mec and synMuv B heterochromatin complexes on MES-4 localization at gene resolution by ChIP-chip and ChIP-seq techniques during embryogenesis will distinguish between these two possibilities . It was hypothesized that MEP-1-LET-418 antagonize MES-4 activity via the NuRD complex , since PIE-1 , which interacts with MEP-1-LET-418 , inhibits the histone deacetylase activity of HDA-1 in NuRD [17] . However , our results strongly suggest that NuRD is not involved . Rather , MEP-1-LET-418 act as the Mec complex in response to SUMO modification . Potentially , this also involves the synMuv B heterochromatin complex , since LIN-61/L3MBT was also shown to mediate SUMO-triggered transcriptional repression [48] . It is unclear how the Mec and synMuv B heterochromatin complexes might prevent MES-4 binding . One possibility is that they change the structure and accessibility of chromatin , given that LET-418/Mi-2 is an ATP-dependent nucleosome remodeling factor while LIN-61/L3MBT and HPL-2/HP1 may cause higher order chromatin compaction [5] , [55] . Alternatively , they may prevent MES-4 binding by recruiting other factors that erase the histone modifications recognized by MES-4 [69] , [70] . Analyzing the states of histone modifications during normal repression vs . misexpression might help to understand the mechanisms . The result that DRM-specific targets are not suppressed by mes-4 inactivation indicates that the DRM complex also functions downstream of MES-4 or in a parallel pathway . The DRM complex might be recruited to these target loci by sequence specific DNA binding ( e . g . , by EFL-1/E2F or by LIN-54/Mip120 ) and carry out repression . Indeed , among the 307 lin-35 L1 stage up-regulated genes , 163 either contain E2F binding sites in their promoter [44] , or were reported to be bound by LIN-54 in a ChIP-chip study [71] , or both . Therefore , loss of DRM-mediated repression at target promoters may account for much of the misexpression that was observed . On the other hand , given the vast extent of gene misexpression in lin-35 mutants , misexpression may also involve antagonizing master regulators of germline expression ( e . g . , other germline chromatin factors ) . Identifying factors that specifically mediate the misexpressions in DRM mutants may uncover novel master regulators of germline expression . It is interesting that the misexpressed germline genes are the most detectable in intestine and hypodermis , suggesting that these two tissues are the most prone to misexpression . This might be caused by a difference in the initial MES-4-MRG-1 activity levels or by a difference in the overall transcriptional activity and/or chromatin configuration among different tissues . These two are not mutually exclusive . Intestine and hypodermis are two tissues that remain replication competent even towards the end of developmental maturity ( L4/adult molt ) or even after that [72] , [73] . As such , they may maintain a different chromatin state than other tissues , making them more susceptible to the action of germline chromatin regulators , and rely more on Mec and synMuv B heterochromatin complexes to prevent misexpression of germline genes . The function of LIN-35/Rb and other synMuv B chromatin factors to prevent germline gene expression in the soma is likely conserved in higher organisms . In Drosophila , misexpression of germ granule and small RNA factors were also detected in Rb mutant cells by microarray analyses [74] and in the brain tumors developed in l ( 3 ) mbt ( lin-61 homologue ) mutant flies [75] . Future work to understand the antagonistic relationship between synMuv B complexes and MES-4-MRG-1 is crucial to understand the process of repressing germline genes in the soma during development . In addition , since the DRM complex functions beyond antagonizing MES-4-MRG-1 , identifying the unknown components specific to the DRM pathway will likely uncover novel players that function in this process . Besides regulating the cell cycle , mammalian Rb also promotes differentiation , and cancer cells are sometimes dedifferentiated [76] . The expression of germline-specific genes in somatic cells in worms and flies may be a form of dedifferentiation , and the role of synMuv B genes to prevent that may be important for tumor suppression . Consistent with this idea , the misexpressed germ granule and small RNA factors were shown to be required for the growth of the l ( 3 ) mbt mutant tumors [75] . Thus , the knowledge of how Rb and other synMuv B chromatin complexes function to prevent misexpression may contribute to our understanding of cancer formation and the identification of new pathway components may lead to the discovery of novel tumor suppressive and oncogenic pathways . Finally , the explicit mapping of C . elegans LIN-35/Rb and other synMuv B chromatin factors to a regulatory pathway for repression of germline encoded small RNA pathways suggests that in some mammalian tumors , most especially the many that bear Rb mutations , there may be specific enhancement of small RNA pathways to empower the dedifferentiation of those tumors towards more multipotent germline-like cells . A specific prediction from our work is that one of the major transcriptional signatures of Rb tumors should be misexpression of germline small RNA cofactors and small RNAs themselves .
synMuv B alleles used in this study are listed in Figure S2 . Other alleles used are glp-4 ( bn2 ) I , smo-1 ( ok359 ) I , dcp-66 ( gk370 ) I , rrf-1 ( pk1417 ) I , rrf-2 ( ok210 ) I , C04F12 . 1 ( tm1637 ) I , sago-2 ( tm894 ) I , hT2[bli-4 ( e937 ) let- ? ( q782 ) qIs48] ( I;III ) , szT1[lon-2 ( e678 ) ] I;X , szT1 ( I;X ) , dpy-18 ( e364 ) III , wago-9/C16C10 . 3 ( tm1200 ) III , dpy-17 ( e164 ) III , eT1 ( III;V ) , eri-1 ( mg366 ) IV , nT1[qIs51] ( IV;V ) , unc-46 ( e177 ) dpy-11 ( e224 ) V , egr-1 ( gk255 ) V , chd-3 ( eh4 ) X . Transgenic strains Is ( sur-5::gfp ) I , JR672 [wIs54 ( scm::gfp ) V] , GR1403 [sur-5::gfp I; eri-1 ( mg366 ) IV] , PD7271{pha-1 ( e2123 ) III; ccEx7271[let-858::gfp pha-1 ( + ) ]} , GR1402 [eri-1 ( mg366 ) IV; wIs54 V] and FR463 [Is ( HPL-2::GFP+pRF4 ) ] were previously described [32] , [56] . The LIN-61::GFP translational fusion consists , in order , of a 2 . 9 kb lin-61 genomic fragment containing the upstream intergenic sequence and coding region minus the STOP codon , an engineered sequence encoding a short peptide linker ( GGAGGSAAA ) followed by GFP sequence amplified from pPD99 . 77 , and then a 244 bp lin-61 3′ genomic fragment starting immediately after the STOP codon . Extra chromosomal complex arrays were generated by injecting wild type animals with a mixture of 10 ng/ul LIN-61::GFP fusion plasmid+1 ng/ul Pmyo-2::NLS::mCherry ( gift from J . Bai and J . Kaplan ) +89 ng/ul sheared salmon sperm DNA . The LIN-61::3xFLAG fusion was generated by replacing the GFP sequence in LIN-61::GFP fusion construct with 3xFLAG sequence . LIN-54::3xFLAG fusion consists of a 2 . 6 kb lin-54 genomic fragment containing the upstream intergenic sequence and coding region minus the STOP codon , the same short peptide linker followed by 3xFLAG sequence , and then a 413 bp lin-54 3′ genomic fragment starting immediately after the STOP codon . Extra chromosomal complex arrays were generated by injecting wild type animals with a mixture of 10 ng/ul 3xFLAG fusion plasmid+1 ng/ul Pmec-7::mRFP+89 ng/ul sheared salmon sperm DNA . All integrated transgenic lines were obtained by UV irradiation , followed by six times backcrossing . LIN-13::GFP was reported to be a functional fusion gene [34] . All other translational fusion genes were tested to be rescuing functional fusions ( data not shown ) . RNAi clones targeting let-418 , lin-13 and lin-15b were made by individually cloning genomic fragments corresponding to part of let-418 exon 5 ( 1025 bp ) , part of lin-13 exon 12 ( 727 bp ) and part of lin-15b exon 5 ( 1053 bp ) into the NcoI site of the Ahringer L4440 feeding vector . The other RNAi clones used were from the Ahringer or Vidal libraries . To measure Is ( sur-5::gfp ) expression , synchronized L1 animals of the appropriate genotype were treated with RNAi at 15C till the L3 stage of the same generation ( P0 clones ) or that of the second generation ( F1 clones ) before scoring . The same was done to measure wIs54 ( scm::gfp ) expression , except that animals were cultured at 22C and GFP was scored at the L4 stage . Germline desilencing using PD7271 was assayed according to Kim et . al [56] . For PGL-1 misexpression , mutant or RNAi treated animals were cultured at 22C for two generations ( one generation for P0 RNAi treatment ) . PGL-1 immunostaining was performed as described [13] with some modifications . Freeze-cracked slides were immediately fixed at −20C in 100% methanol for 120 minutes , and then 100% acetone for 15 minutes . Larvae were stained with monoclonal anti-PGL-1 antibody ( K76 ) at 1∶40 in PBSTB ( PBS containing 0 . 1% Tween-20 and 1% BSA ) overnight at 4C , followed by Alexa Fluor-conjugated goat anti-mouse IgM ( Invitrogen ) at 1∶100 in PBST for 1 hour at 25C . After washing with PBST , slides were mounted with Vectashield mounting medium with DAPI ( H-1200 ) and analyzed using a Zeiss Axioplan2 microscope and Openlab imaging software . Immunostainings of LIN-61::GFP , LIN-13::GFP and LIN-61::3xFLAG in embryos were performed similarly . Rabbit polyclonal anti GFP ( Invitrogen A11122 ) and Sigma M2 anti-FLAG were used at 1∶1000 and 1∶50 , respectively . Alexa Fluor488-conjugated goat anti-rabbit Ig G and Alexa Fluor594-conjugated goat anti-mouse Ig G ( Invitrogen ) were both used at 1∶100 . To show detailed PGL-1 granule morphology ( Figure 1D ) , stained animals were photographed using a Zeiss Imager Z1 microscope and AxioVision software . Images were each deconvoluted from a stack of fifteen optical sections using algorism provided by the software . Synchronized L1 stage wild type , glp-4 , and synMuv B; glp-4 animals were cultured at 25C till L4 stage before collection . For one generation RNAi due to sterility ( let-418 RNAi , smo-1 RNAi , hda-1 RNAi ) , L1 stage glp-4 animals were dropped on RNAi food and allowed to grow at 25C till L4 stage . For all other experiments involving RNAi treatment , synchronized L1 stage glp-4 animals were dropped onto RNAi food and cultured at 15C till gravid adults with plateful of eggs . Adults and F1 larvae were washed off , leaving the eggs behind . These eggs were allowed to hatch at 25C for 3 hours . The newly hatched larvae were washed off , transferred onto new plates with the same RNAi food and allowed to grow to the L4 stage before collection . Total RNA was isolated using Trizol ( Invitrogen ) and cDNA was synthesized using Superscript III reverse transcriptase ( Invitrogen ) . Negative control without reverse transcriptase was performed on a pooled sample of all the RNA samples under study . Real time PCR reactions were performed using SYBR green ( BioRad ) . At least three biological replicate samples were tested in triplicate . For each primer pair , a standard curve was generated using serial dilutions of a pooled sample of all cDNA templates involved . Relative quantities were deduced using the standard curve . Fold changes of relative quantities were calculated and normalized to rpl-32 and act-1 . Fold differences for rpl-32 and act-1 were less than two fold . Primer sequences are available upon request . Embryos were collected by hypochlorite treatment of gravid adults and dropped into liquid nitrogen to freeze . Frozen pellets were freeze ground to fine powder using a mortar and pestle and 1∶6 resuspended in 50 mM HEPES pH 7 . 6 , 140 mM KCl , 10 mM NaCl , 1 mM MgCl2 , 1 mM EDTA , 10% glycerol , 0 . 1% β-octylglucoside , 0 . 5 mM BME , Complete Protease tablets ( Roche ) , 2 mM PMSF . Lysates were cleared by two centrifugations each at 15000× g for 20 min . Immunoprecipitation was performed using monoclonal anti-GFP clone 3E6 ( Invitrogen ) coupled to Affy ProA agarose beads ( BioRad ) at 4C for 2 hours . Immunoprecipitates were washed using the same buffer . Bound proteins were eluted using 0 . 1 M glycine ( pH 2 . 5 ) , and precipitated using TCA . Pellets were washed with acetone , resuspended in 50 mM NH4HCO3 and sent for mass spectrometry . Immunoprecipitations for co-IP-Western analyses were performed similarly , except that 50 ul packed embryos were used and bound proteins were eluted from the anti-GFP matrix by boiling in PAGE gel sample loading buffer . Eluted proteins were analyzed by SDS-PAGE and Western analyses . Roche anti-GFP ( cat#11814460001 ) , Sigma M2 anti-FLAG were used as primary antibodies , Pierce peroxidase-conjugated goat anti-mouse IgG antibody ( cat#31430 ) and Pierce SuperSignal West Pico Extended Duration chemiluminescent detection kit ( cat#34076 ) were used . For the diluted cel-1 ( RNAi ) assay , OD600 matched cel-1 ( RNAi ) and vector RNAi cultures were mixed at 2∶1 ratio by volume before seeding plates . First day gravid adults reared on OP50 food at 20C were exposed to RNAi food for 24 hours , then transferred to new plates with the same food for a 4-hour egg lay . Afterwards , adults were taken off the plate and laid embryos were allowed to hatch and grow at 20C . L2 arrest was scored when animals of the same genotype treated with only vector RNAi reached adulthood . For the myo-2 ( RNAi ) and his-14 ( RNAi ) assays , synchronized L1 animals were treated with RNAi at 20C and were scored when animals of the same genotype reached adulthood on vector RNAi . Synchronized L1 animals were dropped onto pos-1 ( RNAi ) or vector RNAi food and cultured at 20C until gravid adults with plateful of eggs . Adults were collected and total RNA was isolated using Trizol ( Invitrogen ) . 60 ug of total RNA were used for each sample . pos-1 siRNA was detected using body-labeled pos-1 probe ( Ambion MaxiScript kit ) . The stripped blot was reprobed with an end-labeled U6 oligo probe . Hybridization signals were analyzed using ImageQuant software and normalized to wild type samples .
|
In metazoans , soma and germline have specialized functions that require differential tissue-specific gene expression . In C . elegans , explicit chromatin marks deposited by the MES-4 histone methyltransferase and the MRG-1 chromodomain protein allow germline expression of particular suites of target genes . Conversely , the expression of germline-specific genes is repressed in somatic cells by other chromatin regulatory factors , including the retinoblastoma pathway genes . We characterized the distinct profiles of somatic misexpression of normally germline-specific genes in these mutants and mapped out three chromatin complexes that prevent misexpression . We demonstrate that one of the complexes closely counteracts the activity of MES-4 and MRG-1 , whereas another complex interacts with additional regulators that are yet to be identified . We show that these intersecting chromatin complexes prevent the upregulation of a suite of germline-specific as well as ubiquitous small RNA pathway genes , which contributes to the enhanced RNAi response in retinoblastoma pathway mutant worms . We suggest that this function of the retinoblastoma pathway chromatin factors to prevent germline-associated gene expression programs in the soma and the upregulation of small RNA pathways may also underlie their role as tumor suppressors .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"biology"
] |
2012
|
Repression of Germline RNAi Pathways in Somatic Cells by Retinoblastoma Pathway Chromatin Complexes
|
Respiratory syncytial virus ( RSV ) infection can result in severe disease partially due to its ability to interfere with the initiation of Th1 responses targeting the production of type I interferons ( IFN ) and promoting a Th2 immune environment . Epigenetic modulation of gene transcription has been shown to be important in regulating inflammatory pathways . RSV-infected bone marrow-derived DCs ( BMDCs ) upregulated expression of Kdm5b/Jarid1b H3K4 demethylase . Kdm5b-specific siRNA inhibition in BMDC led to a 10-fold increase in IFN-β as well as increases in IL-6 and TNF-α compared to control-transfected cells . The generation of Kdm5bfl/fl-CD11c-Cre+ mice recapitulated the latter results during in vitro DC activation showing innate cytokine modulation . In vivo , infection of Kdm5bfl/fl-CD11c-Cre+ mice with RSV resulted in higher production of IFN-γ and reduced IL-4 and IL-5 compared to littermate controls , with significantly decreased inflammation , IL-13 , and mucus production in the lungs . Sensitization with RSV-infected DCs into the airways of naïve mice led to an exacerbated response when mice were challenged with live RSV infection . When Kdm5b was blocked in DCs with siRNA or DCs from Kdm5bfl/fl-CD11c-CRE mice were used , the exacerbated response was abrogated . Importantly , human monocyte-derived DCs treated with a chemical inhibitor for KDM5B resulted in increased innate cytokine levels as well as elicited decreased Th2 cytokines when co-cultured with RSV reactivated CD4+ T cells . These results suggest that KDM5B acts to repress type I IFN and other innate cytokines to promote an altered immune response following RSV infection that contributes to development of chronic disease .
Respiratory syncytial virus ( RSV ) is a significant burden to healthcare worldwide . In the United States nearly all children have been infected by age two [1 , 2] . Severe infections are the leading cause of bronchiolitis in children , resulting in up to 125 , 000 hospitalizations in the US each year [3 , 4] . Furthermore , infants who are hospitalized with severe disease are 3–4 times more likely to develop asthma later in life [5 , 6] , as RSV interferes with the initiation of the adaptive immune response leading to an altered immune environment in the lungs . One consequence is the recruitment of T cells that produce interleukin ( IL ) -4 , IL-5 and IL-13 , which are cytokines involved in the pathogenesis of lung disease , including asthma . Since RSV infection does not result in a strong memory response , repeated infections are common throughout childhood and for adults [7] . Currently , there is no vaccine for RSV . Attempts to create a formalin-inactivated vaccine resulted in more severe infection upon exposure to the virus in those children who were vaccinated compared to unvaccinated children [8] . Thus , the need to understand the immune response to RSV at the molecular level is critical to develop better therapeutics and aid in the development of effective vaccines . Dendritic cells ( DCs ) residing in the airway are among the first cells to encounter an RSV infection . An important role for DCs is the production of IL-12 , which is critical to initiate a Th1 response following infection with a virus . Viral infection also results in the recognition of pathogen-associated molecular patterns ( PAMPs ) such as viral nucleic acids , including double-stranded RNA . DCs express toll-like receptors ( TLRs ) that recognize these PAMPs . Viral RNA recognition signals induce the production of type I IFNs , including IFN-β , thereby leading to an antiviral state in nearby cells and initiating the proper T cell response . However , while other respiratory viruses such as influenza drive a strong IFN-β response , RSV infection suppresses this response , which is partially due to viral non-structural protein interactions with activation factors , resulting in the suppression of both helicase RIG-I [9] and the transcription factor NF-γB [10] . These events not only allow for viral replication , but also promote an altered immune environment in the lungs , characterized by mucus hypersecretion [11] . As DCs are important cells in directing the T cell response , it is imperative to understand the mechanisms by which RSV interferes with proper DC function . Recent studies have documented that epigenetic modifications of immune cells at the chromatin level contribute to either activation or repression of specific genes . This regulation can occur by several mechanisms , including methylation , acetylation and phosphorylation of histone tails . Histone methylation can occur on lysine and arginine residues , and contribute to activation or inhibition of transcription . For example , the addition of methyl groups on lysine ( K ) 27 of histone ( H ) 3 ( i . e . H3K27 ) is associated with repression of gene transcription , whereas H3K4 methylation correlates with active gene transcription [12–14] . This process is controlled by methyltransferases that add methyl groups and demethylases that remove them . There is accumulating evidence that epigenetic modulation is an important component of immune cell phenotype and function . For example , the cytokines typically produced by Th2 cells , including IL-4 , IL-5 and IL-13 , are silenced in Th1 cells through histone modification [15] . In addition , Th17 cells have increased H3K4 methylation , a gene activating mark , at the Th17 promoter [16] . In DCs , IL-12 production is decreased following severe sepsis due to decreased H3K4 methylation altering subsequent immune responses [17] . Despite these inroads , data is lacking regarding the contribution of epigenetic regulation to immune cell activation . In the present studies an upregulation of Kdm5b , coding for an H3K4 demethylase , following RSV infection of DCs was observed . As H3K4 methylation is an activating mark , this demethylase has the potential to repress transcription . A role for KDM5B in transcriptional repression has been reported in cancer cells including melanoma , breast cancer and prostate cancer [18–20] . Here we show that KDM5B has a critical role in preventing the activation of DCs during RSV-induced immune responses . Our results show that decreasing Kdm5b expression by siRNA , chemical inhibition or genetic deletion prior to RSV infection leads to an increase in the production of IFN-β and other inflammatory cytokines compared to uninfected controls , as well as decreased Th2 pathogenesis in vivo thus linking Kdm5b expression with disease exacerbation during RSV infection .
A previous report has identified a role for epigenetic regulation in immune cells following viral infection [21] . As DCs are critical for priming the T cell response to RSV infection , studies were initiated to determine whether exposing DCs to RSV resulted in changes in the expression of epigenetic factors in the DCs . BMDCs were infected with RSV or activated by p ( I:C ) or imiquimod , the ligands for TLR3 and TLR7 respectively , as RSV is known to activate cells through both TLR3 and TLR7 [22 , 23] , in addition to other mechanisms . In order to observe early gene expression of epigenetic enzymes , RNA was harvested at 4 hours post treatment to examine transcription levels of genes coding for “epigenetic” enzymes by qPCR array . Several classes of enzymes were analyzed including histone deacetylases ( HDACs ) , histone lysine demethylases ( KDMs ) , protein arginine methyltransferases ( PRMTs ) , and histone lysine methyltransferases ( KMTs ) ( Fig 1A ) . A defining observation was the upregulation of Kdm5b demethylase by RSV in contrast to the downregulation of this enzyme by stimulation through TLR3 and TLR7 ( Fig 1A ) . While Kdm6b was upregulated by RSV infection of DCs , this enzyme was also significantly upregulated by treatment of cells with imiquimod . Because Kdm5b was upregulated only by RSV , studies focused on Kdm5b as a potential unique enzyme in the DC response to RSV . PCR analysis confirmed the peak expression of Kdm5b in BMDCs at 12 hours following RSV infection ( Fig 1B ) . Furthermore , while Kdm5b was upregulated in BMDCs infected with RSV , it was not upregulated by influenza ( H1N1 ) virus , nor in RSV-infected epithelial cells or alveolar macrophages ( S1 Fig ) . Therefore , studies focused on H3K4 demethylase Kdm5b and its role on perturbing critical innate immune genes in DCs . To determine whether KDM5B affects DC function , specific siRNA was used to knock down Kdm5b resulting in >70% reduction in expression levels ( Fig 2A ) . Previous reports have indicated that RSV , unlike many viruses , is a poor inducer of type I IFN , including IFN-β [9 , 10] . BMDCs infected with RSV produced low levels of IFN-β at both 4 and 24 hours , whereas H1N1 virus produced very high levels ( S2 Fig ) . We therefore hypothesized that the increase in KDM5B in BMDCs contributed to the suppression of type I IFN production and that knocking down Kdm5b expression would result in increased IFN-β . Following in vitro treatment of BMDCs with Kdm5b-specific siRNA or with a scrambled siRNA control , significantly increased expression levels of Ifnb , as well as the pro-inflammatory cytokines Tnfa and Il6 were observed in in vitro RSV-infected cells compared to sham-infected BMDCs ( Fig 2B ) . To determine whether APC function was affected by Kdm5b siRNA or inhibitor treatment , MHC-II expression on the cell surface of BMDCs was measured , as well as expression of the co-stimulatory molecules CD80 and CD86 . No differences in any maturation markers were noticed in treated cells compared to controls ( S3 Fig ) . Furthermore , when a chemical inhibitor , 2 , 4-pyridinedicarboxylic acid ( 2 , 4-PDCA ) , was used to block the function of KDM5B [24 , 25] prior to RSV infection , significantly higher levels of Ifnb , Tnfa and Il6 transcripts compared to controls were observed ( Fig 2C ) . While this inhibitor also interacts with other KDM family members , it has the highest specificity for KDM5B . Thus , two independent approaches to block KDM5B function demonstrated an altered immune response resulting in increases of critical innate cytokines . KDM5B catalyzes the demethylation of H3K4me3 and H3K4me2 . As H3K4me3 is associated with active gene transcription , the activity of KDM5B in removing a methyl group leads to decreased promoter activity and decreased gene transcription . Since blocking KDM5B activity led to increased levels of proinflammatory cytokines , we hypothesized that blocking the demethylase activity would lead to greater H3K4me3 at the promoters of these cytokines . To test this hypothesis , a ChIP assay using an anti-H3K4me3 antibody was performed on cells treated with 2 , 4-PDCA , and primers designed to recognize the promoter regions of Ifnb , Tnf and Il6 were used . Treatment of DC with 2 , 4-PDCA prior to infection with RSV led to an increase of H3K4me3 compared to controls on all three cytokine promoters ( Fig 2D ) . Conversely , it was found that there were no differences in H3K4 methylation on the promoters of Il10 and Il12 , which are two cytokines that were unchanged following RSV infection of Kdm5b-deficient DCs ( S4 Fig ) . These results indicate that the KDM5B demethylase activity acts on the promoters of specific inflammatory genes , thereby suppressing transcription . The above data show that inhibiting KDM5B function leads to increased innate cytokine production after RSV infection . To determine the role of KDM5B in human cells , human monocyte-derived DCs ( MoDCs ) cultured from peripheral blood monocytes were used . As shown in Fig 3A , KDM5B is significantly upregulated in MoDCs at 12 and 24 hours following RSV infection , and although the degree of upregulation is less pronounced than in mouse DCs , the kinetics are very similar ( Fig 3A ) . To assess the role of KDM5B in human DCs , the function of KDM5B was inhibited by treating the cells with 2 , 4-PDCA for 24 hours , followed by RSV infection . Similar to observations in mouse cells treated with 2 , 4-PDCA , inhibiting KDM5B in human MoDCs led to increased production of IFNB , TNF and IL6 compared to RSV alone or DMSO control ( Fig 3B ) . To determine whether the presence of the inhibitor affected the H3K4me3 status of these genes , a ChIP analysis for H3K4 was performed and examined the promoter regions of specific innate cytokine genes . Interestingly , MoDCs exhibited a slight increase in promoter methylation following incubation with the inhibitor alone , which was further increased when the cells were infected with RSV ( Fig 3C ) . Similar to observations in mouse DCs , no change in methylation at the IL10 an IL12 promoters was observed ( S4 Fig ) . Finally , infected MoDCs that had been treated with DMSO or 2 , 4-PDCA were cultured with autologous CD4+ T cells in the presence of RSV to assess the APC function of the MoDCs ( Fig 3D ) . While the T cells co-cultured with 2 , 4-PDCA-treated DCs had similar levels of IFN-γ production , the Th2 cytokines IL-5 and IL-13 were significantly decreased . These data suggest that inhibiting KDM5B function in human MoDCs results in regulation of Th2 cytokine production , supporting the hypothesis that RSV drives an altered immune phenotype that relies on epigenetic regulation of DC . Pro-inflammatory cytokines , including IFN-β , are important factors in driving Th1 responses following viral infection in vivo [26] . The above data indicated siRNA knockdown of Kdm5b led to increased pro-inflammatory cytokines in vitro , suggesting that priming the immune system in vivo with these DCs would drive a stronger Th1 response to RSV . To test this hypothesis , BMDCs were treated with Kdm5b-specific siRNA or scrambled control siRNA for 48 hours and subsequently infected overnight with RSV . These cells were then transferred intratracheally into naïve C57Bl/6 mice to prime the immune system in the context of Kdm5b-deficient DCs . One week after transfer , mice were challenged with RSV ( Fig 4A ) . Our lab has previously demonstrated that this sensitization protocol with myeloid DC elicits a pathogenic immune environment upon RSV reinfection , and that the RSV-infected DCs that are transferred begin migrating to the lymph nodes within 24 hours [27 , 28] . Eight days following challenge , the mediastinal lymph nodes ( MLN ) were removed and restimulated ex vivo to determine the effect that sensitizing mice with Kdm5b-knockdown DCs would have on the T cell response . The data indicated an increase in IFN-γ production from mice that had been sensitized with Kdm5b-specific siRNA treated DCs , accompanied by a significant decrease in IL-5 and a trend toward less IL-4 production ( Fig 4B ) . Furthermore , it was found that the lungs of mice primed with DCs that had been transfected with Kdm5b-specific siRNA had decreased inflammatory infiltrates compared to mice primed with control siRNA-treated DCs , as observed by H&E stained sections ( Fig 4C ) . High-power images revealed that the inflammation surrounding the airways of control RSV-infected DC-sensitized mice contained numerous eosinophils ( Fig 4C , insets ) . The number of eosinophils surrounding the airways was enumerated by microscopy in 100μm sections demonstrating decreased numbers in the Kdm5b siRNA group ( Fig 4D ) and confirmed with flow cytometric analysis of Gr-1+ and Siglec-F+ cells ( Fig 4E ) . Additionally , it was noted that the total number of cells in the lungs was increased in mice sensitized with control siRNA , but the lungs had fewer cells when the mice received Kdm5b siRNA ( Fig 4F ) . Flow cytometric analysis showed that these differences in total lung cells were due to fewer numbers of both conventional CD11c+CD11b+ DCs , as well as CD11c+CD103+ DCs ( Fig 4G ) [29 , 30] . Furthermore , there was a decrease in total CD4+ T cells as well as activated CD4+CD69+ cells ( Fig 4H ) . Increased numbers of total cells , as well as DCs , T cells and eosinophils were also observed in mice that where sensitized with uninfected DCs , but to a lesser degree . ( S5 Fig ) . In these mice , the numbers of inflammatory cells were similar to the numbers of cells measured in mice that received Kdm5b-deficient DCs , but was lower than the number of cells noted with control-treated , RSV-infected DCs were used to prime mice prior to challenge . Thus , the inhibition of Kdm5b in RSV-infected DCs led to significant protection from immunopathology that often has been associated with immunization to RSV [8] . Mucus production is a common occurrence during RSV infection , and is linked to the Th2 response to the virus . When mice were primed with Kdm5b-specific siRNA transfected DCs prior to RSV challenge , lower levels of the mucus-associated genes Muc5ac and Gob5 were observed in the lungs of mice primed with Kdm5b-specific siRNA ( Fig 4I ) . In addition , a primary inducer of goblet cell metaplasia and mucus hypersecretion , Il13 , was also reduced in mice primed by Kdm5b siRNA inhibited DC ( Fig 4I ) . Furthermore , mucus levels in the airways decreased , as visualized by PAS staining ( Fig 4J ) . Scoring of mucus production on a scale of 1–4 is quantified in Fig 4K and demonstrated a significant decrease in overall mucus production in the Kdm5b-inhibited DC transfer model . These results indicate that priming the immune response to RSV with DCs lacking Kdm5b results in a less pathogenic immune environment in vivo . To provide further genetic evidence for the role of Kdm5b , a DC-specific Kdm5b knockout mouse was developed . Kdm5bf/f mice [31] were crossed with CD11c-Cre+/- mice to create Kdm5b knockout CD11c+ DCs . Initial evaluation of these mice demonstrate no baseline differences in the number of immune cells in the lungs and spleens , including CD3+CD4+ and CD3+CD8+ T cells , CD11c+CD11b+ DCs and CD11b+F4/80+ macrophages ( S6 Fig ) . BMDCs from Kdm5bf/f-CD11c-Cre+ and control Kdm5bf/f-CD11c-Cre- mice that were grown had no identifiable growth or differentiation defects . BMDC from Kdm5bf/f-CD11c-Cre+ mice infected with RSV demonstrated increased expression levels of the innate cytokines Ifnb , Tnf and Il6 at 24 hours compared to Cre- controls ( Fig 5A ) , suggesting that KDM5B acts to suppress cytokine expression in DCs following RSV infection . These latter differences were not due to infectivity as there were no differences in the ability to infect the DCs from the Cre+ mice ( S7 Fig ) . These mice were then infected with RSV , and cytokine transcripts were measured in whole lung tissue early after infection . Innate cytokine transcripts were elevated in Cre+ mice over Cre- controls at both 2 and 4 days post-infection ( dpi ) ( Fig 5B ) . To determine whether this corresponded to the demethylase activity of KDM5B at the promoter regions of these genes , a ChIP assay in Kdm5b-deficient BMDCs and Cre- controls probing for H3K4 methylation was performed . Similar to observations using chemical inhibition of KDM5B , the Kdm5b-deficient DC had increased H3K4me3 at the promoter regions of all three inflammatory genes examined ( Fig 5C ) as assessed by ChIP analyses , but not of Il10 and Il12 ( S4 Fig ) . These results further support the notion that KDM5B demethylates H3K4 at specific innate cytokine promoters , thereby contributing to gene suppression . The previous data above suggested a decreased Th2 response to RSV infection when mice were sensitized with Kdm5b-knockdown DCs . However , to determine whether KDM5B is important in primary RSV infection , Kdm5bf/f-CD11c-Cre+ and control mice were infected with RSV . At 2 and 4 dpi , lungs were removed and RNA was isolated from homogenates . mRNA levels of the RSV proteins F , G and N were measured ( Fig 5D ) . Previous studies indicate that RSV replication peaks at 4 dpi [32]; no differences in RSV gene expression at 2 dpi were found , but by day 4 , viral gene expression was significantly increased in Kdm5bf/f-CD11c-Cre- mice , whereas viral clearance was enhanced in Kdm5bf/f-CD11c-Cre+ mice . These results were consistent with a plaque assay , where there were no differences in the levels of infectious virus at 2 dpi , but at 4 dpi the Kdm5bf/f-CD11c-Cre+ mice had significantly fewer infectious particles in the lungs compared to controls ( Fig 5E ) . At 8 dpi , MLN were removed and restimulated with RSV ex vivo . Similar to results shown in Fig 4 , the data indicated that Th2 cytokines IL-4 , IL-5 and IL-13 were substantially decreased in supernatants from restimulated cells , while IFN-γ production was increased , although not significantly ( Fig 5F ) . Expression of the mucus-associated genes Muc5ac and Gob5 in the lung tissue of infected mice was also measured . Kdm5bf/f-CD11c-Cre+ mice had lower expression of mucus-associated genes compared to the control Cre- mice ( Fig 5G ) . Visualization of mucus by histologic PAS staining demonstrated decreased mucus production in Kdm5bf/f-CD11c-Cre+ mice compared to controls ( Fig 5H ) . Slides were scored on a scale of 1–4 for mucus production , and quantification showed less mucus production in Kdm5bf/f-CD11c-Cre+ mice ( Fig 5I ) . Together , these data demonstrate that genetic deletion of Kdm5b from CD11c+ cells results in increased innate cytokine production by DCs and a correlative decrease in the Th2 response to RSV in vivo . The above studies found a decrease in proinflammatory cytokines in Kdm5bf/f-CD11c-Cre+ BMDCs similar to that observed with siRNA-treated BMDCs . To confirm that our results in the Kdm5bf/f-CD11c-Cre+ mice were mediated by DCs , BMDCs were infected with RSV for 24 hours , then delivered intratracheally into mice . After 7 days , mice were infected with RSV and assessed at 8 dpi . We found that mice that received Kdm5bf/f-CD11c-Cre- DCs had increased inflammation compared to RSV only controls , but that this inflammation was decreased in mice that were sensitized with Kdm5bf/f-CD11c-Cre+ DCs , as determined by cellular infiltrates and H&E staining ( Fig 6A ) . The MLN were removed and restimulated with RSV in vitro for 48 hours . Sensitizing mice with RSV-infected DCs led to increased levels of Th2 cytokines , IL-4 , IL-5 and IL-13 , but cytokine production was significantly lower when mice had been sensitized with Kdm5b/f-CD11c-Cre+ DCs compared to controls , although the decrease in IL-5 was not significant . These mice also had an increase , although not significant , in IFN-γ production from the lymph nodes ( Fig 6B ) . Kdm5bf/f-CD11c-Cre+ DCs sensitized mice had fewer numbers of both CD11c+CD11b+ and CD11c+CD103+ DCs as well as fewer CD4+ T cells in the lungs . Also , fewer activated CD4+CD69+ T cells were observed in the lungs of Cre+ sensitized mice compared to mice sensitized with control DCs ( Fig 6C and 6D ) . Finally , Kdm5bf/f-CD11c-Cre+ DC sensitized mice had less mucus production in the lungs compared to mice treated with control DCs , with a significant decrease in Gob5 expression ( Fig 6E and 6F ) . Together , these results confirm the role of Kdm5b in skewing the immune environment towards increased pulmonary pathology .
The objective of this study was to explore whether epigenetics are involved in determining the immune response during RSV infection . Previous studies have identified that DCs are a primary innate cell altered during RSV-induced pathogenesis , and that these cells play a central role in determining the nature of the immune responses . The concept that our immune system responses are influenced by environmental factors including microorganisms , pollution , diet and pathogen exposure is important as a backdrop for understanding disease progression [33–35] . Thus , we hypothesized that RSV infection of DCs would lead to the differential expression of epigenetic enzymes in these cells . The studies found that KDM5B , an H3K4 demethylase , was upregulated in DCs following RSV infection , but not when the cells were treated with ligands for TLR3 or TLR7 , which are known to be important in recognition of the virus [22 , 23 , 36] . As H3K4me3 is an activation mark , it was further hypothesized that inhibiting KDM5B , which would remove methyl groups and thus remove the activation mark , would alter the activation state of the DC . Indeed , proinflammatory cytokines expressed by DCs were increased when KDM5B was blocked , which in vivo led to a suppression of the Th2 phenotype often associated with RSV infection . Thus , the present study identified several important concepts; 1 ) Epigenetic enzymes are a part of the mechanism by which a pathogen can modify the immune response; 2 ) Specific epigenetic enzymes have distinct effects on the function of DCs; 3 ) By identifying and specifically blocking epigenetic enzyme function a profound effect can be observed on the pulmonary immune environment . These results are summarized in Fig 7 . Changes in the lung environment due to pathogen exposure have been well documented . Viral infections often induce long-term alterations in both immune function and lung physiology . For example , infection with influenza leads to a subsequent increase in susceptibility to bacterial infections , which is mediated in part by excessive IL-10 production in the lungs [37] and increased neutrophil apoptosis [38] . Conversely , influenza infection has been shown to be protective of the Th2 response associated with subsequent RSV infection in a mouse model [39] . Similarly , RSV infection also induces long-term changes in the lung environment , particularly when severe infection occurs early in life . Numerous studies have linked infection in infancy with childhood wheezing and the development of asthma [5 , 6] . While the mechanisms of this connection remain unclear , the cytokine environment that develops during infection may contribute to the pathology . In severe RSV infections during early childhood , the immune response in children requiring hospitalization is often characterized by the presence of Th2 cytokines and associated with increased eosinophil infiltration and mucus overproduction [40 , 41] . During early life , the immune system is predominantly biased toward a Th2 phenotype , and a Th1 balance does not develop until about one year of age [42] , which may contribute to the altered immune response to RSV infection in the lungs of infants . The data presented in these studies highlight a potential Th2 reinforcing response that is induced by RSV in DC that would limit the ability of an individual to develop a more appropriate , less pathogenic anti-viral response . This specific epigenetic mechanism , elicited by KDM5B , may be but one contributing factor that influences RSV pathogenesis . RSV is relatively poor at inducing type I IFNs , especially IFN-β , compared to other viruses including influenza [9 , 43–47] . The production of IFN-β from DCs has been shown to be important in directing DC maturation and cytokine production , indicating that increasing IFN-β production by DCs could alter the immune environment following RSV infection [48] . Previous studies have highlighted the importance of IFN-β production in developing an antigen-specific response to RSV , and have found that STAT1-deficient mice , which cannot initiate signaling from either type I or type II IFN , have significantly increased Th2 responses , as well as increased illness [26 , 49] . On the other hand , IFN-γ-deficient mice are protected , indicating that type I IFNs are important for developing an appropriate anti-viral response to RSV [26] . Furthermore , IFN-β and TNF-α are important in establishing a Th1 response in humans , thus increasing levels of these cytokines could help skew an effective immune response to RSV [50 , 51] . In support of these previous findings , the present study demonstrates that a primary consequence of removing KDMB from DC is the increase in both IFN-β and TNF-α , resulting in reduced Th2 pathology . Alternatively , increasing proinflammatory cytokines from DCs that lack functional KDM5B protein resulted in increased viral clearance from the lungs following primary RSV infection . The resulting decreased antigen levels in the lungs may potentially lead to DCs being able to preferentially drive a Th1 phenotype over Th2 . However , as there were no differences in the antigen presenting ability of DCs deficient in Kdm5b , it is likely that the cytokines produced from the DCs are the primary factor that leads to decreased Th2 pathology . Dendritic cells are the primary cell type responsible for stimulating CD4+ T cells and directing the adaptive immune response . Many of the long-term immunologic changes that occur in the lungs following RSV infection are due to alterations in DC function [27 , 28 , 46] . Although RSV nonstructural proteins ( NS1 and NS2 ) have been specifically implicated in the inhibition of type I IFN production [9 , 52–57] , the molecular mechanisms have not been fully explored . Recent studies in epigenetic mechanisms have led to a greater understanding of the molecular control of immune cells in the context of infectious agents . It has been shown that long-term changes in DC function and cytokine production following an inflammatory insult are due in part to epigenetic changes in the Il12 gene [17] . Others have demonstrated that the ability of DCs to prime Th1 responses is due to histone deacetylase activity [58] . More specifically , epigenetic enzymes can regulate the antiviral response by controlling the interferon pathway , especially IFN-stimulated genes [59] . This modulated immune effect was shown to be mediated by the G9a/GLP enzymatic complex [59] . In our studies , targeted gene arrays identified epigenetic enzymes that may contribute to the altered DC response following RSV infection . By using this unbiased approach , we targeted KDM5B as an important modulator of DC innate cytokine responses , as increased Kdm5b expression correlated with decreased Ifnb expression . By specifically inhibiting KDM5B and by generating a novel CD11c ( DC ) targeted KO mouse , studies identified that KDM5B has a critical role for altering DC function . Even more striking was the fact that the single enzyme alteration had a pathogenic impact on the anti-RSV response , including modifying T cell cytokine profiles and development of goblet cell metaplasia and mucus hypersecretion . In summary , our studies identify a unique mechanism by which RSV infection influences how DCs are activated and subsequently activate the T cell responses leading to increased pulmonary pathogenesis . These studies used three methods to inhibit the expression or function of Kdm5b-siRNA knockdown , chemical inhibition and genetic deletion–and in all cases increased levels of innate cytokine production in DCs that subsequently led to a decrease in the Th2 response was observed . While previous attempts to create vaccines have led to increased Th2 pathology , these studies argue that future vaccine designs may need to consider the epigenetic programs that control pro-inflammatory cytokine production by DCs related to a protective immune response .
All experiments on animal were performed in compliance with the guidelines of the Office of Laboratory Animal Welfare from the National Institutes of Health . Procedures were approved by the University Committee on the Use and Care of Animals at the University of Michigan ( protocol PRO00004817 , exp . 04/03/2016 ) . All protocols using human cells were approved by the University of Michigan Institutional Review Board . For all experiments using human subjects , written informed consent was obtained from all donors . C57BL/6 mice were purchased from Jackson Laboratory ( Bar Harbor , ME , USA ) . Kdm5bfl/fl mice were previously described [31 , 60] , and were crossed with B6 . Cg-Tg ( Itgax-cre ) 1-1Reiz/J ( CD11c-Cre ) mice ( Jackson Laboratory ) . Mice were infected intratracheally with 1 x 105 pfu of RSV strain A2001/2-20 , a clinical isolate originally from Vanderbilt University , was propagated as described [32] . Viral stocks were grown in Hep-2 cells and concentrations determined by plaque assay . In some experiments , cultured dendritic cells were infected with RSV for 24 hours , then thoroughly washed and 2 . 5 x 105 cells were transferred intratracheally into naïve mice . Whole lungs were harvested at two and four days post infection and were ground with a mortar and pestle . Supernatants were diluted and incubated with Vero cells for four days . RSV plaques were detected using a specific polyclonal antibody ( Millipore , Billerica , MA ) . Bone marrow was collected by flushing the femur and tibia of hind legs with PBS + 1% fetal calf serum ( FCS ) . BMDCs were grown in RPMI 1640 supplemented with 10% FCS , L-glutamine , penicillin/streptomycin , non-essential amino acids , sodium pyruvate , 2-mercaptoethanol ( ME ) and 10 ng/ml of recombinant murine granulocyte macrophage-colony stimulating factor ( GM-CSF; R&D Systems , Minneapolis , MN , USA ) . Cells were fed on days 3 and 5 with fresh GM-CSF . On day 6 , cells were cultured with RSV ( MOI = 1 ) , 20 μg/ml of poly ( I:C ) or 1 μg/ml of imiquimod . For some experiments , cells were first treated for 24 hours with 1 mM 2 , 4-PDCA ( Sigma , St . Louis , MO , USA ) , a chemical inhibitor of KDM5B [24 , 25] , or with DMSO ( 0 . 1% ) as a control , then were infected with RSV . Fifty to 70 ml of blood was taken by venous puncture . Peripheral blood mononuclear cells were isolated from heparinized blood by Ficoll-Paque ( GE Healthcare ) purification . Briefly , blood was diluted 1:1 with sodium chloride and tubes were centrifuged at 400g for 30 minutes with the brake off . Cells were harvested from the interface of the Ficoll layer , and were washed and enumerated . Monocytes were isolated using anti-CD14 microbeads , according to the manufacturer’s instruction ( Miltenyi Biotec , San Diego , CA , USA ) . Cells were cultured in RPMI 1640 with 10% human serum and L-glutamine , pen/strep , non-essential amino acids , sodium pyruvate , 2-ME , 40 ng/ml of recombinant human IL-4 and 40 ng/ml of recombinant human GM-CSF ( both from R&D Systems ) . Cytokines were replenished on days 3 and 6 , and cells were used on day 7 . For some experiments , CD4+ T cells were isolated from the CD14- fraction using the T cell isolation II kit ( Miltenyi Biotec ) . Cells were then cultured with MoDCs in the presence or absence or RSV . At 48 hours , RNA was extracted and message levels of IFN-γ , IL-5 and IL-13 were determined by qPCR . BMDCs were transfected with siRNA at day 6 of culture . 1 μM of ON-TARGETplus or scrambled control siRNA ( Dharmacon , Pittsburgh , PA , USA ) was transfected using the Amaxa DC nucloefection kit and an Amaxa Nucleofector ( Lonza Inc , Cologne , Germany ) . Cells were then cultured for 48 hours in complete medium and knock-down was assessed by qPCR . Cells were then infected with RSV for 24 hours . RNA was extracted using TRIzol reagent ( Invitrogen , Carlsbad , CA , USA ) or RNeasy Mini Kit followed by the Cleanup Kit ( Qiagen , Germantown , MD , USA ) and following the manufacturers instructions . RNA isolated from tissues was first homogenized . Complementary DNA was synthesized using murine leukemia virus reverse transcriptase ( Applied Biosystems , Foster City , CA , USA ) and incubated at 37°C for one hour , followed by 95°C for 5 minutes to stop the reaction . Real-time quantitative PCR was multiplexed using Taqman primers with a FAM-conjugated probe and GAPDH with a VIC-conjugated probe ( Applied Biosystems ) to measure transcription of Il6 , Tnf , Il13 , Il4 , Ifng and Kdm5b . Fold change was quantified using the 2-ΔΔCT method . Custom primers were designed to measure Ifnb , Muc5ac , Gob5 and RSV F , G and N RNA levels . All reactions were run on an ABI Prism 7500 Sequence Detection System or ViiA 7 Real Time PCR System ( both from Applied Biosystems ) . Lungs were removed at 8 days post infection . The large left lobe of each lung was inflated by injection with 4% formaldehyde . Lungs were embedded in paraffin , and 5 μm sections were sectioned and stained with hematoxylin and eosin ( H&E ) to visualize inflammatory cells or with periodic acid-Schiff stain ( PAS ) to visualize mucus production . To score mucus production , slides were evaluated by a blinded observer . Sections were scored based on the following scale: 1 –minimal , 2—slight , 3 –moderate , 4 –severe . Lungs and mediastinal lymph nodes were removed and single cells were isolated by enzymatic digestion with 1 mg/ml collagenase A ( Roche , Indianapolis , IN , USA ) and 20 U/ml DNaseI ( Sigma ) . Cells were resuspended in PBS with 1% FCS and Fc receptors were blocked with purified anti-CD16/32 ( clone 93; BioLegend , San Diego , CA , USA ) . Surface markers were identified using antibodies ( clones ) against the following antigens , all from BioLegend , unless otherwise specified: CD11c ( N418 ) , CD11b ( M1/70 ) , CD103 ( 2E7 ) , CD3 ( 145-2C11 ) , CD4 ( RM4-5 ) , CD69 ( H1 . 2F3 ) , I-A/I-E ( M5/114 . 15 . 2 ) , CD80 ( 16-10A1 ) , CD86 ( GL-1 ) , Gr-1 ( RB6-8C5; eBiosciences , San Diego , CA , USA ) and SiglecF ( E50-2440; BD Biosciences , San Jose , CA , USA ) . Mediatstinal lymph nodes were removed and single cells were isolated by enzymatic digestion . 5 x 105 cells were plated in 200 μl of complete medium and were restimulated with RSV for 48 hours . Supernatants were collected and levels of the cytokines IL-4 , IL-5 , IL-13 and IFN-γ were measured by Bioplex assay ( Bio-Rad ) . Chromatin immunoprecipitation ( ChIP ) was performed using an assay kit ( Millipore ) with minor modifications . Briefly , cells were fixed in 1% formaldehyde , then lysed in SDS buffer . Cells were then sonicated using a Branson Digital Sonifier 450 ( VWR , West Chester , PA , USA ) to create 200–1000 bp fragments . The lysate was clarified by centrifugation , and 5% of the supernatant was saved to measure the input DNA . The remaining chromatin was incubated with 1 μg of anti-H3K4me3 antibody ( Abcam ) or control IgG ( Millipore ) and incubated at 4°C with rotation overnight . Immune complexes were precipitated with salmon sperm DNA/protein A agarose beads . Crosslinking was reversed by incubation at 65°C and samples were treated with proteinase K . DNA was purified by phenol:chloroform:isoamyl alcohol separation and ethanol precipitation . Primers for the promoter regions of IFN-β , TNF-α and IL-6 were designed using Lasergene software and DNA was amplified by qPCR using SYBR Green buffer ( Applied Biosystems ) . Results are expressed and mean ± SE . Statistical significance was measured first by one-way or two-way ANOVA as appropriate , followed by a Student Neuman Keuhl’s post-hoc t test . A p value of <0 . 05 was considered significant .
|
Respiratory syncytial virus ( RSV ) is a significant public health concern . Nearly all children are infected by two years of age , and severe infection often results in hospitalization . There is no vaccine for RSV , and previous attempts have resulted in increased disease severity in immunized children once they were exposed to the virus . Therefore , a better understanding of how RSV directs the immune response is needed . In this study , we found that the protein KDM5B regulates an epigenetic mechanism that directs the immune response to RSV . KDM5B suppressed the activation of key antiviral signals in dendritic cells , and inhibition of KDM5B led to gene activation and increased antiviral function . This correlated with decreased pathology in the lungs . Therefore , our data suggest that new attempts at designing a vaccine should consider the effects of vaccination on dendritic cells , and should consider strategies that will increase antiviral signals from dendritic cells .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
RSV-Induced H3K4 Demethylase KDM5B Leads to Regulation of Dendritic Cell-Derived Innate Cytokines and Exacerbates Pathogenesis In Vivo
|
Hepatitis A virus ( HAV ) , an enigmatic and ancient pathogen , is a major causative agent of acute viral hepatitis worldwide . Although there are effective vaccines , antivirals against HAV infection are still required , especially during fulminant hepatitis outbreaks . A more in-depth understanding of the antigenic characteristics of HAV and the mechanisms of neutralization could aid in the development of rationally designed antiviral drugs targeting HAV . In this paper , 4 new antibodies—F4 , F6 , F7 , and F9—are reported that potently neutralize HAV at 50% neutralizing concentration values ( neut50 ) ranging from 0 . 1 nM to 0 . 85 nM . High-resolution cryo-electron microscopy ( cryo-EM ) structures of HAV bound to F4 , F6 , F7 , and F9 , together with results of our previous studies on R10 fragment of antigen binding ( Fab ) -HAV complex , shed light on the locations and nature of the epitopes recognized by the 5 neutralizing monoclonal antibodies ( NAbs ) . All the epitopes locate within the same patch and are highly conserved . The key structure-activity correlates based on the antigenic sites have been established . Based on the structural data of the single conserved antigenic site and key structure-activity correlates , one promising drug candidate named golvatinib was identified by in silico docking studies . Cell-based antiviral assays confirmed that golvatinib is capable of blocking HAV infection effectively with a 50% inhibitory concentration ( IC50 ) of approximately 1 μM . These results suggest that the single conserved antigenic site from complete HAV capsid is a good antiviral target and that golvatinib could function as a lead compound for anti-HAV drug development .
Over the past 2 decades , progress in understanding human infections caused by hepatitis A virus ( HAV ) has been eclipsed by the priority of combating persistent hepatitis B virus ( HBV ) and hepatitis C virus ( HCV ) infections . HAV , the most important agent for enterically transmitted viral hepatitis , is distributed worldwide and infects all age groups [1] . The global burden of HAV has not abated . Approximately 1 . 5 million clinical cases of HAV occur annually despite the availability of an effective vaccine [2 , 3] . Hepatitis A as an infectious disease strongly correlates with income , hygiene , and living conditions [4] . Areas with poor hygiene and living conditions continue to be under constant threat of HAV outbreaks [4] . More recently , HAV has also started to become a new public health concern in well-developed , economically advanced countries due to the lack of natural or vaccine-induced acquired immunity to HAV in many adults [5 , 6] . In the past year , more than 649 people throughout California have been reported to be infected with HAV . Among these , 417 required hospitalization , and 21 patients died , making this the largest outbreak in the United States in the past 20 y [7] . Development of antiviral therapy against HAV infection is urgently needed . HAV , transmitted via the fecal–oral route , is a positive-sense , single-stranded RNA icosahedral virus belonging to the genus Hepatovirus within the Picornaviridae family [8] . The 7 . 5 kb genome of HAV contains a single open reading frame ( ORF ) that encodes a giant polyprotein [9] . The polyprotein is processed by a viral protease ( 3Cpro ) into 3 polypeptide intermediates , namely , P1–P3 [9] . P1 is subsequently further processed into 3 structural proteins , VP0 ( a precursor for VP2 and VP4 ) , VP3 , and VP1 , which self-assemble into a spherical capsid with icosahedral symmetry [10] . Five copies of the VP1 capsid protein surround the icosahedral 5-fold axes . Three copies of VP2 and VP3 alternate at the 3-fold axes , and 2 copies of VP2 abut each other at the 2-fold axes [11] . Although a limited number of antigenic sites located on the HAV capsid have been revealed by escape mutants , the antigenicity of HAV is largely uncharacterized [12 , 13] . Our recent study involving the structure of a complex of HAV with its neutralizing monoclonal antibody ( NAb ) , R10 , extended the previously unreported VP2 antigenic sites [14] . Unlike other picornaviruses , HAV is extremely stable , both genetically and physically . So far , 6 genotypes of human HAV have been identified [15] but with only a single serotype , suggesting that HAV has highly conserved antigenic sites [16 , 17] . The low antigenic variation might be attributed to its highly deoptimized codon usage [18] . A systematic and comprehensive study of the antigenic characteristics of HAV and neutralizing mechanisms could facilitate the design of effective small-molecule antivirals targeting HAV . We set out to clarify the molecular basis for the antigenicity of HAV by characterizing 4 NAbs with varying neutralizing activities against the virus . We sought this information for rationally designing antiviral inhibitors . Here , we report the characterization of 4 highly potent NAbs: F4 , F6 , F7 , and F9 . Furthermore , we have analyzed the experimentally derived high-resolution structures of HAV bound to the 4 NAbs as well as the previously reported R10-HAV structure to identify conserved epitopes for gaining key structure-activity correlates . Using a robust in silico docking method , we have screened the DrugBank Database and have identified 1 promising inhibitor named golvatinib . Cell-based antiviral assays have confirmed the ability of golvatinib to block infections caused by HAV .
To shed further light on the nature of the antigenicity of HAV , 2 rounds of monoclonal antibodies ( mAbs ) were generated . R10 , an NAb with a 50% neutralization concentration value ( neut50 ) of approximately 2 nM [14] , was produced in the first round , and over 30 mAbs were screened during the second round . Of these later antibodies , 4 antibodies , named F4 , F6 , F7 , and F9 , are NAbs . Surface plasmon resonance ( SPR ) experiments showed that the 5 NAbs bind to HAV with a high affinity in the nanomolar range ( Fig 1A , S1 Fig ) . A number of NAbs with similar affinities are virus specific ( e . g . , dengue virus-specific human mAb 5J7; Japanese encephalitis virus-specific mAbs 2F2 and 2H4 ) and exhibit exceptionally potent neutralizing activities [19 , 20] . To investigate whether these 5 NAbs recognize different epitopes or the same patch of epitopes , we performed a competitive binding assay . Briefly , the CM5 chip ( BIAcore , GE Healthcare ) , fully occupied with HAV , was initially saturated with R10 , and additional binding with another NAb was evaluated . The CV60 mAb ( an mAb against Coxsackievirus A16 ) was used as a negative control . Binding of R10 blocks the attachment of other 4 NAbs to HAV ( Fig 1B ) , suggesting that these 5 NAbs may bind to the same patch of epitopes or at least partially overlapped epitopes . To characterize neutralizing activities , these 5 NAbs were evaluated for their abilities to prevent HAV infection . Of note , all 4 NAbs generated from the second round showed potent neutralizing activities , of which F6 exhibited the strongest neutralizing activity ( a neut50 value of approximately 0 . 1 nM , which was 20-fold more potent than R10 [Fig 1C] ) . To define precisely the atomic determinants of the interactions between these 4 NAbs and HAV , structural investigations of HAV in complex with fragment of antigen binding ( Fab ) from its NAbs were carried out . Cryo-EM micrographs of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV complexes were recorded using a Titan Krios electron microscope ( Thermo Fisher ) equipped with a Gatan K2 detector ( Gatan , Pleasanton , CA ) ( S2 Fig ) . The structures of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV were determined at resolutions of 3 . 90 , 3 . 68 , 3 . 05 , and 3 . 79 Å with 4 , 536 , 7 , 245 , 16 , 743 , and 3 , 798 particles , respectively , by single-particle techniques using the gold-standard Fourier shell correlation = 0 . 143 criterion [21] ( Fig 2A , Table 1 , S3 Fig and S4 Fig ) . Densities attributable to residue backbones and side chains were recognizable in maps ( Fig 2A–2C ) . These maps were of sufficient quality to allow the atomic modelling of most of the HAV capsid proteins and NAb Fab . The structures of these 5 complexes are almost indistinguishable . Differences are observed in the residues of the common complementary determining regions ( CDRs ) of NAbs ( r . m . s . d . for 12 , 473 Ca atoms less than 1 . 25 Å ) , which are consistent with the results of the competitive binding assays ( Fig 2A and S5 Fig ) . There are 60 copies of NAb Fabs ( probably fully occupied ) bound to the virus in accordance with the level of electron density for the Fab ( Fig 2A ) . Possibly correlated with its unusual stability , HAV capsid proteins exhibit no notable conformational changes upon binding to any NAbs . Unlike EV71 or other picornaviruses in which several distinct patches for neutralizing epitopes have been reported [20 , 22 , 23 , 24] , all 5 NAb Fabs encircle edges of the pentameric building blocks of the virus , between the 2-fold and 3-fold axes ( Fig 2A and Fig 3A ) . Examination of the possibility of binding of 2 arms of an immunoglobulin-G ( IgG ) molecule to the HAV surface showed that any 2 adjacent Fabs binding to the capsid could indeed mimic the 2 arms of a single IgG molecule ( S6 Fig ) . Therefore , the IgG avidity for all 5 NAbs might be observed due to 2 Fab arms of an IgG on the surface of HAV being sufficiently close . To explore the mechanism of neutralization , real-time reverse transcription PCR ( RT-PCR ) assays were performed to quantify the virus remaining on the cell surface , following exposure to antibodies’ previrus attachment to cells at 4°C . The results reveal that these NAbs prevent HAV attachment to the permissive 2BS cell surface ( S7 Fig ) . In summary , the high potencies of all 5 NAbs could be due to several reasons , including ( 1 ) higher avidity of the bivalent form of antibody , ( 2 ) the ability of the bivalent antibody to aggregate virus particles [14] , and ( 3 ) efficient block viral attachment to the host cell . As expected , all Fabs exhibit a similar mode of binding , in which 1 Fab binds across the interface between the pentamers , interacting with VP2 and VP3′ from different pentamers ( Fig 2 ) . The footprints of the 5 NAbs cover interaction areas ranging from approximately 970 Å2 to 1 , 290 Å2 , of which approximately 60% ( approximately 630 Å2 ) and approximately 40% ( approximately 440 Å2 ) are contributed by the heavy-chain and light-chain variable domains , respectively ( Fig 3A ) . In line with this observation , F6 epitope contains more amino acid residues than other epitopes ( S1 , S2 , S3 , and S4 Tables ) , which is consistent with the results of binding affinities and neutralizing activities ( Fig 1 ) . Given the fact that the F6 exhibits the most potent antiviral activity , the epitope analysis of these 5 NAbs is representative of the F6 mAb . The heavy chain predominantly binds to the BC loop and EF loop of VP3 , whereas the light chain binds to the BC loop of VP2 and the BC loop of VP3 ( Fig 3B and S1 Table ) . The epitopes on HAV capsid include residues S65 , R67 , and T71 in the BC loop and A198 and S201 of VP2; A68 , S69 , D70 , S71 , V72 , G73 , Q74 , Q75 , K77 , and V78 in the BC loop of VP3; and L141 , D143 , T145 , G146 , I147 , T148 , L149 , and K150 in the EF loop of VP3 ( Fig 3B and S1 Table ) . The region of the F6 Fab that binds HAV comprises 4 of the 6 common CDRs: H1 ( residues 28–32 ) , H2 ( residues 52–57 ) , H3 ( residues 100–106 ) , and L1 ( residues 30–31 ) with , unusually , additional interactions contributed by the light-chain framework region ( L-FR; residues 45–55; Fig 3C and S1 Table ) . The antibody components of these interactions include residues Y31 , R45 , Y48 , S51 , R52 , L53 , D55 , and Q59 from the light chain and residues N28 , Q30 , H31 , Y32 , Q52 , T53 , N54 , T56 , Y57 , R98 , N101 , I102 , E103 , C104 , H105 , and Y106 from the heavy chain ( Fig 3C ) . Tight binding between the F6 fab and HAV capsid is facilitated by 33 hydrogen bonds and 9 salt bridges ( Fig 3C ) . Structures of HAV in complex with 5 NAbs reveal that epitopes on HAV locate within the same patch and are extremely conserved ( Fig 4A ) , which is substantially different when compared to other picornaviruses , e . g . , at least 4 regions of the epitopes recognized by its NAbs are mapped in EV71 [26 , 27] . In line with neutralizing activities , F6 epitope has 3 extra residues ( D70 , K77 , and L141 of VP3 ) when compared to others , and R10 possesses the least number of epitope residues ( all the NAbs recognized the A198 of VP2 except R10 ) ( Fig 4A ) . These 5 NAbs share high sequence similarities at the framework region but bear relatively low sequence identities ( approximately 35% ) at the CDRs ( Fig 4B ) . In spite of variations in the sequences , these 5 CDRs involved in the interactions with HAV adopt an indistinguishable configuration and a similar binding mode ( Fig 2 , Fig 4B , and S4 Fig ) . As expected , further tight binding of F6 and F4 to HAV is made possible by the additional hydrogen bonds and charge interactions formed by the antibodies ( Fig 4C ) . Furthermore , residues in R10 that interact with HAV are also fewer in number than those observed for other NAbs . To decipher the structure activity correlates between HAV-NAb interactions and neutralizing activities , the interaction interface areas and binding energies were calculated and then compared with their neutralizing activities ( S5 Table ) . We assembled a data set of 5 NAbs inhibition data for HAV and generated correlation plots between the Neut50 values and the area and energy of interaction , which produced a compelling correlation of 0 . 93 and 0 . 94 , respectively ( Fig 4D–4E ) . These analyses suggest two lessons: ( 1 ) epitopes revealed by NAbs on HAV are good targets for drug design; and ( 2 ) the more robust binding of NAbs to the epitopes , the better the antiviral activities . To date , 6 genotypes of human HAV have been identified but with only a single serotype . This indicates that these 5 NAbs are likely to bind strongly to the 6 human HAVs and could be capable of preventing human HAV infections . Sequence and structural analyses show that the residues constituting the epitopes are 87 . 5% identical and 94 . 6% conserved , with only 3 out of 21 contacting residues ( residue 67 of VP2 , residues 145 and 146 of VP3 ) being moderately conserved ( 50%–85% ) , and the remaining residues completely conserved ( 100% ) ( Fig 5A and Fig 5B ) . The variation rate for the epitopes is even slightly lower than that for the whole capsid ( P1 , approximately 94 . 3% conserved ) , highlighting a single , conserved antigenic site for HAV . Given the fact that a single , conserved antigenic site exists in HAV and the key structure-activity correlates based on the antigenic site have been established , we next used the structural data to rationally design and screen potent compounds against HAV targeting the antigenic site . These residues composing the antigenic site are distributed on both sides of a long “gully , ” which forms a potential inhibitor binding pocket ( Fig 6A ) . Results of our previous study have also indicated that the “gully” area might be critical for HAV receptor binding [14] . We postulated , on the basis of inspection of the HAV-NAbs binding interface , that a tight binder ( a compound ) mimicking the NAbs might efficiently block HAV entry and infection . To test this hypothesis , we scanned in silico the DrugBank database ( https://www . drugbank . ca/ ) , using Phase version 3 . 7 [30] , Glide version 6 . 1 [31] to identify potential tight binders . Briefly , the 4 key residues 31 , 32 , 101 , and 102 from the heavy chain of the 5 NAbs , which made the greatest contributions to the specific action of antigen–antibody , were selected as a reference structure for pharmacophore modeling . The generated pharmacophore was used to screen the drugs database of DrugBank . A total of 2 , 588 drugs were screened . Then , all selected drugs were docked to the antigenic site , and the top-ranked 4 molecules were selected . Distinct from the others , compound 3 has the best glide score ( S6 Table ) . Therefore , compound 3 , named golvatinib ( DB11977 ) , was predicted to bind to the “gully” much stronger than others ( Fig 6B ) . As expected , in this docking pose , golvatinib contacts with the epitope residues , including S65 from VP2 and V72 , G73 , Q74 , Q75 , V78 , P79 , T144 , T148 , L149 , and Q246 from VP3 , via hydrophilic and hydrophobic interactions ( Fig 6A ) . We therefore also measured the inhibitory activities of golvatinib by in vitro studies in 2BS cells . We used 100 50% tissue culture infective dose ( TCID50 ) virus in the presence of different concentrations of the compounds and exposed control wells to the equivalent concentration of solvent ( DMSO ) to ensure no effects on uninfected cells or on virus titer ( S8 Fig ) . The compound golvatinib exhibited a potent antiviral activity , with a 50% inhibitory concentration ( IC50 ) of approximately 1 μM , inhibiting the viral titer to below 15% at concentrations over 8 μM ( Fig 6B ) . Meanwhile , no notable cytotoxic effect of golvatinib at concentrations of 0 . 0005 to8 μM was observed ( Fig 6B ) . The measured antiviral activity of golvatinib is in agreement with that predicted in silico . As expected , golvatinib , like the 5 NAbs , inhibits HAV infection by blocking attachment to the host cell ( S9 Fig ) . Due to the partial overlapped binding sites of the NAbs and of golvatinib ( Fig 6A ) , it is quite possible that they are capable of competing each other to attach the HAV surface . In addition , the binding of golvatinib to the HAV does not alter its particle stability ( S9 Fig ) , which is consistent with our previous results that stabilization or destabilization is unlikely to be the major neutralization mechanism in our study systems [14] .
Attachment of the virus to its cellular receptors located on the surface of the host cell and uncoating of the virus leading to the release of the viral genome into host cells are regarded as the 2 key steps for the successful entry of nonenveloped viruses , including picornaviruses , into host cells [32] . Neutralizing antibodies block the entry of viruses into host cells by blocking the attachment of the virus to the cellular receptor [33] , overstabilizing the virus [34] , preventing the release of viral genome [27] , or physically destabilizing the capsid of the virus [22] . In our previous study , we demonstrated that R10 , a HAV-specific neutralizing antibody , neutralizes HAV infection by preventing the binding of the virus to its putative receptor T-cell immunoglobulin and mucin-containing domain 1 ( TIM-1 ) [14] . Recent evidences suggests that TIM-1 is not an essential receptor for the naked ( unenveloped ) HAV but rather an attachment factor for quasi-enveloped virions [35] , making the bona fide receptor ( s ) elusive . Therefore , it is challenging to verify whether the binding of NAbs or golvatinib blocks the interactions between HAV and its bona fide receptor . Many picornaviruses use cell-surface molecules belonging to the immunoglobulin superfamily ( IgSF ) as their cellular receptors , which usually consist of tandem repeats of between 1 and 5 Ig-like domains to interact with viruses [36] . Given the fact that R10 competitively blocked TIM-1 Ig V binding to HAV [14] , it is possible that the bona fide receptor might be from the IgSF . In this study , the antigen binding site of the newly screened NAbs ( F4 , F6 , F7 , and F9 ) maps to the same epitope on the surface of HAV as that identified for R10 , suggesting that the binding sites of the 4 NAbs and the bona fide receptor may overlap . Previous studies have shown that residues S102 , V171 , A176 , and K221 of VP1 and D70 , S71 , Q74 , and 102–121 of VP3 are part of the neutralizing epitopes [13] . However , structural analysis reveals that these putative epitope residues are forming 2 clusters that are separated by a distance of 40 to 50 Å on the HAV surface , suggesting that these residues are unlikely to form a single antigenic site . The high-resolution structures of HAV in complex with 4 NAbs described in this study coupled with the results of our previous studies on HAV-antibody complexes [14] further verify the fact that VP2 ( but not VP1 ) , as well as VP3 , form a single , conserved antigenic site on the surface of HAV , which differs radically from the architecture of the antigenic sites of other picornaviruses [37 , 38] . However , our studies cannot exclude the possibility of the likely existence of a second neutralizing antigenic site involving residues in VP1 , yet ill-defined on the viral capsid . Additionally , the neutralizing epitopes should differ with those of binding but non-neutralizing antibodies , which needs to be further investigated . The single , conserved antigenic site we identify could serve as an excellent target for structure-based drug design . Although hepatitis A is a vaccine-preventable disease [39] , an anti-HAV drug would be indispensable for treating fulminating infections . In this study , about 2 , 588 drug candidates ( compounds ) from the DrugBank database were selected for in silico docking studies . One of the candidates predicted to interact with the conserved antigenic site by the docking studies exhibited excellent antiviral activity without any notable cytotoxicity . Therefore , based on the preclinical evaluation of its cytotoxicity and pharmacodynamics , golvatinib , previously investigated for the treatment of platinum-resistant squamous cell carcinoma of the head and neck [40] , could act as a lead compound for anti-HAV drug development . In summary , we have used a combined experimental and computational approach starting from a number of NAbs targeting a single , conserved antigenic site located on the surface of a complete viral capsid to obtain , in a single round of design , a potent micromolar-range drug candidate that is effective and safe and has many drug-like properties .
Animals were bred and maintained under specific pathogen-free ( SPF ) conditions in the institutional animal facility of the Institute of Biophysics , Chinese Academy of Sciences . All animal experiments were performed with protocols ( protocol numbers VET102 , VET201 , VET203 , and VET301 ) approved by the Animal Care and Use Committee of Institute of Biophysics , Chinese Academy of Sciences . HAV virus genotype TZ84 ( HAV IA genotype ) was used to infect 2BS cells at a multiplicity of infection ( MOI ) of 0 . 2 at 34°C . Particle production and purification have been described previously [11] . F4 , F6 , F7 , and F9 were purified from mouse ascites with a protein A affinity column ( GE ) . The Fab fragment was generated using a Pierce FAB preparation Kit ( Thermo Scientific ) , according to the manufacturer’s instructions . Briefly , after removal of the salt using a desalting column , the antibody was mixed with papain and then digested at 37°C for 6 h . The Fab was separated from the Fc fragment by using a protein A affinity column . Then , Fab was loaded onto a Hitrap Q FT column ( GE ) . Fractions corresponding to the major peak were collected and concentrated for cryo-EM analysis . The binding affinities of the 5 NAb assays were determined by SPR . These experiments were performed using a BIAcore 3 , 000 machine ( BIAcore , GE Healthcare ) in the buffer solution containing 10 mM HEPES ( pH 7 . 4 ) , 150 mM NaCl , and 0 . 005% v/v Tween 20 at 25°C . The purified HAV full particles were directly immobilized onto CM5 sensor chips ( BIAcore , GE Healthcare ) at concentrations equivalent to approximately 950 response units ( 0 . 3 mg/ml ) . Subsequently , gradient concentrations ( 0 . 0315 , 0 . 0625 , 0 . 125 , and 0 . 25 μM ) of purified Fab fragments of F4 , F6 , F7 , F9 , and R10 were used to flow over the chip surface . To regenerate the chip , 100mM NaOH was used . The binding affinities were analyzed using steady state affinity with the software BIAevaluation version 4 . 1 . Binding competition between HAV antibodies was determined using SPR ( BIAcore 3 , 000 , GE ) . The entire experiment was performed at 25°C in the buffer solution containing 10 mM HEPES ( pH 7 . 4 ) , 150 mM NaCl , and 0 . 005% v/v Tween 20 . The CM5 biosensor chip ( BIAcore , GE Healthcare ) immobilized with HAV full particles ( 0 . 3 mg/ml ) was first saturated with R10 for 5 min . Afterward , the other NAbs were injected in the presence of R10 for another 3 min . CV60 , an irrelevant antibody , was used as a negative control . Except for R10 , all other NAbs were evaluated at a concentration of 300 nM for saturation . R10 was applied at a concentration of 900 nM . The chip was regenerated with 100mM NaOH ( GE Healthcare ) . For the neutralization assay , purified mAbs at a concentration of 0 . 2 mg/ml were initially diluted 8-fold as stocks and then serially diluted 2-fold with DMEM containing 2% FBS; 100 μl of 2-fold antibody dilutions were mixed with 100 μl of HAV containing 100 TCID50 for 1 h at 37°C and then added to monolayers of 2BS cells in cell culture flasks ( T25 CM2 ) . Meanwhile , maintaining medium was provided as well . Each dilution was replicated 3 times along with one control that contained no antibody dilution . After 21 d of growth at 34°C , the medium was removed , and the cells were washed three times using PBS buffer; 1 ml of Trypsin/EDTA was added , and the flask was left for 3 min at 37°C . The suspended cells were freeze-thawed 5 times to collect the virus . Enzyme-linked immunosorbent assay ( ELISA ) was used to measure HAV antigen content . The percent inhibition was determined relative to the mean OD450 values of the control wells in which the virus has been incubated with medium alone . Purified F4 , F6 , F7 , and F9 Fab fragments were incubated with purified HAV ( at a concentration of 2 mg/ml ) separately on ice for 10 min at a ratio of 120 Fab molecules per virion . A 3-μl aliquot of the mixtures of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fb-HAV were transferred onto a freshly glow-discharged 400-mesh holey carbon-coated copper grid ( C-flat , CF-2/1-2C; Protochips ) . Grids were blotted for 3 . 5 s in 100% relative humidity for plunge-freezing ( Vitrobot; FEI ) in liquid ethane . Cryo-EM data sets were collected at 300 kV using a Titan Krios microscope equipped ( Thermo Fisher ) with a K2 detector ( Gatan , Pleasanton , CA ) . Movies ( 25 frames , each 0 . 2 s , total dose 30 e− Å−2 ) were recorded with a defocus of between 1 and 2 . 5 μm using SerialEM [41] , which yields a final pixel size of 1 . 35 Å . The frames from each movie were aligned and averaged for the correction of beam-induced drift using MOTIONCORR [42] . Particles from micrographs were picked automatically using ETHAN [43] and then manually screened using the boxer program in EMAN [44] . The CTF parameters for each micrograph were estimated by using a GPU accelerated program Gctf [45] . Cryo-EM structures were determined with Relion 1 . 4 [46] with the application of icosahedral symmetry . The initial model was created by EMAN2 [47] . A total of 4 , 536 , 7 , 245 , 16 , 743 , and 3 , 798 particles of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV were used to determine structures at resolutions of 3 . 9 , 3 . 68 , 3 . 05 , and 3 . 79 Å , respectively , as evaluated by the so-called gold standard FSC procedure between 2 half maps ( threshold = 0 . 143 ) [21] . The crystal structure of HAV full particle ( PDB ID code 4QPI ) was used to fit the complex EM maps , and the atomic models of F4- , F6- , F7- , and F9-Fabs were built de novo into densities with the structure of R10 ( PDB ID code 5WTG ) as a guide , using COOT [48] . All models were further refined by positional and B-factor refinement in real space using Phenix [49] and rebuilding in COOT [48] iteratively . The final models were evaluated by Molprobity [50] functions integrated in Phenix . Data and refinement statistics are summarized in Table 1 . In antigen binding systems , the chain C , D , E , and G of F4 Fab-HAV , F6 Fab-HAV , F7 Fab-HAV , and F9 Fab-HAV structures were fetched out to perform MD simulation and binding energy calculations . In the 4 chains , the chain D and E were set to ligand and the chain C and G were set to receptor . The complex was solvated to TIP3P waters , and 0 . 1 M NaCl was added to systems as salt with soft tleap in AmberTools 16 [51] . Amber 16 was used to perform MD simulation . All 4 systems were first relaxed by 5 , 000-step minimization ( 2 , 000 steps , steepest descent minimizations; 3 , 000 steps , conjugate gradient minimization ) . After minimization , the system was gradually heated from 0 K to 300 K in the canonical NVT ensemble with a Langevin thermostat using a collision frequency of 2 . 0 ps−1 . Initial velocities were assigned from a Maxwellian distribution at the starting temperature . Then 100 ps of density equilibration with weak restraints on the complex was followed by 500 ps of constant pressure equilibration at 300 K and 1 atm . Finally , 10 ns MD simulations for each system was conducted with the target temperature at 300 K and the target pressure at 1 . 0 atm . In 4 systems , Na+ was selected as the counter ions; the concentration of NaCl was set to 0 . 1 M , and ions’ parameters of Joung and Cheatham [52] were used . Electrostatics was handled using the particle mesh Ewald ( PME ) algorithm [53] with a 10 . 0 Å direct−space nonbonded cutoff . All bonds involving hydrogen atoms were constrained using the SHAKE algorithm [54] , using a time step of 2 . 0 fs . The coordinates’ trajectories were saved every 2 ps during the whole MD runs . MM-GBSA [55] was used to calculate the binding energy of F4 , F6 , F7 , F9 , and R10 Fab ( chain D and chain E ) and HAV VP2 and VP3 ( chain C and chain G ) . In each system , 100 snapshots of the last 6 ns MD simulation were fetched out to calculate the binding energy . The entropy contributions were neglected because the same receptor was used and because the normal mode analysis calculations are computationally expensive and subject to a large margin of error that introduces significant uncertainty in the result . The free energy for each species ( ligand , receptor , and complex ) is decomposed into a gas-phase MM energy , polar , and nonpolar solvation terms , as well as an entropy term , as shown in the following equation: ΔG=ΔEMM+ΔGsolv−T∙ΔS=ΔEbat+ΔEvdw+ΔEcoul+ΔGsolv . p+ΔGsolv . np−T∙ΔS . EMM is composed of Ebat ( the sum of bond , angle , and torsion terms in the force field ) , a van der Waals term , EvdW , and a Coulombic term , Ecoul . Gsolvp is the polar contribution to the solvation free energy , often computed via the Generalized-Born ( GB ) approximation . Gsolvnp is the nonpolar solvation free energy , usually computed as a linear function of the solvent-accessible surface area ( SASA ) . The Phase program of the Schrodinger Suite 2013 [30] was used for the pharmacophore modeling . Four key residues at positions of 31 , 32 , 101 , and 102 from the heavy chain of the 5 NAbs were selected as a reference structure for modeling . Pharmacophore sites were generated using the default set of chemical features: hydrogen bond acceptor ( A ) , hydrogen bond donor ( D ) , hydrophobe ( H ) , negative ionizable ( N ) , positive ionizable ( P ) , and aromatic ring ( R ) . The size of the pharmacophore box was set to 1 Å to optimize the number of final common pharmacophore hypotheses . The generated pharmacophore was used to screen the drugs database of Drugbank . The distance matching tolerance was set to 2 . 0 Å . A total of 2 , 588 drugs were screened out using this procedure . The docking algorithm Glide [31] , which is based on descriptor matching , was used to perform virtual screening and learn the interactions between small molecules and the protein structure . The structure of HAV epitopes was prepared and then used to build the energy grid . For Glide docking , the docking box was centered on the position of mass center of the 4 selected residues , and its outer box size was set to 40 × 40 × 40 Å . The scaling factor for protein van der Waals radii was set to 1 . 0 . All 2 , 588 drugs were docked to the antigen binding site on HAV capsids , and the 4 molecules were selected . The compound 3 , which has the best glide score , was selected finally . Approximately 2 × 105 2BS cells were seeded into each well of a 24-well plate and incubated overnight in a CO2 incubator supplemented with 5% CO2 . Before virus infection , HAV ( 100 TCID50 ) was incubated with serially diluted concentrations ( 0 , 0 . 008 , 0 . 032 , 0 . 125 , 0 . 5 , 2 , and 8 μM ) of golvatinib ( MedChemExpress ) for 1 h at room temperature with gentle rocking and then transferred to the plate containing 2BS cells . After adsorption for 1 h , the inoculum was removed , and the cells were supplied with fresh maintenance medium and incubated at 34°C . At 7 d post infection , the cells were lysed for ELISA to determine HAV antigen content . Approximately 2 × 105 2BS cells were seeded into 24-well plates and incubated overnight in a CO2 incubator . Inhibitors were serially diluted concentrations ( 0 , 0 . 008 , 0 . 032 , 0 . 125 , 0 . 5 , 2 , and 8 μM ) and then transferred to the plate containing 2BS cells . Seven days after addition of drug , CCK-8 kit ( Sangon Biotech ) was used according to the manufacturer’s protocol . In brief , each well of the plate had10 μl CCK-8 solution added and was incubated 2 h at 37°C . Absorbance at 450 nm was measured by SynergyH1 microplate reader ( BioTek ) . An MX3005p RT-PCR instrument ( Agilent ) was used for the thermofluor assays . SYTO9 ( Invitrogen ) was used as a fluorescent probe to detect the presence of single-stranded RNA [56 , 57] . A 50 μL reaction solution was set up in the PCR plate ( Agilent ) , containing 1 . 0 μg of virus plus serially diluted concentrations of golvatinib ( 0 , 0 . 512 , 5 . 12 , and 51 . 2 μM ) and 5 μM SYTO9 in PBS buffer solutions , and ramped from 25°C to 99°C with fluorescence recorded in triplicate at 1°C intervals . The RNA release ( Tr ) temperature was taken as the minimum of the negative first derivative of the RNA exposure . HAV ( 100 TCID50 ) was mixed with serially diluted concentrations of golvatinib or NAbs before the virus attached to cells ( 2 × 105 ) and then added to 2BS cells and incubated at 4°C for 1 h . The cells were washed three times and total cellular RNA purified using RNeasy mini kit ( Qiagen ) , as described in the manufacturer’s instructions . Real-time quantitative PCR ( qPCR ) was performed using One Step SYBR PrimeScript RT-PCR Kit ( TaKaRa ) in a MX3005p RT-PCR instrument ( Agilent ) . The 20-μL reaction contained 12 . 5 μL 2 × One Step SYBR RT-PCR Buffer III , 0 . 5 μL TaKaRa Ex Taq HS , 0 . 5 μL PrimeScript RT Enzyme Mix II , 0 . 5 μL each of 10 μM forward ( 5′-TGG AAT CAC ATT AAA GCA AGC AA-3′ ) and reverse ( 5′-GGA ACA CGA AAT CTC AAA GTT GAC T-3′ ) primers , 2 μL total RNA , and 4 μL RNase-free H2O . The thermal profile for qPCR was 42°C for 5 min for reverse transcription , 95°C for 10 s for reverse transcription inactivation; this was followed by 40 cycles of denaturation at 95°C for 10 s and annealing and extension at 60°C for 30 s . GAPDH was used as the housekeeping gene to normalize samples ( forward 5′-CTG TTG CTG TAG CCA AAT TCGT-3′ , reverse 5′-ACC CAC TCC TCC ACC TTT GAC-3′ ) . The analysis of relative levels of HAV RNA in different samples was performed by comparative 2−ΔΔCT method [58] . The cryo-EM maps of the F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV complexes were deposited in the Electron Microscopy Data Bank with accession number EMD-9827 , EMD-9828 , EMD-9829 , and EMD-9830 , respectively . The atomic coordinates for F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV complexes were deposited in the PDB with accession numbers: 6JHQ , 6JHR , 6JHS , and 6JHT , respectively .
|
Hepatitis A virus ( HAV ) is a unique , hepatotropic human picornavirus that infects approximately 1 . 5 million people annually and continues to cause mortality despite a successful vaccine . There are no licensed therapeutic drugs to date . Better knowledge of HAV antigenic features and neutralizing mechanisms will facilitate the development of HAV-targeting antiviral drugs . In this study , we report 4 potent HAV-specific neutralizing monoclonal antibodies ( NAbs ) , together with our previous reported R10 , that efficiently inhibit HAV infection by blocking attachment to the host cell . All 5 epitopes are located within the same patch and are highly conserved across 6 genotypes of human HAV , which suggests a single antigenic site for HAV , highlighting a prime target for structure-based drug design . Analysis of complexes with the 5 NAbs with varying neutralizing activities pinpointed key structure-activity correlates . By using a robust in silico docking method , one promising inhibitor named golvatinib was successfully identified from the DrugBank Database . In vitro assays confirmed its ability to block viral infection and revealed its neutralizing mechanism . Our approach could be useful in the design of effective drugs for picornavirus infections .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"antimicrobials",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"enzyme-linked",
"immunoassays",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"drugs",
"immunology",
"microbiology",
"electron",
"cryo-microscopy",
"viruses",
"microscopy",
"pharmacology",
"antibodies",
"immunologic",
"techniques",
"research",
"and",
"analysis",
"methods",
"monoclonal",
"antibodies",
"immune",
"system",
"proteins",
"proteins",
"medical",
"microbiology",
"antigens",
"microbial",
"pathogens",
"immunoassays",
"hepatitis",
"viruses",
"viral",
"packaging",
"viral",
"replication",
"hepatitis",
"a",
"virus",
"biochemistry",
"virology",
"viral",
"pathogens",
"physiology",
"electron",
"microscopy",
"microbial",
"control",
"biology",
"and",
"life",
"sciences",
"antivirals",
"organisms"
] |
2019
|
Structural basis for neutralization of hepatitis A virus informs a rational design of highly potent inhibitors
|
The complete connectional map ( connectome ) of a neural circuit is essential for understanding its structure and function . Such maps have only been obtained in Caenorhabditis elegans . As an attempt at solving mammalian circuits , we reconstructed the connectomes of six interscutularis muscles from adult transgenic mice expressing fluorescent proteins in all motor axons . The reconstruction revealed several organizational principles of the neuromuscular circuit . First , the connectomes demonstrate the anatomical basis of the graded tensions in the size principle . Second , they reveal a robust quantitative relationship between axonal caliber , length , and synapse number . Third , they permit a direct comparison of the same neuron on the left and right sides of the same vertebrate animal , and reveal significant structural variations among such neurons , which contrast with the stereotypy of identified neurons in invertebrates . Finally , the wiring length of axons is often longer than necessary , contrary to the widely held view that neural wiring length should be minimized . These results show that mammalian muscle function is implemented with a variety of wiring diagrams that share certain global features but differ substantially in anatomical form . This variability may arise from the dominant role of synaptic competition in establishing the final circuit .
The nervous system's connectivity is believed to be a fundamental determinant of its function [1 , 2] , but in general it is not readily accessible . One way to characterize neural circuits is to extract statistical properties of connectivity , often by pooling data from multiple animals [3–6] . This method assumes that connectional specificity at the level of classes of cells suffices to account for the properties of circuits [7–9] . It also assumes that within a class , each neuron's connectivity is established independently , without correlations with that of other cells . While such models may provide interesting ideas about how the nervous system works , their underlying assumptions are probably oversimplified . Neurons , for example , often innervate a nonrandom subset of cells within their target , rather than stochastically innervating such a group of cells [10 , 11] . In many circuits , neurons innervating the same group of cells do not establish connections independently , as evidenced by interneuronal competition observed during development [12 , 13] . Thus some neuroscientists have concluded that “any attempt to interpret neuronal connectivity purely in terms of probabilities … must be doomed to failure [14] . ” The obvious alternative is to obtain complete wiring diagrams ( connectomes ) of either the entire nervous system of an individual animal , or a well-defined subnetwork of the nervous system , based on direct observation rather than statistical inference . It is possible that such maps might ultimately reveal that neural circuits are stochastic in certain aspects and thus amenable to probabilistic descriptions . On the other hand , such maps may reveal organizational specificity that may not be detectable by statistical analysis , especially when the structure and connectivity of individual neurons need to be characterized in the context of the entire circuit ( see below ) . The first attempt to directly describe a connectome was undertaken in the parasitic nematode Ascaris lumbricoides with optical microscopy [15–17] , but it produced only “enigmatic wiring diagrams” [18] because of inadequate resolution . The only successful connectomic reconstruction was accomplished in another nematode , C . elegans , using serial electron microscopy [2 , 18–20] . This map has proven to be a valuable resource for further analysis of circuits underlying C . elegans behaviors [21–23] . Therefore , it is likely that mammalian connectomes will also provide important information . The advent of transgenic technologies to label neurons [24] , combined with automated optical microscopy and computer-assisted image analysis tools [25] , provides an avenue for the reconstruction of mammalian connectomes . Nevertheless , given the enormous complexity of mammalian nervous systems , it is necessary to begin this endeavor with tractable circuits . In this work we attempt to generate the complete wiring diagram of a peripheral neuromuscular circuit . This circuit consists of the full set of α-motor axons and the full complement of muscle fibers in the single muscle innervated by these axons . It can be captured in its entirety because each muscle's innervation is nonoverlapping . In contrast , any finite volume of circuitry in the central nervous system ( CNS ) contains neuronal processes entering and leaving the volume , so completeness of reconstruction cannot be achieved locally . Another advantage of the neuromuscular circuit is that its functional organization has been studied intensively , which culminated in the discovery of the size principle [26] , namely , the recruitment of motor neurons proceeds in the order of increasing twitch tensions . The anatomical underpinnings of the graded tensions elicited by the group of motor neurons , however , have not been demonstrated . An additional rationale for studying the neuromuscular connectome is that the mature wiring diagram emerges from an extensive postnatal reorganization of axonal arbors known as synapse elimination . Previous imaging studies [27 , 28] suggested that the fate of different axons that co-innervate the same NMJ is influenced by the interactions of these axons at other NMJs with other axons . Therefore , predicting which branches are retained and which are pruned requires analyzing the competitive relationships among the entire group of neurons . In this work we took the first step of unraveling the rules of this competition by generating the adult neuromuscular connectome , which is the end product of this aforementioned process . Lastly , comparing corresponding connectomes between different animals or in the same animal ( e . g . , left versus right side ) may help clarify the extent to which genetics , epigenetic factors , and random fluctuations impact circuit structure .
The reconstructed interscutularis connectomes ( Figure 3 and Figure S1 ) provided an atlas of neuromuscular connectional diagrams of all the axons within the muscle . This atlas included information about the number and position of all the postsynaptic targets , as well as branching topology , neighbor relations , and segmental geometry of each axon . Some of the information , such as motor unit sizes ( number of neuromuscular junctions [NMJs] innervated by one axon ) and statistical properties of axonal tree structures , may be obtainable by pooling single axon data from many sparsely labeled muscle samples , if homogeneity among animals and unbiased sampling are assumed . In this case , the connectomic approach provides a compendium of such information efficiently . More importantly , however , other aspects of neuromuscular circuit organization , such as neighbor relations in the fasciculation and innervation pattern , can only be understood by placing each individual axon into the context of the whole circuit's structure , and thus require the connectomic approach . Moreover , comparison between identified neurons across mice cannot be achieved by random , sparse labeling ( see below ) . To summarize the data: each connectome contained 14 . 5 ± 1 . 5 axons and 198 ± 11 muscle fibers . The axons exhibited a wide range of motor unit sizes , with a predominance of smaller motor units over larger ones ( Figure 4A ) . The smallest motor units had only one NMJ ( 2/87 axons ) ; the largest motor units had 37 NMJs ( 2/87 axons ) . We found that among the 979 instances of axonal branching , most ( 88 . 5% ) were binary , with progressively smaller fractions of higher degree branching ( tri-furcations 10 . 7% , 4-furcations 0 . 6% , 5-furcations 0 . 2% , see Figure S2 ) . We then analyzed branching symmetry of axons that innervated more than three NMJs ( n = 83 ) . This symmetry was evaluated with the imbalance index I [31] , which is 0 for a completely symmetric tree ( e . g . , each branching point gives rise to exactly two daughter branches ) , and one for a completely asymmetric tree ( e . g . , each branching point gives rise to one terminal branch and one branch that further bifurcates ) . Most axons were relatively symmetric ( I = 0 . 31 ± 0 . 21 ) , and axons with larger motor unit sizes tended to be more symmetric ( Spearman test , p < 0 . 0001 ) . The total intramuscular length of axonal arbors ranged from 1 , 583 μm to 13 , 320 μm ( 7 , 256 ± 2 , 352 μm , mean ± standard deviation [SD] ) , with a positive correlation with motor unit sizes . Axonal segments between branching nodes became progressively shorter with increasing branch orders ( Figures 4B and S3 ) . We noted that the range of motor unit sizes ( 7 . 7 ± 2 . 8-fold ) was similar to that of twitch tensions recorded in previous physiological studies of mammalian muscle contraction ( e . g . , 8 . 3-fold [32] , 12 . 3-fold [33] ) . Furthermore , both the twitch tension distribution and the motor unit size distribution shared the same shape: unimodal and skewed towards the smaller end ( Figure 4A ) . These results strongly argue that motor unit sizes are the anatomical underpinning of the observed distribution of twitch tensions . As the entire collection of motor units in each muscle was known in our dataset , we could ask whether all connectomes follow the same motor unit size distribution . Indeed we found that motor units in all six connectomes were distributed in the same way ( p > 0 . 2 , Kruskal-Wallis test ) . Given that motor neurons are recruited in a fixed order ( weak to strong , see [26] ) , the correspondence between motor unit size and twitch tension mentioned above allowed us to establish the functional correspondence between individual axons in different muscle samples . Based on conduction velocity studies , axons generating larger twitch tensions appear to possess larger calibers [33 , 34] . We thus anticipated that axonal cross-sectional area should correlate with motor unit size . We measured the mean cross-sectional area of each axon right before its first intramuscular branch and normalized the area to the total cross-sectional areas of all axons innervating the same muscle . We found that the normalized cross-sectional area A was correlated with the motor unit size M obeying a power law: A scaled approximately as the square root of M ( Figure 4C ) . Furthermore , the cross-sectional area of first order axonal branches was correlated with the number ( N ) of downstream NMJs by a similar scaling relationship: A ∼ N0 . 536 , ( n = 47 , 95% confidence interval [CI] of the exponent: 0 . 4663–0 . 6051 ) . This similarity suggests that the scaling relationship is a fundamental property of motor axon branching . In order to better understand the origin of this relationship , we measured the total axonal arbor length L distal to the point where the axon enters the muscle . We found that L scaled linearly with A ( Figure 4D ) . Furthermore , L also scaled as the square root of M ( Figure 4E ) . Taken together , these results support the idea that the principal determinant of axonal cross-sectional area is the energy cost associated with axonal membrane ( see Discussion for details ) . As the arrangement of different motor units in a muscle affects the mechanical properties of force delivery , we proceeded to address how motor units are deployed relative to each other in the interscutularis . The positions of NMJs in most motor units were distributed uniformly in the endplate band ( Figure S4 ) , both across the width of the muscle ( 80/83 , 96 . 4% ) and along the muscle's length ( 79/83 , 95 . 2% ) . Statistical test also suggested that some motor units ( 18/83 across the muscle , 4/83 along the muscle ) were “super-uniform , ” i . e . , the distribution was too regular to be from a random uniform sample . Therefore , the interscutularis muscle does not seem to possess compartments as described in certain larger muscles [35 , 36] . In distinction to entire motor units , primary subtrees of individual axons were not uniformly distributed . In most cases they appeared to invade nonoverlapping territories ( for example , see Figure S5 ) . We compared the distribution of subtree terminals of 27 axons in which each subtree had at least four terminals . We found that in 20 axons ( 74% ) the distribution of terminals of the two subtrees was different ( p < 0 . 05 , generalized Wald-Wolfowitz test [37] ) . In particular , in 12 axons ( 44 . 4% ) the territories of the two subtrees were completely segregated . On the other hand , when primary subtrees belonging to different axons were compared , their territories tended to be overlapping ( 78/112 pairs , 69 . 4% , p < 0 . 05 , generalized Wald-Wolfowitz test ) . This arrangement of subtrees suggests that developmental mechanisms prevent multiple branches of the same axon from projecting to the same region , while permitting branches of different axons to intermingle in the same region . Such mechanisms may explain the observation that multiple axons innervate the same muscle fiber at early developmental stages [13] , whereas rarely do two branches of the same axon innervate a single muscle fiber . The intramuscular nerve fasciculation patterns reflect the collective behavior of all the axons . We found that the relationship between branching structures of individual axons and nerve fascicles was surprisingly complicated . Individual axons' branching behavior was not strictly coupled to the fasciculation pattern of the nerve . At some nerve branching points no axons branched; different axons simply followed one of the paths ( Figure 5A ) . On the other hand , some axons branched inside a nerve segment and the resultant branches traveled in parallel along the same segment over some distance ( Figure 5B ) . The most conspicuous example of such behavior was the extramuscular branching of axons discussed previously ( Figure 2 ) . Moreover , although most fasciculated nerve segments travel in a proximal-distal direction , some axonal branches contained in them did not follow the same direction . For example , in Figure 5C three axons entered the nerve fascicle from the left and two axons traveled in the opposite direction . Taken together , among 85 branching axons , 69 ( 81 . 2% ) branched at least once within a nerve segment , and 29 ( 34 . 1% ) contained at least one branch that traveled against the direction of some nerve segments . Overall , 89 . 4% of the axons deviated in some way from being a proper subgraph of the nerve fasciculation pattern . It is possible that intramuscular nerve fasciculation reflected predetermined patterning similar to the highly stereotyped nerve structures seen more proximally ( e . g . , the brachial plexus ) . We thus tested whether there might be a conserved core fasciculation pattern in the interscutularis muscle . We assigned to each segment of the nerve a weight proportional to the total number of downstream NMJs ( Figure 5D ) . We found that the extracted “skeletons” were topologically distinct in each muscle including left-right pairs in the same animal ( Figure 5D insets ) . Therefore it seems unlikely that nerve fasciculation patterns in a muscle are genetically predetermined . As mentioned above , the knowledge of motor unit sizes of all axons allowed us to identify exact neuronal counterparts in different muscle samples . This knowledge enables exploration of a question that has been investigated in invertebrates but , to our knowledge , never in terrestrial vertebrates: the degree to which an individually identified neuron shares the same branching structure with its counterparts in other samples . Although nerve fasciculation patterns differed from sample to sample as shown above , the possibility that the branching structures of axonal counterparts be identical could not be ruled out , as axonal branching structures are not necessarily subgraphs of the nerve fasciculation pattern ( see section above ) . In order to compare neuronal counterparts , we first determined whether there were systematic differences between left and right copies of the interscutularis muscle . The number of muscle fibers on left and right sides was not significantly different ( left 201 . 7 ± 20 . 7 versus right 199 . 0 ± 18 . 0 , n = 6 pairs , two-tailed p = 0 . 53 , paired Student's t-test; Figure S6A ) , nor was there a difference in the distribution of muscle fiber types ( type I: left 40 . 3% , right 42 . 4%; type IIA: left 20 . 1% , right 19 . 1%; type IIB+IIX: left 39 . 6% , right 38 . 5%; two pairs ) . The number of innervating motor neurons was not significantly different either ( left 14 . 7 ± 1 . 5 versus right 14 . 0 ± 1 . 4 , six pairs , two-tailed p = 0 . 47 , paired Student's t-test; Figure S6B ) . We thus proceeded to identify each neuron and its counterparts based on motor unit size and/or its rank within the connectome . We analyzed four connectomes in two animals ( left-right pair for each animal ) . We first compared the largest motor unit with its contralateral counterpart in the same animal . In one case their sizes were similar ( animal M3 , left 25 NMJs [12 . 8% of all NMJs in the muscle] versus right 29 [14 . 8%] , Figure S1B ) , but in the other case less so ( animal M4 , left 37 [18 . 8%] versus right 28 [15 . 2%] , Figure 3B ) . Moreover , their appearances did not exhibit appreciable similarity upon visual inspection ( e . g . , Figure 3B , L1/R1 pair ) . We then compared smaller motor units with their contralateral counterparts and again found no appreciable similarity in the branching structures . Whether the counterpart was defined by rank order ( Figures 3B and S1B ) or by absolute motor unit size made no difference; in each case there was no evidence for a common branching pattern . In order to investigate whether left-right pairs of axons with the same rank or same motor unit size are similar in less obvious ways , we focused on their topologies , ignoring geometric features ( e . g . , length and angle of branches ) . We found a wide range of different topologies between axons with the same rank ( Figure 6A ) and even among axons with the same motor unit size ( Figure 6B ) . We used tree-editing distance ( TED [38] ) to quantify the topological difference between axons . We found that left-right pairs of axons in the same animal ( intra-animal pairs ) were no more similar to each other than interanimal pairs of same-sized axons ( Figure 6C ) . Furthermore , intra-animal pairs of axons were no more similar than pairs of synthetic “axons” randomly selected from an ensemble of tree structures generated by a Monte Carlo simulation ( Figure 6D ) ( intra-animal pair TED , 9 . 00 ± 2 . 62; Monte Carlo TED , 8 . 87 ± 2 . 50; two-tailed p = 0 . 89; unpaired Student's t-test ) . These results indicate that the topologies of intra-animal left-right pairs of axons were not correlated . In addition to the variability of branching topology we found that axonal trajectories did not adhere to the principle of minimization of total wiring length , initially proposed by Cajal [1] and supported by the full reconstruction of the C . elegans nervous system [39–41] but see [42] . Even superficial visual inspection of the interscutularis connectome showed that almost every axon's total length could be shortened by following different nerve fascicles or altering the location of branching points . Remarkably , some axons took highly tortuous routes to their target muscle fibers even when more direct paths seemed possible ( Figure 7 ) . Moreover , ∼6% of axons branched extramuscularly as previously mentioned ( Figure 2 ) , which is also suboptimal , as the two resultant branches of the same axon invariably continued together into the same muscle . In these cases , removal of the extramuscular branching could have saved 2 , 072 ± 1 , 116 μm ( n = 4 axons ) of wiring length , which is equivalent to 25 ± 9 . 5% of the intramuscular wiring length of these axons . As the intramuscular wiring length is comparable to the distance from the cell body to the muscle , its contribution to the total metabolic cost of the cell is substantial . Therefore , the 25% extra wiring length imposes significant additional metabolic load to the cell . However , from the perspective of the neuromuscular system as a whole , this additional cost may be insignificant , since the metabolic load of muscle contraction far exceeds that of axonal conduction .
There are several reasons for mapping the connectome , i . e . , the entire wiring diagram of a neural circuit . Most importantly , this map represents the complete inventory of connectional information in one particular sample . The conventional approach to circuit analysis , in contrast , infers connectivity by pooling partial data from many samples , and thus relies on assumptions such as homogeneity of neurons in a population , or stereotypy of circuits among different individuals . The connectomic approach is advantageous because it does not require such assumptions . Moreover , as every cell is identified , the way in which each cell integrates into the organization of the circuit is revealed . Lastly , comparison between different instantiations of the same circuit can reveal those aspects of connectivity that are physiologically relevant , and those that are not . In this work we present an initial attempt to reconstruct mammalian subnetwork connectomes . We chose the mouse interscutularis muscle as the starting point because of its simplicity and accessibility . Its simplicity lies in the fact that being an end organ , it does not have strong recurrent components . This feature allows us to achieve “completeness” within a finite volume , as opposed to the situation in most CNS circuits , where recurrent connections originating from distant sites are commonplace . Furthermore , it is simple because the input is purely divergent: each axon innervates multiple muscle fibers but each muscle fiber has only one input . This pattern is only present at a few places of the CNS . Although the interscutularis muscle represents one of the smallest possible connectomes in mammals , it still presented significant technical challenges for reconstruction . Often axonal branches were tightly fasciculated with each other , the distance between which approached the resolution limit of confocal microscopy . This problem was aggravated by scattering especially when imaging deeper structures . Therefore axonal profiles sometimes bled into each other , so computer segmentation had to be monitored and complemented by manual intervention , which significantly reduced the speed of reconstruction . Optimally , the Reconstruct program we modified for automatic segmentation traced out 4 mm of axonal length per hour . In practice , however , the requirement of human monitoring and editing reduced it to ∼0 . 5 mm per hour . We anticipate that the automated imaging and semi-automated reconstruction undertaken in this work will also be generalized to the study of CNS connectomes . However , the aforementioned technical difficulties in imaging and image analysis would be greater in the CNS , where the length scale of neural structures is much smaller and the packing of neuropil is much denser . Thus future technical innovations are required to facilitate fully automated reconstruction . For example , different colors may be introduced to spectrally separate different neurons [29]; imaging resolution may be improved through super-resolution techniques [43–46]; aberrations induced by scattering in deep tissues may be overcome by serial sectioning followed by either electron microscopy [47 , 48] or optical microscopy [49] , by adaptive optics [50] , or by tissue clearing [51] . The reconstructed connectomes demonstrated four organizational principles of neuromuscular circuits . First , the motor unit size distribution in each connectome paralleled previous results from physiological recordings of twitch tensions , providing an anatomical correlate for Henneman's size principle , which until now was a physiological concept . The skewed distribution of twitch tensions , obtained by pooling data from many different samples , demonstrated that statistically most motor units generate small twitch tensions , and a few generate large twitch tensions . However , the degree to which the set of motor units within each muscle obeys the same distribution has not been directly demonstrated . Our measurement of all motor units in each muscle shows that the skewed motor unit size distribution holds for each sample . Second , we found robust , quantitative relationships between axonal caliber , arbor length , and motor unit size . The cross-sectional area of an axon proximal to its intramuscular arborization scaled linearly with its intramuscular length ( Figure 4D ) . Because axonal caliber is proportional to axoplasmic transport [52] , it may scale with downstream metabolic expenditure . The energy expenditure is primarily devoted to resting and action potentials instead of synaptic transmission , therefore proportional to the surface area of the axonal membrane [53] . As long as the axonal caliber remains relatively constant , the surface area is proportional to arbor length . This may explain the linear relationship between axonal caliber and arbor length . In addition , we found that the total intramuscular length of an axon scaled with the square root of its motor unit size ( Figure 4E ) , akin to the prediction based on optimization considerations [54] . This power law scaling may be the result of the fact that average branch lengths progressively decreased as branch orders increased ( Figure 4B ) . Therefore as motor unit size increases , the required increment in axonal arbor length is reduced . This relationship , combined with the proportionality between axonal caliber and arbor length , explains why axonal caliber scales sublinearly with motor unit size ( Figure 4C ) . Third , the axonal branching structure of each motor neuron was unique . We compared each axon with its functional counterparts , as defined by the size principle , in other muscles , and found substantial topological differences . Left-right pairs of corresponding neurons in the same animal showed no less variation than ipsi- or contralateral pairs from different animals . Such intra-animal variance is surprising , as each pair of neurons had identical genetic background and presumably experienced an identical environment . This result suggests that the branching pattern of these neurons was not predetermined , which contrasts strongly with the situation in invertebrates . For instance , the C . elegans connectome revealed remarkable stereotypy in the structure of the neural circuit . Worm neurons that are ontogenetic counterparts share almost identical branching patterns and connectivity both within an individual and across different animals , even though they may not be exact replicas of each other [18 , 19] . In annelids [55 , 56] , insects [57–62] , and crustaceans [63 , 64] individual neurons can also be identified , and their axonal branching patterns are stereotyped . In particular , this mammalian result contrasts with the stereotypy of neuromuscular innervation in invertebrates . For example , although there are fine structural differences in the terminal branching of axons at NMJs of any particular muscle fiber in insects , even these branches seem to have morphological regularities that are recognizable between different animals [65 , 66] . In mammals not only is the preterminal branching highly variable ( as shown in this paper ) , but our experience suggests that no two NMJs look the same . Thus axonal branching in this mammalian system seems fundamentally different from that found in invertebrates . Fourth , many axons exhibited tortuous trajectories en route to target muscle fibers , contrary to the notion that neural circuits should minimize total wiring length [67] . The layout of axonal arbors did help to minimize wiring length by preventing significant overlaps between territories of subtrees . However , other aspects of wiring , in particular extramuscular branches , wasted substantial wiring length . In contrast , in C . elegans neural wiring approximates the optimal solution fairly well [40] . The suboptimality in wiring length found in this work does not imply that the optimization principle per se is inapplicable; it rather suggests that factors other than wiring length also play a significant role . For instance , invertebrate nervous systems are under tight genetic control , and particular mutations in a single gene can lead to stereotyped alterations in neural wiring [62] . The mammalian neuromuscular system , on the other hand , may rely more strongly on activity-dependent reorganizations for each individual neural circuit to settle down on a particular wiring scheme . This strategy does not guarantee the establishment of optimal wiring , but only arrives at a solution that is functionally acceptable . In conclusion , the interscutularis connectome reveals that in mammals , muscle function is implemented with a variety of wiring diagrams that share certain global features but differ substantially in anatomical form . Even the left and right copies of this neuromuscular circuit in the same animal exhibited significant variation . Nevertheless , the multitude of wiring diagrams exhibited no appreciable functional difference . Does this fact imply that , a posteriori , the observed variability in this system is inevitable , as there is no functional reason to impose a particular wiring diagram ? We believe that this may not be the case . In general , the nervous system contains features that may have no adaptive value but tend to remain conserved [68 , 69] . This conservation of structure may be due to tight developmental constraints , as random changes during development may lead to dysfunction . Therefore , the rationale for the observed wiring variability may lie beyond the lack of functional significance . This variability may result from the peculiarities of the nervous system of terrestrial vertebrates [13] . The neuromuscular circuit , for example , has a reduplicated arrangement of elements: each neuron belongs to a group of similar cells ( the motor neuron pool ) and projects to a population of similar postsynaptic targets ( muscle fibers ) . At early developmental stages there is extensive fan-out ( each neuron innervates a large number of muscle fibers ) and fan-in ( each muscle fiber is innervated by many neurons ) . The final circuit , however , retains only a small fraction of the initial connections—those that survived the pruning phase of synapse elimination [12 , 70] . This reorganization process is unidirectional ( connections are lost but never regained ) , and the fate of an axonal branch is related to the identity of its competitors [27 , 28] . If a different input were eliminated from even one muscle fiber early on , there might be substantial divergence in the structure of the connectome when synapse elimination is complete . This sensitivity to developmental history may be the engine that generates diversity in neural wiring . From this perspective , the variability is not a sign of lack of regulation , but rather indicates a different developmental strategy . Instead of genetically specifying the optimal wiring diagram for all individuals , this strategy allows a different instantiation to emerge in each case . Given the important role of interneuronal competition in the developing CNS [71–74] , this strategy could well be a common theme in the entire mammalian nervous system . The value of this vertebrate innovation may be that it unfetters the structure of the nervous system from strict genetic determinism .
All animal experiments were performed according to protocols approved by Harvard University Institutional Animal Care and Use Committee ( IACUC ) . Young adult ( ∼30 d old ) transgenic mice of thy-1-YFP-16 line received IP injections of 0 . 1 ml/20g ketamine-xylazine ( Ketaset , Fort Dodge Animal Health ) or 0 . 2 ml/30 g sodium pentobarbital ( 64 . 8 mg/ml in sterile water ) . Once anesthetized , the animals were transcardially perfused with 4% paraformaldehyde ( PFA ) in 0 . 1 M phosphate-buffered saline ( PBS [pH 7 . 4] ) . The interscutularis muscle along with a segment of its innervating nerve was removed and postfixed in 4% PFA for 30 min . Muscles were rinsed in PBS ( 25 °C , 30 min × 2 ) and mounted on slides with Vectashield mounting medium ( Vector Laboratories ) . Mounted slides were slightly squeezed between a pair of small magnets for 12 h in order to flatten the tissue so that the distance from tissue surface to the coverslip was minimized and roughly constant . For identification of muscle fiber type , muscles were removed as above and postfixed with 1% PFA for 7 min , frozen , and sectioned at 20 μm using a Leica Cryostat . Then sections were incubated with blocking solution ( 2% BSA + 1% goat serum + 0 . 3% triton ) at 25 °C for 3 h , and incubated with monoclonal antibodies against myosin type I and 2A ( mouse anti-myosin I IgG1 , 1:20 , Novocastra; mouse anti-myosin 2A IgG1 , 1:10 , Iowa Hybridoma Bank ) at 4 °C for 6–8 h . After several washes in 0 . 1% PBS-triton , samples were incubated with secondary antibody ( Alexa-488 anti-mouse IgG1 , 1:1 , 000; Molecular Probes ) for 3 h . Finally muscle sections were rinsed in PBS ( 25 °C , 30 min × 2 ) and mounted on slides with Vectashield mounting medium . Samples were imaged using a confocal laser scanning microscope ( Zeiss Pascal , Carl Zeiss ) equipped with a motorized stage . We used a 63× 1 . 4 NA oil-immersion objective and digitally zoomed-in so that each pixel was 0 . 1 μm ( Nyquist limit ) . YFP florescence was excited with a 488-nm Argon laser and detected through a band-pass emission filter of 530–600 nm . The images were oversampled by a factor of 1 . 5 in the Z direction ( Z-step sizes = 0 . 2 μm ) , with 12 bit dynamic range . Stack montages were obtained using the motorized stage controlled by the MultiTimeZ macro ( Carl Zeiss ) , which set up the coordinates and imaging conditions for each stack . Adjacent stacks had 10% overlap to guarantee the precision of later alignment and tracing . Using custom-written Matlab ( The MathWorks , Inc . ) programs , image stacks were median-filtered and resized to 512 × 512 in XY to have cubic voxels ( 0 . 2 × 0 . 2 × 0 . 2 μm ) . Each stack was then digitally resampled along either x- or y-axis to generate a series of cross-sections that were approximately orthogonal to the direction of most axons . All axons were reconstructed from the series of cross-sections using custom-modified Reconstruct program ( freely available from http://synapses . clm . utexas . edu/tools/reconstruct/reconstruct . stm; J . Lu , J . C . Fiala , J . W . Lichtman , unpublished data ) . Briefly , the modification incorporated a region-growing algorithm based on intensity threshold . The user presets the threshold and selects a point ( seed ) in an axon on one cross-section image . The program applies the region-growing algorithm to detect the contour of the axon on the image . Then it calculates the centroid of the contour , and propagates the centroid to the next image as the seed for the next cycle of edge-detection . The user can interrupt the progress of tracing at any time if aberrant region-growing occurs , and resets the threshold . Once all axons were traced out , they were rendered in 3D in Reconstruct and projected into 2D images ( one image per axon ) , which were manually assembled into complete montages for each axon in Adobe Photoshop ( Adobe Systems Inc . ) . The 2D montage of each axon was retraced with the NeuronJ plug-in ( http://www . imagescience . org/meijering/software/neuronj/ [75] ) to ImageJ ( http://rsb . info . nih . gov/ij/ , NIH ) to label and measure each axonal segment . The length and connectivity of axonal segments were transformed into a tree representation with custom-written Matlab programs . Reconstruction accuracy was confirmed in three different ways . First , different persons independently traced a series of overlapping stacks , and the tracing results contained no gross-level discrepancy that would have led to different interpretations of the connectivity relationship between axonal branches . Second , we traced individual axons from tri-color mice using only one channel , and compared the results to the tri-color images . In the line of tri-color mouse ( thy-1-KOFP × thy-1-YFP-H × thy-1-CFP-S ) , Kusabira-Orange fluorescent protein ( KOFP ) is expressed in 100% of motor axons ( J . Livet and J . W . Lichtman , unpublished data ) ; CFP and YFP are expressed each in a random subset of motor axons ( 24 ) . Images were taken from all three fluorescent channels , and gray-level data from the KOFP channel only ( Figure 1E ) was used for reconstruction of axonal profiles . The monochromatic tracing results for doubly labeled axons ( Figure 1F , yellow , KOFP + YFP; lavender , KOFP + CFP ) were identical to that shown in the RGB images ( Figure 1G , CFP , YFP , and KOFP were mapped to blue , green , and red , respectively ) . Third , we checked for abnormalities in tracing results such as axons looping back onto themselves or branches unconnected to any axon , and did not find any . We quantified the symmetry level of axonal arbors using the imbalance index I , which is defined as I = 2 × ( ∑all interior nodes |TR − TL| ) / ( n − 1 ) ( n − 2 ) . Here TR and TL are the number of terminals belonging to the right and left subtrees of the branching node , respectively , and n is the total number of terminals ( NMJs ) in the entire axon . Intramuscular axonal arbor length was defined as the total length downstream of a reference point common to all axons in the muscle . This reference point was chosen to be the first branching point of the axon that branched most proximally ( close to cell body ) in the muscle . Arbor length of axons that branched extramuscularly was defined to be the sum of arbor length of primary branches distal to the reference point . Axonal caliber was measured by dividing the volume V of an axonal segment ( length ∼ 50 μm ) slightly proximal to the reference point by the length of the segment . Volume was calculated as V = ∑i Ai × d , where Ai was the area of the axonal profile on the i-th image section and d is section thickness . The calculated axonal caliber was normalized to the sum of calibers of all axons entering the muscle . Relationships between axonal caliber , arbor length , and motor unit size were fitted with GraphPad Prism 5 for Windows ( GraphPad Software , Inc . ) . The spatial distribution of NMJs in each motor unit was parameterized along the lateral-medial axis ( along muscle ) and the rostral-caudal axis ( across muscle ) . The relative position of a NMJ was defined as its rank order in the connectome along the axis . These relative positions were used to test whether a motor unit is distributed uniformly in the endplate band ( Kolmogorov test [76] ) . In order to test whether the two primary subtrees of an axon tend to “exclude” each other , we implemented the generalized Wald-Wolfowitz test in Matlab . For each axon , a minimal spanning tree ( MST ) was constructed from the distance between NMJs using Prim's algorithm [77] . Edges connecting NMJs of different subtrees were removed , and the number of resultant disjoint subgraphs was counted . Significance level p was obtained through a Monte Carlo simulation in which NMJs were reshuffled between subtrees . Two subtrees were considered completely segregated if removing one edge partitioned the MST into 2 disjoint subgraphs , each corresponding to a subtree . TED is defined as the minimal number of operations ( insertion , deletion , and relabeling of nodes ) required to transform one tree into another . TED was calculated with a custom implementation in Matlab of a dynamic programming algorithm . Tree structures of axons belonging to the right-side muscles were flipped horizontally so as to compare them with the ones on the left side , with which they would overlap if there were no branching differences . We used Monte Carlo simulation to generate a large ensemble of axons all with the same number of terminals but random topologies consistent with the data . The simulation was based on a branching process model with level-dependent branching probabilities . In particular , probabilities of axons to terminate , bifurcate , trifurcate , etc , at each branching level were calculated from the full ensemble of reconstructed axons . 50 random “axons” were generated and their pair-wise TEDs ( 1 , 225 pairs in total ) were compared to that of real axons with the same number of terminals .
|
Conventionally , the organization of a neural circuit is studied by sparsely labeling its constituent neurons and pooling data from multiple samples . If significant variation exists among circuits , this approach may not answer how each neuron integrates into the circuit's functional organization . An alternative is to solve the complete wiring diagram ( connectome ) of each instantiation of the circuit , which would enable the identification and characterization of each neuron and its relationship with all others . We obtained six connectomes from the same muscle in adult transgenic mice expressing fluorescent protein in motor axons . Certain quantitative features were found to be common to each connectome , but the branching structure of each axon was unique , including the left and right copies of the same neuron in the same animal . We also found that axonal arbor length is often not minimized , contrary to expectation . Thus mammalian muscle function is implemented with a variety of wiring diagrams that share certain global features but differ substantially in anatomical form , even within a common genetic background .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"neuroscience"
] |
2009
|
The Interscutularis Muscle Connectome
|
A 26 year-old female patient presented to the Tropical Medicine outpatient unit of the Ludwig Maximilians-University in Munich with febrile illness after returning from Southern Africa , where she contracted a bite by a large mite-like arthropod , most likely a soft-tick . Spirochetes were detected in Giemsa stained blood smears and treatment was started with doxycycline for suspected tick-borne relapsing fever . The patient eventually recovered after developing a slight Jarisch-Herxheimer reaction during therapy . PCR reactions performed from EDTA-blood revealed a 16S rRNA sequence with 99 . 4% similarity to both , Borrelia duttonii , and B . parkeri . Further sequences obtained from the flagellin gene ( flaB ) demonstrated genetic distances of 0 . 066 and 0 . 097 to B . parkeri and B . duttonii , respectively . Fragments of the uvrA gene revealed genetic distance of 0 . 086 to B . hermsii in genetic analysis and only distant relations with classic Old World relapsing fever species . This revealed the presence of a novel species of tick-borne relapsing fever spirochetes that we propose to name “Candidatus Borrelia kalaharica” , as it was contracted from an arthropod bite in the Kalahari Desert belonging to both , Botswana and Namibia , a region where to our knowledge no relapsing fever has been described so far . Interestingly , the novel species shows more homology to New World relapsing fever Borrelia such as B . parkeri or B . hermsii than to known Old World species such as B . duttonii or B . crocidurae .
Relapsing fever , a bacterial disease caused by microaerophilic spirochetes of the genus Borrelia , can be found worldwide . Transmission is based on vectors such as body lice ( louse-borne relapsing fever ( LBRF ) ) or ticks ( tick-borne relapsing fever ( TBRF ) ) . Depending on the geographical region as well as the vectors present , many different Borrelia spp . are capable of infecting humans . Relapsing fever can be responsible for various febrile presentations that are clinically impossible to distinguish from other febrile diseases like malaria [1] . Symptoms include recurrent fevers , tachycardia , headache , conjunctivitis , hepatomegaly , splenomegaly , urine discoloration , asthenia , vomiting , myalgia and arthralgia . The mainstays of diagnosis are patient history , physical examination results as well as stained thin and thick blood films with the microscopic confirmation of spirochetes . Species differentiation is impossible by morphologic means and is dependent on molecular methods such as polymerase chain reaction ( PCR ) and sequencing [2–4] . Relapsing fever borrelioses can easily be treated with tetracyclines or penicillins [5–8] . So far , antibiotic resistance has not been reported . Mortality of untreated TBRF is generally in the low percentage range and may be associated with Jarisch-Herxheimer reactions occurring in less than half of the cases , however convincing data are missing especially for African TBRF [1 , 9] . Imported cases by returning travellers which could be studied , are also rarely reported [10] , although relapsing fever borrelioses are well known and common on the African continent [1 , 11] . Unfortunately , many African laboratories lack the ability to perform biomolecular tests , thus the exact species distribution as well as potential animal reservoirs are frequently unknown . Within Central , Southern and East Africa , mainly B . duttonii has been described , while further North also B . crocidurae and B . hispanica can be found as significant human pathogens [11] . Herein , we describe the case of a TBRF detected in a patient returning to Germany from a trip to the Kalahari Desert that was apparently not caused by any of the well-known spirochetes . The spirochetes detected in the blood film were examined by DNA amplification methods and were found to be more closely related to New World relapsing fever species than to the expected Old World species .
Written informed consent for this publication was obtained from the patient . The need for an Institutional Review Board statement has been waived by LMU Ethics committee .
A 26-year old German-native female presented to the OPD of the division of infectious diseases & tropical medicine at Munich university hospital with fever accompanied by headache , fatigue , generalized body pain and nausea for one day after returning from a four week holiday in Southern Africa . The patient had spent about seven days each in South Africa , Botswana , Zimbabwe and Namibia . She had not taken any malaria chemoprophylaxis , but travel related vaccinations such as those against hepatitis A and rabies had been given . The patient indicated that when travelling through the Kalahari , she had noticed during a stay in the Buitepos area between Namibia and Botswana , a large mite-like creature of about 5-6mm in size resting on the anterior part of the foot above the metatarsal bones 2/3 while wearing sandals . The bite occurred eight days prior to the onset of symptoms . The patient indicated similarity between the arthropod and pictures of soft ticks . The arthropod easily detached upon wiping it off the skin . A few days later , the patient detected a small , painless coin shaped erythema with a central brightening at the site of the bite , which again lasted several days until it disappeared . Throughout the four week travel period , no other health issues occurred . On the day of presentation , the patient had been travelling from Zimbabwe via Johannesburg to Munich and reported sudden fevers with chills , headaches and generalized muscle and body pain . During a stop-over in Johannesburg the patient presented to the local airport clinic . There , paracetamol was prescribed to ease symptoms for the rest of the journey after a rapid test for malaria was found to be negative . The patient denied suffering from diarrhoea , exanthema , cough or any other symptoms . The medical history of the patient was significant for a pelvic vein thrombosis with subsequent pulmonary embolism in 2003 . Hypercoagulability due to antithrombin III deficiency was diagnosed and pharmacological long-term therapy with phenprocoumon daily was initiated . There are no other medical conditions known in the patient , especially neither allergies , nor substance abuse or immunosuppression . At presentation in the outpatient clinic , the patient was afebrile ( 37 . 6°C ) and in stable general condition . The physical examination was without any pathological findings , except for light renal angle tenderness on palpation of the left side . Laboratory results were unremarkable apart from a slightly elevated C-reactive protein ( CRP ) with normal leukocyte count and erythrocyte sedimentation rate ( ESR ) ( see Table 1 ) . Rapid diagnostic tests for Dengue-fever and Malaria were negative . Urine analysis was within normal limits . In the microscopic evaluation of Giemsa stained thin and thick blood films , tiny spiral shaped bacteria were seen . The size was determined to be about 10 μm in length with a diameter as small as 0 . 5 μm . The spirochete-like organisms were loosely wound with only about 5–6 turns and suspected to be Borrelia spp . ( Fig 1 ) . Because of the risk for a Jarisch-Herxheimer reaction upon initiation of therapy , the patient was transferred as an inpatient to the department of infectious diseases at the neighbouring hospital Klinikum Schwabing . Upon arrival on the ward the patient was febrile ( 38 . 4°C ) and stable ( RR 110/70mmHg , pulse 100/min , SpO2 97% at room air ) . An electrocardiogram and abdominal sonography were unrevealing . After i . v . administration of 500 ml of 0 . 9% NaCl , doxycycline 100 mg ( i . v . ) was started . Shortly after the first dose of doxycycline , the patient complained about sudden increase of fever ( >40 . 0°C ) with chills as well as nausea and vomiting . Due to the clinical suspicion of a light Jarisch-Herxheimer reaction , the patient received prednisolone 100 mg , pethidine , antiemetics as well as further intravenous fluid substitution . Her condition improved rapidly within the following 30 minutes . The following doses of doxycycline 100 mg were administered orally ( p . o . ) twice daily without any further incidents . In particular , no further fevers or chills occurred . The daily laboratory examinations showed a transient dip in thrombocyte counts down to a minimum of 125 , 000/μl . After four days , the patient could be discharged in good physical condition and the antimicrobial treatment was continued as an outpatient for a total of ten days . Two months after the initial treatment , the patient presented again to the outpatient clinic with severe left-sided headaches , which had occurred for the first time about one month ago . After extended infectious disease workup and neurological examinations including cranial MRI , EEG and Doppler sonography of the brain vessels , the patient was diagnosed with migraine . A correlation with the previously undergone “Candidatus Borrelia kalaharica” infection was considered unlikely . The patient reported improvement of the headaches after osteopathic treatment; no documented sequelae remained six months after the infection . In BLAST searches using 16S rRNA and flaB PCR sequences top hits included B . anserina as well as New World relapsing fever species such as B . parkeri and B . hermsii . Genetic distance analyses conducted in MEGA revealed a value of 0 . 004 compared to strain VS4 from Tanzania ( for which the 16S rRNA sequence was available but did not turn up as hit in BLAST searches and was downloaded from GenBank ) and values of 0 . 006 compared to the 16S rRNA fragment of other known Borrelia species ( Table 2 ) . Genetic distances of the flagellin gene ( flaB ) fragment ( 685 bp ) were 0 . 059 , 0 . 064 and 0 . 066 to B . anserina , B . turicatae and B . parkeri , respectively ( Table 3 ) . This was also reflected in phylogenies ( Fig 2A and 2B ) . In the 16S rRNA phylogeny strain VS4 isolated from Mvumi , Tanzania [19] clustered next to “Ca . B . kalaharica” . In the flaB phylogeny “Ca . B . kalaharica” formed a sister clade to B . anserina ( Fig 2A ) . Using a flaB fragment of approximately 300 bp , comparison between the Borrelia strain investigated here with strains from the Mvumi region in Tanzania [19 , 20] revealed that several of the strains ( designated B . duttonii in GenBank ) showed the highest similarity to “Ca . B . kalaharica” ( S1 Table ) . Furthermore , phylogenetic analysis using the short flaB fragment indicated that several strains from Tanzania formed a sister clade to “Ca . B . kalaharica” ( S1 Fig ) . Using primers targeting glpQ produced a very small fragment ( 350 bp ) and readable sequences were not obtained suggesting non-specific amplification . It has been shown for many bacterial species that use of sequence fragment of several conserved housekeeping loci increases the discriminatory power between bacterial species and strains [21] . Housekeeping loci that were used successfully for multilocus sequence typing of Lyme borreliosis group spirochetes and relapsing fever spirochetes included clpA , clpX , nifS , pepX , pyrG , recG , rplB and uvrA [22 , 23] . For uvrA a good PCR amplification product and suitable sequence reads were obtained . Due to limitation of the amount of available DNA , analyses of additional genes were not possible . Fragments ( 900 bp ) of the uvrA gene revealed an identity of 92% to B . hermsii in a GenBank BLAST search with no hits to any of the Old World relapsing fever species . Similarly , the genetic distance to B . hermsii was 0 . 08 , the closest value for any of the analysed species ( Table 4 ) . Phylogenetic analysis suggests a closer relationship to B . anserina and New World RF species than to Old World RF species ( Fig 2C ) .
Sequences have been submitted to GenBank with accession numbers KT970516 ( flaB ) ; KT970517 ( uvrA ) ; KT954008 ( 16S rRNA ) .
|
A patient reported an arthropod bite on her dorsal metatarsal region 2/3 of her anterior foot during a trip to the Kalahari Desert in Southern Africa . Eventually , a rash developed at the site of the bite and high-grade fever started 8 days later . Spirochetes were detected in the blood smear and the patient was treated with the suspected diagnosis of tick-borne relapsing fever . The presence of Borrelia spp . in fresh patient blood could be confirmed using DNA amplification and sequencing techniques . Homology searches of the obtained sequences from 16S rRNA , flaB , and uvrA revealed surprisingly distant relationships to known Borrelia species . It is concluded that the infection was caused by a new species of tick-borne relapsing fever Borrelia capable of infecting humans that we propose to name “Candidatus Borrelia kalaharica” , which is described within this manuscript . The blood sample was discarded after initial analysis; therefore , no successful culture could be obtained .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"sequencing",
"techniques",
"borrelia",
"infection",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"tropical",
"diseases",
"microbiology",
"animals",
"relapsing",
"fever",
"bacterial",
"diseases",
"phylogenetic",
"analysis",
"neglected",
"tropical",
"diseases",
"molecular",
"biology",
"techniques",
"cellular",
"structures",
"and",
"organelles",
"bacteria",
"bacterial",
"pathogens",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"infectious",
"diseases",
"medical",
"microbiology",
"microbial",
"pathogens",
"borrelia",
"molecular",
"biology",
"molecular",
"biology",
"assays",
"and",
"analysis",
"techniques",
"ribosomes",
"arthropoda",
"biochemistry",
"rna",
"ribosomal",
"rna",
"cell",
"biology",
"nucleic",
"acids",
"spirochetes",
"biology",
"and",
"life",
"sciences",
"non-coding",
"rna",
"organisms"
] |
2016
|
"Candidatus Borrelia kalaharica" Detected from a Febrile Traveller Returning to Germany from Vacation in Southern Africa
|
Resequencing is an emerging tool for identification of rare disease-associated mutations . Rare mutations are difficult to tag with SNP genotyping , as genotyping studies are designed to detect common variants . However , studies have shown that genetic heterogeneity is a probable scenario for common diseases , in which multiple rare mutations together explain a large proportion of the genetic basis for the disease . Thus , we propose a weighted-sum method to jointly analyse a group of mutations in order to test for groupwise association with disease status . For example , such a group of mutations may result from resequencing a gene . We compare the proposed weighted-sum method to alternative methods and show that it is powerful for identifying disease-associated genes , both on simulated and Encode data . Using the weighted-sum method , a resequencing study can identify a disease-associated gene with an overall population attributable risk ( PAR ) of 2% , even when each individual mutation has much lower PAR , using 1 , 000 to 7 , 000 affected and unaffected individuals , depending on the underlying genetic model . This study thus demonstrates that resequencing studies can identify important genetic associations , provided that specialised analysis methods , such as the weighted-sum method , are used .
New technologies allow sequencing of parts of the genome of large groups of individuals [1] , and hereby initiate the next generation of large scale association studies . Resequencing studies can directly identify millions of rare mutations in the genome , and may therefore be able to identify disease-mutations that are not tagged by panels of common SNPs [2] . Resequencing may thus hold the key to detecting associations in the presence of genetic heterogeneity , where the genetic component of disease-risk is determined by multiple rare mutations , each with a low marginal effect on disease-risk ( i . e . low population attributable risk; PAR ) . Recent studies support the hypothesis that multiple rare mutations , each with a low marginal effect , may be a major player in genetic determination of susceptibility for some complex diseases [3]–[13] . Examples of genetically heterogeneous diseases include cystic fibrosis [14] , [15] , colorectal cancer [16] and probably schizophrenia [13] . Different genetic models may underlie genetic heterogeneity . One possibility is that multiple different variants located across the genome have independent influence on disease risk , such that each variant explains only a small fraction of all affected individuals . Another scenario is that the function of each haplotype of a gene is destroyed if one ( or more ) lethal mutations occur on the haplotype . In this manner , an individual must have at least one mutation on each of the two haplotypes to be predisposed for the disease ( see the Recessive-Set model in Figure 1 ) . In both of these models , the marginal PAR of each mutation may be very low , even when the disease is highly heritable . Association studies using panels of common SNPs are well suited for identifying variants each with a relatively high PAR , whereas multiple rare variants , each with a small PAR , are difficult to identify using these methods [17]–[24] . In cases where a single ( or very few ) common variants are expected to be associated with a disease , a variant-by-variant approach using the strongest marginal signal for each tested variant may be beneficial ( as discussed in [25] and [26] ) . On the other hand , when multiple rare mutations are expected to influence disease risk , an obvious approach is to group the variants according to function , such as genes , pathways and ultra conserved regions , and compare the group counts rather than the counts for each variant in the group . The rationale behind this grouping approach is that if many different mutations in a group affect disease risk , it may be beneficial to focus on the group rather than on each variant individually . The cohort allelic sums test ( CAST ) is an existing grouping method in which the number of individuals with one or more mutations in a group ( e . g . gene ) is compared between affected and unaffected individuals [5] , [26] , [27] . An alternative method using a grouping approach is the Combined Multivariate and Collapsing ( CMC ) method [26] . In this method all rare variants are collapsed , as in the CAST method , and the collapsed variants are treated as a single common variant which is analysed together with the other common variants using multivariate analysis [26] . In the CMC version used in [26] , rare variants are defined as those having a minor allele frequency ( MAF ) of at most 1% . In this study , we focus on a scenario in which a group of multiple rare mutations has been identified . In functional regions , one may choose to include only probable disease susceptibility mutations ( non-synonymous substitutions , frame shift mutations , etc ) in the group of mutations . Using only probable disease susceptibility mutations has the benefit that random variation due to non-associated variants may decrease . In this manner , association studies of groups of rare probable disease-susceptibility variants may be able to identify genetically heterogeneous mutations , and hence complement genome-wide analysis of common SNPs . Grouping of mutations according to functional elements , such as genes , has the added advantage of focusing on causal relations between genes and diseases , rather than just identifying highly associated genomic regions . Furthermore , since many ( millions of ) mutations are expected to be identified in a resequencing study of thousands of individuals [28] , grouping lowers the burden of multiple testing . We propose a weighted-sum method in which mutations are grouped according to function ( e . g . gene ) , and each individual is scored by a weighted sum of the mutation counts . To test for an excess of mutations in affected individuals , we use permutation of disease status among affected and unaffected individuals . By using permutation , the method adjusts for the weighting of the mutations and the requirement that a mutation must be observed to be included in the study . Note that permutation of disease status results in correct type I error even in the presence of linkage disequilibrium ( LD ) [29] , [30] , although relatively low LD is expected between rare variants [26] , [31] , [32] . The weighted-sum method deviates from the CAST method [5] , [27] by weighting the variants differently when determining the genetic load of an individual . By weighting the signals from each mutation , the weighted sum method accentuates mutations that are rare in the unaffected individuals , so that the test is not completely dominated by common mutations . In the CAST method , common variants will have a high impact on the group signal , and if many common mutations are present in a group , almost all individuals will have one or more mutations . To avoid this effect it may be necessary to use a threshold on the mutation-frequencies , as suggested in the CMC method [26] . A drawback of such frequency thresholds is that it can be difficult to select them in a biological meaningful way , and the outcome of the test will depend on the selection of thresholds . In the weighted-sum method we include mutations of all frequencies , but mutations are weighted according to their frequency in the unaffected individuals .
The weighted-sum method compares the number of mutations in a group of variants between samples of affected and unaffected unrelated individuals . It is designed to identify an excess of mutations in the affected individuals , compared to the unaffected individuals . Each variant belongs to a group ( gene , pathway , ultra conserved area , etc . ) and , for a group with L variants , the method is comprised of the following steps: Alternatively a p-value can be found by using a standard permutation test , where the p-value is found by ( k0+1 ) / ( k+1 ) , and k0 is the number of the k permutations that are at least as extreme as x . In such a testing framework , the permuting routine can be stopped if the estimated p-value ( and its precision ) reaches a certain level; e . g . if the p-value , minus three times the estimated standard deviation of the p-value , is above the significance threshold . Such a permutation strategy may be as fast as the approximation strategy , since fewer than 1000 permutations are needed to reject the hypothesis of association in many cases . Throughout this paper , the approximation strategy is used because it runs fast for power simulations . Another reason for using the approximation strategy ( rather than standard permutation with a stopping rule ) is to produce Uniform ( 0 , 1 ) distributed p-values ( under the null hypothesis; see Figure S2 ) for all the tests conducted , which is preferred if further analyses of the p-values are conducted in e . g . a pathway analysis . The standard permutation approach can only produce uniformly distributed p-values under the null hypothesis if no stopping rule is used , which is a computationally expensive approach . Whether using the approximation or standard permutation strategy , permutation of the case-control labels maintains the LD structure of the genetic data . Thus , the test is valid ( i . e . has correct false positive rate ) whether or not the variants are in LD . The weighted-sum method is compared to the CAST , CMC , and variant-by-variant methods , which were discussed in the introduction and are described in more detail in Comparison with other Methods . For each set of parameters , 100 datasets are simulated , the four methods are applied , and the proportions of significant outcomes are used as the power estimates . To mimic a genome wide study of about 20 , 000 fairly independent human genes , we calculate a p-value for each gene , and use a significance threshold of 0 . 05/20000 = 2 . 5×10−6 in all power simulations . To evaluate the weighted-sum method on rare variants with the frequency-spectrum of a naturally occurring population , we used resequencing data from the Encode III project ( ftp://ftp . hgsc . bcm . tmc . edu/pub/data/Encode ) . In the Encode III project ten 100 kb Encode regions were resequenced in different human populations , and all substitutions were identified ( see http://www . hgsc . bcm . tmc . edu/projects/human/ ) . To mimic a disease-resequencing study , we grouped all exonic variants of each Encode region , and compared the number of rare variants between the two largest populations: the African YRI population ( 120 individuals; including 60 individuals from HapMap phase I and II ) and the Central European CEU population ( 119 individuals; including 60 individuals from HapMap phase I and II ) . Only variants that passed the quality control filter for the ENCODE III study were used ( see http://www . hgsc . bcm . tmc . edu/projects/human/ ) . The genotype data were downloaded as the ENCODE III draft release I ( on August 11th , 2008 ) , and the “Gencode Ref ( encodeGencodeGeneKnownMar07 ) ” track in the UCSC Genome Browser [37] was used to define exon positions in each ENCODE region . Exonic variations were reported for only five of the ten ENCODE regions , and hence only these five regions were used . The CAST method , as described in [27] , corresponds to the method used in [5] . In brief , for each group of variants , it compares the number of individuals with one or more mutations between affected and unaffected individuals , using a standard χ2 or Fisher exact test . In this study , we use the Fisher exact test throughout to avoid bias due to distributional approximation . In the variant-by-variant approach the genotype frequencies of each variant are compared using the one-sided Fisher's exact test , and the significance level of the group is found by Dunn-Sidak correction [38] of the smallest p-value in the group . Note that the Dunn-Sidak correction is very similar to the Bonferroni correction , as the Bonferroni correction is an approximation of the Dunn-Sidak correction . Whereas the Bonferroni correction is slightly conservative for independent tests ( such as the independent variants in the power simulations ) , the Dunn-Sidak correction has the benefit of being exact . The CMC method is implemented according to the description in [26] . In brief , for the CMC method all rare variants are collapsed , as in the CAST method , and the collapsed variants are treated as a single common variant which is analysed together with the other common variants using multivariate analysis [26] . We used the Fisher product method [42] , [43] for multivariate analysis , rather than the Hotelling's T2 method , because it allows for one-sided testing , and hence allowed a fair comparison for the CMC method . Note that if a two-sided test were used for the CMC method , the power estimates would then have been too low compared to the variant-by-variant and weighted-sum methods . The weighted-sum method is implemented as described above , using k = 1000 permutations in step C . In all power simulations Iij∈{0 , 1 , 2} is used in step B ( even when the dataset is simulated under a recessive or dominant model ) .
The mutation frequencies are sampled according to Wright's formula ( see Methods ) , and hence mutations are very rare for some variants . Using 1000 affected and 1000 unaffected individuals , mutations are on average observed at only 49 . 4% of the variants ( sd: 4 . 9% ) . This means that when e . g . 100 variants are sampled , on average 49 . 4 variants contain at least one mutation , and are hence tested for association . This level is in concordance with the level from human resequencing studies [5] , [7] , [16] . Under the baseline parameter settings ( see Methods ) it is seen that the CMC method , as reported in [26] , has better performance than the variant-by-variant and CAST methods , but the weighted sum method has even better performance ( Figure 2 ) . The weighted-sum method identifies groups with a PAR of 10% , with at least 80% power , for all genetic models ( Figure 2 ) . To investigate whether the weighted-sum method is robust under other model parameters , we fix the group PAR at 10% , and vary the other parameters one by one . The number of variants that contribute to the disease-risk ( D-variants ) determines the marginal PAR of each variant in the group , such that a low number of D-variants yields a high marginal PAR . Accordingly , all investigated methods perform well when the number of D-variants is low , and hence the marginal PAR is high ( Figure 3 ) . When the number of D-variants rises , and hence the marginal PAR of each variant drops , the power to identify a disease-group falls ( Figure 3 ) . For the weighted-sum method , the effect of the number of D-variants depends on the genetic model . For the recessive models , it is able to identify even large groups of variants , whereas it is more sensitive to the number of D-variants when the heterozygote contributes to disease-risk ( Figure 3 ) . The proportion of D-variants likewise influences the power . Under the Recessive-Set model , both the CAST and the CMC methods perform well when a reasonably high proportion of the variants contribute to disease-risk , whereas both the variant-by-variant and the CAST method are unable to identify disease-groups under the other scenarios ( Figure 4 ) . On the other hand , the weighted-sum method is generally robust to a low proportion of D-variants in the group , but a higher proportion of D-variants yields higher power ( Figure 4 ) . Note that the probability of mutant-haplotypes ( pM ) in unaffected individuals under the Recessive-Set model does not have a large impact on the power ( Figure S3 ) . The number of individuals needed to identify a disease-associated group depends strongly on the underlying genetic scenario . With n = nA = nU = 1000 individuals , a group with a PAR of 1% can be identified under the Recessive-Set model , while a group with a PAR of 5%–10% can be identified under the other models . A study with n = 7000 individuals can identify a group with a PAR of 2% under all genetic models ( Table 1; see Tables S1 and S2 for equivalent tables for the CMC and CAST methods ) . To cover a scenario where the mutation-frequencies are distributed according to a natural existing population , we used resequencing data from 120 individuals from the African YRI population and 119 individuals from the Central European CEU population . In this example , we test for overrepresentation of rare exonic variants in the YRI population compared to the CEU population in each Encode region . Such an overrepresentation is expected since the YRI population generally shows higher diversity than the CEU population [39] , and hence more rare variants are expected . Exonic variants are grouped for each ENCODE region , to mimic a disease-resequencing study like the ones reported in human resequencing studies [5] , [7] , [16]; as a result , 5 groups of 2–72 polymorphic variants are obtained ( see Table 2 ) . As with the simulated data , the weighted-sum method generally shows higher power than the alternative methods to identify an excess of rare variants in the Encode data ( Table 2 ) . Table 2 shows that large groups of variants generally yield lower p-values than small groups . This is expected in the case of heterogeneity , where inclusion of more variants will lead to a stronger combined signal , and hence a lower p-value . In the current un-optimized implementation of the weighted-sum method , a genome wide analysis of 20 , 000 groups , with 50 polymorphic variants each , using nA = nU = 1000 individuals can be completed in approximately 600 CPU hours on a standard stand-alone machine ( Intel Pentium Dual 2 GHz , 2GB RAM ) . When the number of permutations ( k ) is 500 instead of 1000 , the results are unaffected ( results not shown ) but the computing time is halved , however since the test is fast we use k = 1000 in this study . Note that the computation time is linear in number of individuals and number of permutations ( see Table S3 ) .
In this work , we propose a specialised method to identify multiple rare mutations underlying a genetically heterogeneous disease . Analysis of real data and power simulations show that the proposed weighted-sum method performs very well compared to existing methods . This demonstrates that the use of specialised analytical methods can improve power to identify genetic components of complex ( genetically heterogeneous ) diseases . On the other hand , it must be kept in mind that the power of such specialisation is at the cost of generality , and therefore the methods must be used in combination with other strategies covering other biological scenarios such as the common variant common disease scenario . It must further be noticed that all methods using the grouping approach ( i . e . CMC , CAST and weighted-sum ) are sensitive to misclassification of which allele is treated as the mutation ( i . e . disease-related allele ) . If disease-related alleles from some variants are grouped with wild-type alleles from other variants it may hide a true signal . As stated in the Background section , it may be natural to treat e . g . non-synonymous substitutions , frame shift indels and very rare alleles as mutations , but when there is no information to classify the alleles , grouping methods may not be useful . Instead the idea from the CMC method can be used , such that the variants that can be grouped are analysed with a grouping statistic ( e . g . the weighted-sum method ) , and all other variants are analysed variant by variant or by multivariate analysis . The weighted-sum method is designed for resequencing data , since this technology allows rare mutations to be observed directly . The use of inferred haplotypes from tag SNP studies is a current approach to evaluation of unobserved variants , but this approach fails when the unobserved variants are rare; the tag SNP approach is hence not suited for the scenario of multiple rare disease-mutations [2] . Alternatively , familial linkage studies are a strategy to identify mutations underlying genetically heterogeneous diseases , but when the marginal effect of each mutation is low , it may be difficult to obtain a sufficient number of affected individuals to detect a disease association [40] , [41] . The weighted-sum method can be adapted to a wide range of study designs , by e . g . the following: ( A ) Using the posterior probability of each genotype rather than the most probable genotype . ( B ) Analysing mutations in conserved areas by weighting each mutation according to the measure of conservation; this is an extension of the conservation base selection criterion from [7] . ( C ) Analysing continuous traits by testing for correlation between genetic ranks ( or scores ) and the trait measure . Furthermore , the weighted-sum method can be used for other types of data that can be grouped according to function . Such data include for example methylation measures , where multiple regions/sites can be methylated in promotor regions ( i . e . the CpG islands ) . Note that ranking can be omitted in the test procedure , so the test statistic is the sum of the genetic scores ( γi ) of all affected individuals , rather than the sum of ranks . In the tests performed in this study , the two procedures yield very similar results ( results not shown ) , but we prefer to use the ranking procedure because it is robust to outliers . The mutation weights ( ) can be chosen in an infinite number of ways . We suggest using the estimated standard deviation of the total number of mutations in the sample ( including affected and unaffected individuals ) , under the null hypothesis of no frequency differences between affected and unaffected individuals . This choice of weight ensures that all variants in a group contribute equally to the weighted sum , under the null hypothesis . The weight of each mutation is determined by its frequency in the population of unaffected individuals only . In this way , a mutation which is common among unaffected individuals has lower weight than a mutation which is rare among the unaffected individuals . If further information about the mutations is available , it may be incorporated in the weights . Such information could include the estimated impact of a mutation or a measure of conservation of the surrounding region ( as discussed above ) . Analysis of pathways can be done in two different ways . One way is to use the pathway as a group , and run the test on the entire pathway . On the other hand , for large pathways , it may be beneficial to use a method that allows a gene with a strong signal to have a high impact on the combined pathway test-statistic ( T ) . If a pathway contains G non-overlapping genes , a method to do this is to use the weighted-sum method on each gene , and combine the resulting p-values ( π1 , … , πG ) with the Fisher product test statisticSince π1 , … , πG are i . i . d . uniformly ( 0 , 1 ) distributed under the null-hypothesis , T is χ2-distributed with 2G degrees of freedom , and can be evaluated accordingly [42] , [43] . This method allows for fast analysis of different pathways , using the results from the gene-analysis , and can thereby assist in the functional analysis of a disease association study . Simulating inheritance of a genetically heterogeneous disease can be performed in different ways . To ensure that all variants have a low effect , we have chosen to simulate all variants within a group with the same PAR . An alternative scenario is to simulate all variants , in a group , with the same relative risk ( RR ) , and let the PAR vary according to the mutation-frequency . Under this scenario , a single , or few , common mutations may carry a large part of the total risk , and this scenario is hence equivalent to a scenario with a single , or few , disease-contributing variants . A few common variants carrying a relatively large risk is exactly the what studies using panels of SNPs are designed for , and our focus has therefore been on scenarios where the disease risk can not be explained by a few variants . Note further that all investigated methods are able to identify cases where a few mutations carry a large part of the total risk ( see Figure 3 ) . We have further included the comparison of the Encode populations , to cover a scenario where the mutation-frequencies are distributed according to an actual population . In summary , we show that the weighted-sum method is powerful for identifying multiple rare mutations underlying genetically heterogeneous diseases . Under some genetic scenarios , 1000 affected and 1000 unaffected individuals are sufficient to identify e . g . a gene with a PAR of only 1% , corresponding to an odds ratio of 1 . 1 . These findings thus demonstrate that resequencing studies have the potential to identify important genetic associations , provided specialised analysis methods are used .
|
Resequencing is an emerging tool for the identification of rare disease-associated mutations . Recent studies have shown that groups of multiple rare mutations together can explain a large proportion of the genetic basis for some diseases . Therefore , we propose a new statistical method for analysing a group of mutations in order to test for groupwise association with disease status . We compare the proposed weighted-sum method to alternative methods and show that it is powerful for identifying disease-associated groups of mutations , both on computer-simulated and real data . By using computer simulations , we further show that resequencing a few thousand individuals is sufficient to perform a genome-wide study of all human genes , if the proposed method is used . This study thus demonstrates that resequencing studies can identify important genetic associations , provided that specialised analysis methods , such as the proposed weighted-sum method , are used .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"genetics",
"and",
"genomics/disease",
"models",
"genetics",
"and",
"genomics/genetics",
"of",
"disease",
"public",
"health",
"and",
"epidemiology/epidemiology",
"mathematics/statistics",
"genetics",
"and",
"genomics/medical",
"genetics"
] |
2009
|
A Groupwise Association Test for Rare Mutations Using a Weighted Sum Statistic
|
The role of sensory systems is to provide an organism with information about its environment . Because sensory information is noisy and insufficient to uniquely determine the environment , natural perceptual systems have to cope with systematic uncertainty . The extent of that uncertainty is often crucial to the organism: for instance , in judging the potential threat in a stimulus . Inducing uncertainty by using visual noise , we had human observers perform a task where they could improve their performance by choosing the less uncertain among pairs of visual stimuli . Results show that observers had access to a reliable measure of visual uncertainty in their decision-making , showing that subjective uncertainty in this case is connected to objective uncertainty . Based on a Bayesian model of the task , we discuss plausible computational schemes for that ability .
Every single human action happens in a context of uncertainty , being based on incomplete knowledge and undertaken despite unpredictable consequences . When faced with uncertainty , humans employ heuristics [1] , [2] and show characteristic biases in their decision [3] . The neural structures involved in some of these decisions are now being identified [4]–[6] . Before one can make decisions that depend on uncertain information , the degree of uncertainty must be evaluated . The basic question of how well humans do at evaluating their own uncertainty remains largely understudied . Uncertainty is a familiar concept in cognitive science , in particular thanks to Signal Detection Theory ( SDT; Green and Swets 1966 ) . In a typical psychophysical task , an observer has to detect small contrast increments near threshold . The uncertainty in this task comes mostly from internal variability: because of fluctuations in her internal representation of contrast , the observer makes mistakes and is uncertain about the correctness of her decisions . Unfortunately for the experimenter , this source of the uncertainty is internal to the observer and therefore only indirectly controllable . Now consider another difficult perceptual task: listening to a speaker among cocktail-party chatter . Here the difficulty depends not so much on variability in the brain , but rather on interactions between the different voice signals: the one emitted by the speaker you aim to listen to , and the sound of other voices . Even with the volume of the other voices staying the same over time , difficulty will depend on the languages spoken , the gender of the speakers , and other sources of confusion . More generally , background chatter plays the role of noise , and difficulty will vary based on how much signal and noise covary . An analogous visual task can be obtained by adding visual noise to a signal –random perturbations to the stimuli shown to the observer . Using visual noise , we are in a position to manipulate the objective uncertainty: objective uncertainty is inversely related to the amount of task-relevant information available in the stimulus . Concurrently , we can measure the perceived uncertainty of the observer , the level of confidence she actually reports . We introduce three experiments where we manipulate objective uncertainty and study its relationship with perceived uncertainty . In the first two experiments , observers were presented with pairs of images of oriented objects embedded in high levels of noise , and had to report the orientation of the image of their choice . Even though the two images contained the same level of noise , the particular noise structure made one image orientation more certain than the other . We found that observers reliably chose the more certain of the two images , thereby providing evidence of a capacity to accurately evaluate objective uncertainty . We confirmed this in another experiment , in which we held the objective uncertainty of one of two stimuli fixed while varying the other , and asked observers to pick the less uncertain one . The greater the difference in uncertainty was , the greater the chance that observers picked the less uncertain stimulus , showing that uncertainty discrimination behaves similarly to normal psychophysical tasks . In a third experiment , we extend our results to a letter discrimination task . We discuss plausible computational mechanisms for achieving these results .
To determine whether observers did effectively pick the less uncertain stimuli , we contrasted two conditions . In the so-called True Choice ( TC ) condition , the two stimuli presented resulted from independent draws from the same noise distribution . Note that two stimuli with the same average noise level , as is the case here , can still vary in the objective uncertainty they induce , because different realizations of the same noise distribution can make the stimulus more or less ambiguous . In that case there is a benefit to be had in choosing the less uncertain of the two: this gives observers a higher chance of responding correctly than if only one stimulus is available . In the other condition , the False Choice ( FC ) condition , we removed that benefit: the first stimulus was computed the normal way , but the second was obtained by flipping the top one either once or twice ( Figure 2 ) . We took advantage of the underlying symmetry of our templates: flipping the first template left-to-right yields the second , and flipping the second bottom-top yields back the first . By applying these transformations to a noisy version of our template , we were able to create two stimuli that differed pixel-to-pixel , but were equivalent from the point of view of the classification task and thus carried equal objective uncertainty in that context . In the False Choice case , there is therefore nothing to be gained by choosing one rather than the other . At no point in the experiment were observers aware of the existence of the two conditions . The two stimuli presented always had equal contrast , preventing observers from using a heuristic of selecting the lower-contrast stimulus as the most certain . The False Choice condition therefore provides the performance baseline that will be used to determine whether or not observers are able to successfully compare objective uncertainties . We measured observers' performance , defined as proportion of correct classifications , in the two conditions across five different signal-to-noise ratios , chosen to span a range of performance between approximately 60 to 85% . Both the signal-to-noise ratio and the condition each trial belonged to were randomized . If observers are able to make accurate judgments of objective uncertainty , then we expect that measured performance will be higher in the TC than in the FC condition . As expected given the nature of the task , mean performance for all observers grew with increased signal-to-noise ratio . More interestingly , however , mean performance is higher in the TC condition than in the FC condition , which translates into lower performance thresholds in the TC condition ( Figure 3 a and b ) . To establish that the effect is genuine we used a model comparison technique . We used a likelihood-ratio test to evaluate the effect of True Choice versus False Choice ( details in Text S1 ) . Using two psychometric functions , one per condition , rather than one psychometric function for both conditions provides a significantly better fit to performance data ( Nested hypotheses test [10]: p = 0 . 0004 , , d . f . = 24 ) . It appears then that observers were able to take advantage of the True Choice condition , by choosing the less uncertain stimulus a majority of the time . It seems reasonable that , should the ability to pick the less uncertain stimulus be present , the probability of choosing the correct stimulus ought to be an increasing function of the magnitude of the difference: the more the two stimuli differ in their uncertainty , the more likely observers are to choose the right one . We evaluate that by regressing observers' choices of stimuli on the difference of log-entropies ( Text S1 ) . We found a highly significant effect ( details in Text S1 ) of the difference in uncertainty on the probability of choosing the bottom stimulus: in other words , the more uncertain the bottom stimulus compared to the top one , the less likely observers were to choose the bottom one . This last result hints at a more general property: in all psychophysical discrimination tasks , the larger the difference between two stimuli , the more reliable discrimination is . For example , when asked to compare the length of two lines , an observer's responses are likely to be better predictable when the two lines differ by 20 cm rather than 1 . In a second experiment , we sought to confirm our findings by checking that discrimination of uncertainty behaves in the same way . The task was identical to that of experiment 1 , but instead of introducing a False Choice condition , we manipulated the stimuli such that one – the standard – had always the same level of uncertainty and the other – the test – had lower uncertainty . We show in the supplementary material that generating random stimuli with a controlled level of uncertainty can be achieved using a simple orthogonal projection . Mathematically , the space of all possible stimuli of the kind used here can be described in terms of the contrast of individual pixels by having one dimension ( one axis ) for each pixel . Then the two templates are two points u , v in that space , and stimuli obtained by adding white noise to a template are other points , forming Gaussian point clouds around the templates . To decide whether a point is more likely to belong to the left-tilted template rather than the right-tilted one , a simple geometrical rule describes the ideal strategy . Imagine drawing a line between u and v , as in figure 4 , where we illustrate the problem for stimuli with only 2 pixels . Now draw the plane ( in higher dimensions; the hyperplane ) that is orthogonal to the line and cuts through it at the mid-point . Then any stimuli falling on the same side of the plane as u we will call “left-tilted” and any falling on the side of v we will call “right-tilted”: the plane represents the decision boundary . Stimuli falling right on the hyperplane are completely ambiguous: both categories are equally likely . In fact , it is possible to show that the uncertainty of a stimulus is given by its ( unsigned ) distance to the decision boundary . Then the set of stimuli of fixed uncertainty is the set of points that are of the same distance to the decision boundary , and that set is simply the union of two parallel planes . We therefore generated our stimuli by constraining them to lie on a plane of distance d to the decision boundary . Standard stimuli were always on a plane of distance dstandard and test simuli were on a plane of distance dtest . The difference between dstandard and dtest was varied parametrically between 4 different levels: we expected the observers to more reliably choose the test stimulus as the difference increased . The results appear in figure 5: the larger the difference in uncertainty between standard and test , the more likely observers were to choose the test stimulus . We adapted the noise level to each observer's performance , so the distances used varied between observers . We normalise them with respect to the expected distribution of the distance to the hyperplane for the noise level chosen ( see Text S1 ) . The effect of the difference is significant for every observer as modeled by logistic regression of stimulus choice on difference in uncertainty ( t-test for Generalised Linear Models coefficients , all p-values at 10−3 or below ) . This confirms that uncertainty behaves in that respect just like other psychophysical quantities: the more dissimilar two stimuli are on that scale , the more predictable observers' judgments are . In experiments 1 and 2 , the underlying visual task is orientation discrimination under noise , with templates identical in every way except for one basic attribute – their orientation . To check that our results were sufficiently general , we ran a variant of experiment 2 using a letter discrimination task . Observers had to discriminate between the letters ‘T’ and ‘X’ ( shown on figure 5 ) , a pair chosen because the corresponding characters correlate very little . Except for the nature of the templates , experiment 3 was identical to experiment 2 and we replicated its results ( figure 5 ) : observers were more likely to pick the less uncertain stimulus when the difference in uncertainty was larger . Our results thus generalize to more sophisticated visual tasks . Our results imply that observers had access to some estimate of the uncertainty in the orientation task . How is that estimate computed ? Do observers have effective access to a probability distribution over perceptual hypotheses , from which they can estimate their own uncertainty ? Or do they rely on more limited information ? To investigate that question we evaluated two distinct families of models that compute uncertainties globally over the full distribution for the first , and locally for the second . We begin by defining the following quantities: let r and s be two stimuli , represented as vectors of pixel luminances . Call u and v the left-tilted and right-tilted templates . Then and are measures of how “different” r is to u and v , respectively . If r is more like u than v ( i . e . , ) , then it is more likely to have been generated from u , and hence the observer should respond “left-tilted” for stimulus r . In comparing the uncertainty between two stimuli - choosing between r and s - the following procedure is exactly equivalent to the strategy of the “ideal observer” ( i . e . , the strategy that maximizes performance , see Text S1 ) . Compute as ( 4 ) and choose r if , r otherwise . This corresponds to evaluating uncertainty based on the full posterior distribution ( see equation 1 ) : uncertainty is low if one hypothesis corresponds to the data much better than the other , and high otherwise . We call this model the difference of responses model . Another strategy , perhaps simpler for the observer , is to evaluate uncertainty based only on how well the best hypothesis fits the data . We call this the maximum response model . The same measures of distances are computed as in the first model , but only the maximum is retained for each stimulus . The observer then compares the two maxima ( 5 ) Put into perceptual terms , this corresponds to a strategy of picking the stimulus that seems to have a more salient dominant orientation , when the templates were Gabor patches , or the stimulus that was more “letter-like” , when the templates were characters . In statistical terms this is equivalent to evaluating uncertainty based on the magnitude of the likelihood of the maximum-likelihood hypothesis ( Methods ) , a strategy that is sub-optimal for our task but still gives an improvement over choosing between the two stimuli at random . Both hypotheses are realistic from a neural-computation point of view . Computing and is nothing more than a linear filtering of the neural input: although some important non-linearities have been identified in visual orientation discrimination , linear filtering remains the basic operation in all models [11] , [12] . Computing the decision variables , whether dabs and dmax , is a simple non-linear step readily implementable in a neural system . To test those models we make the same assumption we did for regressing choice on difference in log-entropy: the higher dabs and dmax , the more likely observers are to choose the bottom stimulus . As above , we compute the decision variables for every trial and we fit a linear binomial regression model to the responses ( Text S1 ) . Our models give for each trial a choice probability . On figure 6 we plot the percentage prediction correct ( i . e . , the proportion of trials where the model predicted with p> . 5 the choice the observer actually made ) . The two models have the same number of degrees of freedom , and can be directly compared . Both predict the data significantly better than chance , but the maximum response has a significant lead . Our data therefore point to a likelihood-based evaluation of visual uncertainty , rather than one based on the full posterior distribution .
In summary , we demonstrate here that humans display second-degree knowledge of a visual discrimination task: not only are they able to detect what signal is in the noise ( first-degree knowledge ) , but also to estimate how uncertain that knowledge is , at least comparatively . Why humans should be so well calibrated to what is in essence a laboratory task rather than a natural one is a question that deserves attention . It is possible that they learn the statistical properties of the task over time , although we find no conclusive evidence for that in our data ( see Text S1 ) . Previous research lacked an objective standard to compare subjective judgements to , and relied on ratings [13] . Various biases have been reported in human confidence judgments , including over- and under-confidence , global/local inconsistencies , as well as inter-cultural differences [14]–[17] . The forced-choice method we outlined here allows one to test human observers' objective capacity to detect differences in uncertainty contained in a task , and to evaluate possible computational mechanisms much more rigorously . It is a potentially important methodology in the study of discrepancies between visual performance and confidence , a topic many believe to be connected to the wider issue of awareness [18] , [19] , but potentially also in investigations of metacognition in non-human species [20] , [21] . Our work is in tune with a variety of current research that tries to understand visual function as a form of Bayesian inference [22]–[25] . These theories posit that the visual system explicitly encodes probability distributions over perceptual hypotheses . In that context , it makes intuitive sense that the system should be able to measure the uncertainty of such a distribution: comparing two uncertainties as we do here is rarely needed as such , but comes into play in more complicated decisions . Just as a low feeling of confidence in an item to be memorized is a clue that further study is needed [26] , high visual uncertainty signals that more information is needed , making precise evaluation of visual uncertainty an essential aspect of exploration mechanisms [27] . The results given here agree with other studies that have found unexpectedly accurate decision-making in perceptual [28] , [29] and motor systems [30] , [31] . These results imply that uncertainty is dealt with at an implicit level: unlike them , we require observers to make explicit comparisons between levels of uncertainty . The observers who took part in our experiment nevertheless found the task quite intuitive: indeed , we often make comparative judgments of visual uncertainty “in the wild” , as when we judge if we see better from one vantage point than another . Generally , we expect that confidence measures have the potential to play a larger role in computational investigations of perceptual decision-making . The evaluation of uncertainty is a necessary first step in any statistical decision-making system , and biases and approximations in evaluating uncertainty will cause sub-optimal decisions . A systematic study of the evaluation of uncertainty in the visual system will help uncover the shortcuts taken by the brain in making perceptual decisions . Our method can be generalized to other noise models , other sensory modalities , and other tasks . But showing that fine-grained discrimination of uncertainty can be done is of course not an end in itself: uncovering how that essential operation is achieved in the brain is a natural next step .
This study was conducted according to French guidelines on research involving human participants . All participants gave informed consent . The experimental method was the same as in experiment one , unless indicated otherwise . The experimental method was the same as in experiment 2 , unless indicated otherwise .
|
Most work in vision science focuses on the question of why we perceive what we do , and we now have many models explaining what physical properties of a stimulus make us see depth , colour , etc . Here we ask instead what makes us feel confident in our visual perception: in the context of a visual task , what are the physical properties of the stimulus that will make us think we are doing the task well ? The mathematical framework of Bayesian statistics provides an elegant way to frame the problem , by assuming that the visual system is trying to estimate physical properties of the world from incomplete , sometimes unreliable visual information . Objective uncertainty will therefore depend on the quality of the information available in the stimulus . In our experiments we compare objective uncertainty—as computed using the Bayesian framework—with subjective uncertainty , the confidence observers report about their visual percepts . To this end , we use a visual task with well-defined statistical properties , discrimination under noise . We report a surprising degree of agreement between objective and subjective uncertainty , and discuss possible computational models that could explain this ability of the visual system .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"neuroscience/psychology",
"neuroscience/natural",
"and",
"synthetic",
"vision"
] |
2009
|
Evaluation of Objective Uncertainty in the Visual System
|
Enterotoxigenic Escherichia coli ( ETEC ) are common causes of diarrheal morbidity and mortality in developing countries for which there is currently no vaccine . Heterogeneity in classical ETEC antigens known as colonization factors ( CFs ) and poor efficacy of toxoid-based approaches to date have impeded development of a broadly protective ETEC vaccine , prompting searches for novel molecular targets . Using a variety of molecular methods , we examined a large collection of ETEC isolates for production of two secreted plasmid-encoded pathotype-specific antigens , the EtpA extracellular adhesin , and EatA , a mucin-degrading serine protease; and two chromosomally-encoded molecules , the YghJ metalloprotease and the EaeH adhesin , that are not specific to the ETEC pathovar , but which have been implicated in ETEC pathogenesis . ELISA assays were also performed on control and convalescent sera to characterize the immune response to these antigens . Finally , mice were immunized with recombinant EtpA ( rEtpA ) , and a protease deficient version of the secreted EatA passenger domain ( rEatApH134R ) to examine the feasibility of combining these molecules in a subunit vaccine approach . EtpA and EatA were secreted by more than half of all ETEC , distributed over diverse phylogenetic lineages belonging to multiple CF groups , and exhibited surprisingly little sequence variation . Both chromosomally-encoded molecules were also identified in a wide variety of ETEC strains and YghJ was secreted by 89% of isolates . Antibodies against both the ETEC pathovar-specific and conserved E . coli antigens were present in significantly higher titers in convalescent samples from subjects with ETEC infection than controls suggesting that each of these antigens is produced and recognized during infection . Finally , co-immunization of mice with rEtpA and rEatApH134R offered significant protection against ETEC infection . Collectively , these data suggest that novel antigens could significantly complement current approaches and foster improved strategies for development of broadly protective ETEC vaccines .
The enterotoxigenic Escherichia coli ( ETEC ) are among the most common causes of infectious diarrhea worldwide . Importantly , ETEC are disproportionately represented in cases of severe diarrheal illness as well as in deaths due to diarrhea among young children in developing countries [1] . These pathogens cause diarrhea by the elaboration and effective delivery of heat-labile and/or heat-stable enterotoxins to intestinal epithelial cells where they stimulate production of cyclic nucleotides ultimately activating the cystic fibrosis transmembrane regulator ( CFTR ) with resulting net efflux of fluid into the intestinal lumen[2] . Plasmid-encoded colonization factors ( CFs ) , discovered [3] shortly after these organisms were identified as a causative agent of cholera-like diarrheal illness[4–6] , are thought to be essential for effective colonization of the small intestine and required for ETEC pathogenesis . Following early studies suggesting a pivotal role for these structures[7 , 8] , CF antigens have defined the basis for most subsequent ETEC vaccine efforts [9 , 10] . However , one factor complicating development of a broadly protective vaccine for ETEC has been the general plasticity of E . coli genomes[11] , and the significant antigenic heterogeneity of the CFs . To date , at least 26 antigenically distinct CF antigens have been described[12] . The lack of appreciable cross-protection afforded by these antigens combined with the complex landscape of CFs portrayed in ETEC molecular epidemiology studies continue to complicate rational CF antigen selection[13] . Antigenic heterogeneity , recent failure of LT-toxoid-based vaccine strategies[14 , 15] , as well as the need to optimize the performance of live-attenuated vaccines currently in clinical trials [16–18] have highlighted the need to identify additional virulence molecules that might be targeted in ETEC vaccines . Recent efforts led to the identification of two loci discovered on the same virulence plasmid of ETEC strain H10407 that encodes the CFA/I colonization factor . These include the etpBAC two partner secretion locus responsible for production and export of EtpA[19] , a novel adhesin molecule which bridges highly conserved regions of flagellin and the eukaryotic cell surface[20] . Also located on this plasmid is the eatA gene that encodes the EatA serine protease autotransporter molecule[21] capable of degrading EtpA[22] as well as MUC2[23] , the major gel-forming soluble mucin in the small intestine[24] . Recent immunoproteomic[25] and transcriptomic [26] analyses of H10407 have also highlighted two chromosomally encoded antigens that are not specific to the ETEC pathovar , but which nonetheless appear to be involved in the pathogenesis of these organisms . Conceivably , these molecules , YghJ[27] , a secreted mucin-degrading metalloprotease , and EaeH [28] , an adhesin , act in concert with colonization factors and other pathovar-specific virulence proteins like EatA and EtpA to promote toxin delivery . While emerging data suggests that these novel proteins are highly immunogenic[25] and that EtpA and EatA are protective antigens[29–31] in a murine model of ETEC infection , additional data regarding their conservation among ETEC strains are needed to determine their suitability as vaccine targets . Here we demonstrate that these antigens are broadly represented in a diverse collection of ETEC isolates suggesting that they could be employed to augment existing approaches to ETEC vaccine development .
ETEC strains used in this study are detailed in S1 Dataset . All strains were grown at 37° in Cassamino acids yeast extract media[32] ( CAYE: 2 . 0% Casamino Acids , 0 . 15% yeast extract , 0 . 25% NaCl , 0 . 871% K2HPO4 , 0 . 25% glucose , and 0 . 1% ( v/v ) trace salts solution consisting of 5% MgSO4 , 0 . 5% MnCl2 , 0 . 5% FeCl3 ) from frozen glycerol stocks maintained at −80°C . Strains from the International Centre for Diarrhoeal Disease Research ( icddr , b ) in Dhaka were selected based on their associated disease severity using modified WHO guidelines as previously outlined[33] . Expression of individual CFs was determined by dot immunoblotting with monoclonal antibodies specific to each respective CFs ( CF-MAb ) as previously described [34] . Briefly , 2 μl of a PBS suspension containing ∼106 colony forming units of each ETEC strain was dotted onto nitrocellulose , air-dried , blocked with BSA in PBS , followed by detection with CF-MAbs and goat anti-mouse IgG_HRP conjugate . Bound MAbs were then detected with 4-chloro-1-naphthol chromogen and H2O2 . We screened a total of 181 ETEC available isolates currently maintained as frozen glycerol stocks in our laboratories . The majority of these strains were collected between 1998 and 2011 in Bangladesh , and were obtained from the icddr , b in Dhaka . Complementing this collection were geographically disparate strains associated with severe diarrheal illness including strains from the Amazon region in Brazil [35] , and ThroopD , an isolate from a patient with severe ETEC diarrheal illness who presented in Dallas in the 1970s[36] . Strains encoding eatA and etpA were identified by PCR using primers directed against conserved regions of these genes as previous described [37] . Briefly , a small amount of frozen glycerol stock from each strain was introduced with a sterile pipette tip into a PCR mixture containing the respective primers and a master mix . Toxin genotypes were confirmed in these isolates using multiplex PCR screening for genes encoding heat-labile ( LT ) , and heat-stable toxins ( STp , and STh ) as previously described[34] . Primer sequences are listed in S1 Table . To determine production of secreted virulence antigens by different ETEC strains , supernatants from overnight cultures were first precipitated with trichloroacetic acid ( TCA ) [19] and resuspended in sample buffer before polyacrylamide gel electrophoresis . Western blotting was then performed using polyclonal rabbit antisera against recombinant versions of either EatA[21] , EtpA[19] , or YghJ[27] that were pre-absorbed against an E . coli lysate column ( Pierce ) and affinity-purified using the antigen immobilized on nitrocellulose membranes as previously described [31 , 38] , followed by detection with affinity-purified secondary goat anti-rabbit- ( IgG ) -HRP conjugate ( Santa Cruz Biotechnology , SC2004 ) . To examine antigenic conservation of EatA among ETEC isolates for which genomic DNA sequences are currently available , BLASTP[39] was used to search GenBank https://www . ncbi . nlm . nih . gov/genbank/ using the full length sequence of the EatA protein from strain H10407 ( https://www . ncbi . nlm . nih . gov/protein/AAO17297 . 1 ) as the query sequence . To construct alignments of EatA from positive strains , the 1042 residue passenger domain ( corresponding to amino acids 57–1098 of EatA from H10407 ) was compared with EatA of ETEC isolates derived from different phylogenic lineages using a CLUSTAL Omega ( release 1 . 2 . 0 AndreaGiacomo ) [40] algorithm plugin for CLC Main Workbench v6 . 9 . 1 . A similar approach was used to compare the amino-terminal sequence of EtpA ( amino acids 1–600 , GenBank accession number AAX13509 . 2 ) . Virulence protein expression data from the collection of 181 strains under study were included in the analysis . Heat maps were configured using R[41] version 3 . 1 . 0 ( 2014 , http://www . R-project . org/ ) using gplots[42] and RColorBrewer[43] packages installed from http://CRAN . R-project . org using the heatmap2 function within gplots ( see S2 Dataset ) . The antigens used in these studies were produced as polyhistidine-tagged recombinant proteins and purified by immobilized metal ion affinity chromatography ( IMAC ) as previously described[27 , 29 , 44 , 45] . Additional polishing steps including size exclusion or ion exchange chromatography were performed as needed to produce highly purified antigens . Purity of each antigen was assessed by SDS-PAGE followed by sensitive Coomassie Blue staining . Purified recombinant antigens were stored at −80°C . To quantitfy antibody concentrations directed at novel recombinant antigens , kinetic ELISA was performed on dilutions of plasma samples previously obtained from patients hospitalized at the International Centre for Diarrhoeal Disease Research in Dhaka , Bangladesh ( icddr , b ) with acute symptomatic ETEC infections . Plasma samples from non-infected adults and children obtained at icddr , b , or specimens obtained from children at Saint Louis Children’s Hospital were used as negative controls . Samples from human volunteer ETEC H10407 challenge studies were kindly provided by Dr . Robert Gormely and Dr . Stephen Savarino of National Naval Medical Center , Bethesda Maryland . Use of these clinical materials was approved by the Institutional Review Boards of both icddr , b and Washington University School of Medicine . All plasma samples were maintained at 4°C in a humidified chamber prior to use in ELISA . Immune responses to purified recombinant proteins ( rYghJ , rEaeH , rEtpA , rEatAp ) were assessed by kinetic ELISA[46] as previously described [30 , 47] . Antigen binding to ELISA wells ( Corning , Costar 2580 ) was first optimized to determine the optimal coating concentration and buffer system , using highly antigen-specific polyclonal rabbit antisera to detect binding by ELISA . Purified antigens were then diluted either in 50 mM carbonate buffer ( pH 9 . 6 ) ( rEtpA-myc-His6 , 1 μg/ml; rEatAp , 10 μg/ml; rYghJ-myc-His6 , 1 μg/ml ) ; or in phosphate buffered saline ( PBS , pH 7 . 4 ) ( rEaeH-myc-His6 , 1 μg/ml ) . ELISA plate wells were coated with 100 μl/well overnight at 4°C , washed with PBS containing 0 . 05% Tween-20 ( PBS-T ) , and blocked for 1 h at 37°C with 1% BSA in PBS-T . All plasma samples were diluted at 1:4096 in blocking buffer . After incubation for 1 hour at 37°C , plates were washed with PBS-T , and secondary goat anti-human IgG ( H+L ) -HRP conjugated antibody ( Pierce , 31410 ) was added at a final concentration of 1:10 , 000 . After incubation for 30 minutes at 37°C , plates were washed and developed with TMB microwell peroxidase substrate [3 , 3’ , 5 , 5’-Tetramethylbenzidine] ( KPL , 50-76-00 ) . Kinetic absorbance measurements were determined at a wavelength of 650 nm , and acquired at 40 s intervals for 20 minutes using a microplate spectrophotometer ( Eon , BioTek ) . All data were recorded and analyzed using Gen5 software ( BioTek ) and reported as the Vmax expressed as milliunits/min . Statistical calculations were performed using Prism v4 . 0c ( GraphPad Software ) , using nonparametric Mann-Whitney ( two-tailed ) comparisons of data . These studies were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health , using an established protocol approved by the Washington University School of Medicine Animal Studies Committee . Four groups of twelve CD-1 mice were immunized intranasally with either 1 μg of LT ( adjuvant only controls ) , or 1 μg of LT + 15 μg of rEatAp ( H134R ) , or 1 μg of LT + 15 μg of rEtpA , or 1 μg of LT + 15 μg of rEatA ( H134R ) +15 μg of rEtpA on days 0 , 14 , 28 . On day 40 , mice were treated with streptomycin [5 g per liter] in drinking water for 24 hours , followed by drinking water alone for 18 hours . After administration of famotidine to reduce gastric acidity , mice were challenged with 106 cfu of the kanamycin-resistant ( lacZYA::KmR ) strain jf876[48] by oral gavage as previously described[47] . Fecal samples ( 6 pellets/mouse ) were collected on day 42 before oral gavage , re-suspended in buffer ( 10mM Tris , 100mM NaCl , 0 . 05% Tween 20 , 5mM Sodium Azide , pH 7 . 4 ) overnight at 4°C , centrifuged to pellet insoluble material , and recover supernatant for fecal antibody testing ( below ) . Twenty-four hours after infection , mice were sacrificed , sera were collected , and dilutions of saponin small-intestinal lysates were plated onto Luria agar plates containing kanamycin ( 50 μg/ml ) . Murine immune responses to LT , EatA and EtpA were determined using previously described kinetic ELISA . Briefly , ELISA wells were coated with 1 μg/ml GM1 , or 10 μg/ml of rEatAp ( H134R ) , or 1 μg/ml rEtpA in carbonate buffer ( 15 mM Na2CO3 , 35 mM NaHCO3 , 0 . 2 g/L NaN3 , pH8 . 6 ) overnight at 4°C . Wells were washed three times with phosphate-buffered saline containing 0 . 05% Tween 20 ( PBS-T ) , blocked with 1% bovine serum albumin ( BSA ) in PBS-T for 1 h at 37°C , and 100 μl of fecal suspensions ( undiluted ) or sera ( diluted 1:100 in PBS-T with 1% BSA ) was added per ELISA well and incubated at 37°C for 1 h . Horseradish peroxidase-conjugated secondary antibodies were used and signal detected with TMB ( 3 , 3′ , 5 , 5′-tetramethylbenzidine ) -peroxidase substrate ( KPL ) substrate . All animal studies were performed under protocols approved by the Animal Studies Committee of Washington University School of Medicine ( protocol number 20110246A1 ) . All procedures complied with Public Health Service guidelines , and The Guide for the Care and Use of Laboratory Animals . All human studies included were performed under a protocol approved by the Institutional Review Board of Washington University School of Medicine ( IRB ID# 201110126 ) . All of the human studies here report anonymous analysis of de-identified pre-existing sera previously stored from earlier studies for which no additional consent was obtained .
Two novel antigens , the EtpA adhesin , and the passenger domain of the EatA serine protease are encoded on the large 92 kilobase virulence plasmid of the prototypical ETEC strain H10407 . Both of these secreted proteins[22 , 30] are required for H10407 to efficiently deliver heat-labile toxin to target epithelial cells . Furthermore , both of these antigens are immunogenic [25] , and induce protective immune responses in a murine model of ETEC intestinal colonization[29 , 31] . To further assess their utility as potential vaccine antigens , we examined a large collection of ETEC strains that were well characterized with respect to associated clinical metadata pertaining to disease severity and which had not undergone repeated serial passage in the laboratory . Altogether , we found that these antigens are relatively conserved in the ETEC pathovar , confirming the results of earlier studies that focused on strains from different phylogenies obtained in Guinea Bissau and Chile [37 , 49] . Of the 181 strains examined in the present study ( Fig . 1 ) , we found that more than half of all strains produced EtpA ( 102/181 , 56% ) and/or EatA ( 106/181 , 59% ) ( S1 Dataset ) , and that more than three quarters of all strains produced at least one of these antigens . Both EtpA and EatA were identified more than twice as frequently as the most commonly identified CF ( CS6 ) , which was identified in 22% of strains in this collection ( Table 1 ) . Importantly , although the genes encoding the etpBAC secretion system[19] and the EatA autotransporter[21] were initially discovered on the same large virulence plasmid of H10407 , which also encodes the colonization factor ( CF ) CFA/I , we found that these loci were not restricted to strains expressing this particular CF , but were widely distributed among the different CFs , and were also present in strains for which no CF could be identified ( Fig . 2A ) . Indeed , half of the strains for which no CF could be identified expressed either EtpA or EatA , suggesting that these antigens could complement existing vaccination strategies centered on CFs . As expected by the association with multiple CFs , we also found that EtpA and EatA were secreted by strains from multiple phylogenic lineages ( Fig . 2B , C ) . Interestingly , however we found a negative association between the etpBAC locus and strains expressing CFA/IV antigens [50 , 51] including CS5 in that none of the 23 strains possessing CS5 fimbriae secreted the EtpA adhesin . Similarly , among strains expressing CS6 , which is frequently co-expressed with CS5 , only a minority secreted EtpA . These data are also consistent with our earlier observation that the prototype B7A strain , which also expresses CS6 , lacks the etpBAC locus and does not secrete EtpA[19] . Interestingly , both the eatA and etpBAC loci were originally identified in ETEC strain H10407 , originally isolated from an adult with severe , cholera-like illness in Bangladesh[52] . As has been noted previously , this strain also causes more severe illness in human clinical challenge studies relative to other strains like B7A that lack these loci[53] . Because we had clinical metadata pertaining to disease severity for all of the strains in our collection , we questioned whether the production of either of these antigens was associated with strains isolated from more severe forms of infection . However , we did not find any clear association between either of these putative virulence loci and clinical outcome ( S1 Dataset ) . We also examined the conservation of two chromosomally-encoded antigens which are not specific to the ETEC pathovar , but have recently been shown to play a role in virulence . The eaeH gene was originally identified on the chromosome of ETEC strain H10407 by subtractive hybridization with E . coli MG1655[54] , is transcriptionally activated by cell contact [26] , and under these conditions EaeH is produced by a diverse group of strains belonging to different phylogenies[28] . Using the EaeH peptide sequence from H10407 ( GenBank accession AAZ57201 ) , BLASTP searches of recently sequenced ETEC strains from Bangladesh and elsewhere ( http://gscid . igs . umaryland . edu/wp . php ? wp=comparative_genome_analysis_of_enterotoxigenic_e . _coli_isolates_from_infections_of_different_clinical_severity ) also revealed that the eaeH gene was present in 63 out of 91 distinct isolates ( 69% ) ( S1 Dataset ) . BLASTP searches of these data for another chromosomally encoded molecule , YghJ , a type II secretion system effector[55] recently shown to be involved in mucin degradation and toxin delivery[27] demonstrated that the yghJ gene was present on the chromosomes in 83 of 91 ( 91% ) isolates . Similarly , we identified the YghJ protein in a majority ( 161/181 , 89% ) of ETEC culture supernatants ( S1 Dataset ) . This antigen was produced across ETEC strains expressing multiple CF types including 31/36 strains that were CF-negative by monoclonal antibody screening . Ideally , putative vaccine targets should be specific to the pathovar under study or restricted to pathogenic isolates , but not subject to significant antigenic variation . Therefore to further examine the potential utility of two ETEC pathovar specific antigens , EtpA and EatA , as vaccine candidates , we used recently obtained DNA sequence information from multiple ETEC genomes belonging to different phylogenies and from temporally and geographically disparate sources to compare the predicted amino acid sequences of these proteins . For the prototype EatA molecule , first described in ETEC H10407[21] , the 1042 residue region from amino acids 57–1098 is predicted for the secreted passenger domain that contains the serine protease catalytic triad[21] as well as protective epitopes[23] . We therefore compared this region of the molecule to those derived from the recently released genome sequences of multiple ETEC strains . Altogether , we found that the sequence of the EatA passenger domain ( EatAp ) was very highly conserved across strains , and exhibited between 95–100% identity to the prototype H10407 Eatp ( Table 2 ) . Likewise , the predicted serine protease catalytic motif formed by the histidine , aspartic acid and serine residues at positions 134 , 162 , and 267 , respectively were universally conserved within the passenger domains of these proteins ( S1 Fig . ) . Similarly , the predicted amino acid sequences of the secreted EtpA adhesin molecules from multiple strains exhibited between 94 and 100% identity to the H10407 prototype antigen ( Table 3 , S2 Fig . ) . Despite the fact that the comparator strains included here spanned isolates collected over nearly 40 years , belonging to different phylogenies and that strains originated in diverse locations in Asia , Africa and the Americas , both proteins appear to exhibit remarkably little antigenic variation . Likewise , in analysis of the genomes of strains isolated recently within Bangladesh both proteins demonstrated similar degrees of sequence conservation ( S1–S2 Fig . ) . Earlier immunoproteomic studies suggested that a variety of conserved E . coli proteins as well as ETEC pathovar specific proteins are recognized during the course of experimental infections in mice , and these responses parallel those observed using pooled convalescent sera from ETEC patients[25] . To further characterize the immune response to novel antigens we focused on four proteins that have recently been shown to play a role in ETEC pathogenesis , including two plasmid-encoded secreted ETEC pathovar-specific antigens: EatA protease , and the EtpA adhesin , as well as the highly conserved chromosomally-encoded YghJ metalloprotease and the EaeH adhesin protein . In comparing convalescent plasma from patients hospitalized at icddr , b to uninfected controls from Bangladesh , we found that patients in general exhibited significantly greater total antibody ( IgG , IgM , IgA ) responses to each of these antigens following diarrheal illness ( Fig . 3 ) suggesting that these proteins are expressed during the course of infection . Similar results were obtained in comparing plasma from un-infected children from a non-endemic area in the United States ( S3 Fig . ) . We also examined the immune response to EtpA following infection by examining sera obtained before and after challenge of human volunteers with ETEC H10407 . In sera obtained from two independent volunteer challenge studies , we also observed significant increases in immune responses to EtpA ( S4 Fig . ) , strongly suggesting that this secreted protein is specifically recognized following infection by ETEC strains including H10407 that secrete this antigen . Previous studies have demonstrated that individually , vaccination with either EtpA[31 , 56] or the passenger domain of the EatA serine protease[29] affords protection against intestinal colonization in mice . The data above suggest that collectively these antigens might significantly extend coverage presently offered by classical approaches to ETEC vaccine development . We therefore questioned whether these two antigens could be successfully combined in a subunit approach . Because we have previously demonstrated that the native secreted EatA passenger domain will degrade intestinal mucin[29] as well as the EtpA adhesin molecule[22] , we elected to vaccinate animals with a modified recombinant version of the EatA passenger that lacks protease activity ( rEatApH134R ) . Co-vaccination with rEtpA and the mutant rEatApH134R molecule elicited robust serologic responses to both molecules that were comparable to vaccination with either antigen alone . As anticipated , each of the groups mounted strong serologic responses to the LT adjuvant ( Fig . 4A ) , and both antigens retained their immunogenicity following co-immunization of EtpA with the rEatAH134R passenger domain ( Fig . 4B , C ) with responses that were at least comparable to those obtained following immunization with either antigen alone ( Fig . 4C ) . Likewise , mice immunized with both antigens were significantly protected against colonization by ETEC ( Fig . 4D ) , although co-vaccination with both antigens did not appear to be more effective than vaccination with either antigen alone . Collectively , however , these data suggest that co-immunization with these two antigens is feasible , and could be employed to expand present approaches to ETEC vaccine antigen selection .
Enterotoxigenic Escherichia coli remain one of the most common causes of infectious diarrhea worldwide , and severe disease caused by these pathogens persists as leading cause of death among young children in developing countries[1] . Despite recognition of these toxin producing E . coli as a cause of severe cholera-like diarrheal illness more than forty years ago[57] , there remains no effective broadly protective vaccine for ETEC . Most vaccinology efforts to date have focused almost exclusively on a subset of plasmid-encoded antigens , namely the colonization factors ( CFs ) and heat-labile toxin[9] . Vaccines based on this strategy have faced several impediments . First , the CFs are quite diverse with more than 26 distinct antigens described to date . In addition , a number of recent vaccine studies have suggested that simply engendering immune responses to CFs and/or heat-labile toxin may not be sufficient to provide sustained broad-based protection[14–16] . Recent studies of ETEC pathogenesis suggest that a number of virulence factors in addition to the CFs are involved in efficient delivery of toxins to their cognate receptors on the epithelial surface[2 , 22 , 29 , 30 , 48 , 58] . Similarly , the immune response to ETEC infection appears to involve many proteins[25 , 47] in addition to the classical antigens that are the present focus of most vaccines . Collectively , these findings suggest that there may be additional molecules that could be targeted to interdict toxin delivery by these pathogens , expand the list of potential protective antigens , and complement existing approaches to vaccine development for ETEC[59 , 60] . A major challenge to ETEC vaccine development in general is that the most highly conserved antigens of ETEC , typically encoded on core regions of the chromosome , are also shared with commensal E . coli[60] . Included among these chromosomally encoded conserved proteins are two antigens studied here , YghJ[27] and EaeH[28] that were recently shown to be important for ETEC virulence . While the present studies also demonstrate that these proteins are recognized during the course of ETEC infection , the degree to which these antigens can be safely targeted in vaccines without inadvertent disruption of the intestinal microflora remains to be studied . The inherent plasticity of E . coli genomes contributes substantially to the difficulty in defining antigens unique to the ETEC pathovar that are widely conserved . No single antigen exclusive to these pathogens , but universally conserved in this pathovar , has been described to date . Some have suggested that this might be predicted based on the fact that the plasmid-encoded heat-labile and/or heat-stable toxins , which define the ETEC pathovar , could form a minimal complement of virulence genes in wide variety of E . coli host strains[61] . Nevertheless , earlier studies conducted on phlyogenicaly disparate strains from Guinea Bissau[49] and Chile[37] suggested that genes encoding two pathogen-specific antigens EatA and EtpA were present in a majority of strains . In this context , we examined the gene conservation and the actual production of these proteins in a large collection of well-characterized strains from Bangladesh , complemented by strains from other locations that were associated with severe disease and for which there were available clinical metadata . Notably , two plasmid-encoded ETEC pathotype-specific antigens , the EatA serine protease and the secreted EtpA adhesin molecule were shared broadly among strains belonging to different CF groups with the exception of strains that produced CFA/IV antigens CS4 , CS5 , CS6 which only infrequently produced EtpA . The studies reported here represent the largest screen for EtpA and EatA secretion in ETEC performed to date . Earlier studies reporting that genes encoding both proteins were highly conserved relied on either PCR[37] or screening of draft genomes[49] for the presence of the corresponding loci . In general , we found high degrees of concordance between the presence of these genes by PCR and production of the corresponding protein . We should point out however that draft genome assemblies typically fail to encompass the entire etpA gene as automated assembly algorithms cannot faithfully incorporate the multiple repeat regions comprising two thirds of etpA . This could impact interpretation of gene prevalence in ongoing large-scale ETEC genome sequencing projects . The prevalence of EtpA and EatA ( 56 and 59% , respectively ) as determined by examination of protein expression in our study was slightly lower than previously reported in earlier studies that analyzed strains from Guinea Bissau , where both genes were present in 75% of strains [49]; or Chile , where etpA and eatA , were present in 71 and 75% of strains , respectively[37] . Nevertheless similar to these earlier studies , the strains that produced these antigens belonged to many different phylogenies suggesting that genes encoding these antigens have been widely dispersed . The analyses of strains in these studies largely focused on isolates from Bangladesh . However , these data are potentially relevant for vaccine development for a number of reasons . First , Bangladesh is highly endemic for enterotoxigenic E . coli infections , and consequently remains an important site for vaccine field trials . In addition , ETEC has been under study in this region since the discovery of this pathotype , permitting us to compare sequence variation in candidate antigens over four decades . Understanding both current prevalence and sequence conservation of potential novel vaccine antigens in this population over time will be particularly important for making rational decisions about their inclusion in future iterations of ETEC vaccines . Finally , the geographic and temporal dispersal of genes encoding EtpA and EatA in multiple phylogenic backgrounds , further attests to importance of studying these molecules as potential vaccine targets as previously suggested by others[37 , 49] . The optimal formulation of an ETEC vaccine has yet to be defined , and many questions pertaining to the nature of protective immunity that develops following infections with these pathogens remain . Nevertheless , the data presented here do suggest that the novel pathovar-specific antigens could complement existing strategies for ETEC vaccine development by broadening the antigenic valency . Whether the expanded coverage afforded by inclusion of additional pathotype specific antigens would enhance vaccine efficacy beyond that presently achieved by targeting CFs and LT will need to be determined empirically .
|
Infectious diarrhea is one of the leading causes of death among young children in developing countries , and a major cause of morbidity in all age groups . The enterotoxigenic Escherichia coli contribute substantially to this burden of diarrheal illness , and have been a focus of vaccine development efforts for more than forty years following their discovery as a cause of severe diarrheal illness . The heat-labile , and/or heat stable enterotoxins that define ETEC are produced by a diverse population of Escherichia coli . This inherent genetic plasticity of E . coli has made it difficult to identify antigens specific to ETEC that are highly conserved . Therefore , identification of protective antigens shared by many ETEC strains will likely play an essential role in development of the next iteration of vaccines .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[] |
2015
|
Conservation and Immunogenicity of Novel Antigens in Diverse Isolates of Enterotoxigenic Escherichia coli
|
With the recent success of genome-wide association studies ( GWAS ) , a wealth of association data has been accomplished for more than 200 complex diseases/traits , proposing a strong demand for data integration and interpretation . A combinatory analysis of multiple GWAS datasets , or an integrative analysis of GWAS data and other high-throughput data , has been particularly promising . In this study , we proposed an integrative analysis framework of multiple GWAS datasets by overlaying association signals onto the protein-protein interaction network , and demonstrated it using schizophrenia datasets . Building on a dense module search algorithm , we first searched for significantly enriched subnetworks for schizophrenia in each single GWAS dataset and then implemented a discovery-evaluation strategy to identify module genes with consistent association signals . We validated the module genes in an independent dataset , and also examined them through meta-analysis of the related SNPs using multiple GWAS datasets . As a result , we identified 205 module genes with a joint effect significantly associated with schizophrenia; these module genes included a number of well-studied candidate genes such as DISC1 , GNA12 , GNA13 , GNAI1 , GPR17 , and GRIN2B . Further functional analysis suggested these genes are involved in neuronal related processes . Additionally , meta-analysis found that 18 SNPs in 9 module genes had Pmeta<1×10−4 , including the gene HLA-DQA1 located in the MHC region on chromosome 6 , which was reported in previous studies using the largest cohort of schizophrenia patients to date . These results demonstrated our bi-directional network-based strategy is efficient for identifying disease-associated genes with modest signals in GWAS datasets . This approach can be applied to any other complex diseases/traits where multiple GWAS datasets are available .
Genome-wide association ( GWA ) studies have , during the past decade , become a powerful tool to study the genetic components of complex diseases [1] . Although an increasing number of genes/markers have been uncovered in GWA studies , which have provided us important insights into the underlying genetic basis of complex diseases such as schizophrenia [2] , [3] , [4] , it has also become evident that many genes are weakly or moderately associated with the diseases . Most of these variants have been missed in single marker analysis , as investigators typically employ a genome-wide significance cutoff P-value of 5×10−8 . Alternatively , the gene set analysis ( GSA ) of GWAS datasets provides ways to simultaneously examine groups of functionally related genes for their combined effects and thus have improved power and interpretability [5] . Many GSA methods have been reported to date , such as the gene set enrichment analysis [6] , the adaptive rank-truncated product [7] , the gene set ridge regression in association studies ( GRASS ) [8] , etc . Most of these methods were designed to use pre-defined gene sets such as the KEGG database [9] or the Gene Ontology ( GO ) annotations [10] . Alternatively , studies are emerging by incorporating protein-protein interaction ( PPI ) networks into GWAS analysis . Baranzini et al . [11] first adopted a network-based method that was initially designed for gene expression data [12] to analyze the GWAS data for multiple sclerosis . Recently , Rossin et al . [13] developed the Disease Association Protein-Protein Link Evaluator ( DAPPLE ) ; it tests whether genes that are located at association loci in a GWAS dataset are significantly connected via PPIs . We have also developed the dense module search ( DMS ) method [14] , which overlays the gene-wise P values from GWAS onto the PPI network and dynamically searches for subnetworks that are significantly enriched with the association signals . The advantages of network-based analysis of GWAS data in comparison with the standard GSA methods lie in many aspects . First , most GSA methods test on pre-defined gene sets , which heavily rely on a priori knowledge and are incomplete . For example , the popular KEGG database has pathway annotations covering only ∼5 , 000–5 , 500 genes [15] , accounting for less than 30% of the genes in GWAS datasets . In contrast , the annotations of PPI data cover a much larger proportion of human proteins . For example , a recent integrative analysis of PPI data from multiple sources has reconstructed the human PPI network by recruiting ∼12 , 000 proteins and ∼60 , 000 protein interaction pairs with experimental evidence [16] . There are other assembled PPI datasets that include both experimentally supported and computationally predicted interactions; thus , they could annotate even more proteins and interactions [17] , [18] . Second , the standard GSA methods are typically based on canonical definitions of pathways or functional categories , but the association signals from GWAS might converge on only part of the pathway [19] . In such cases , analysis of the whole pathway as a unit would reduce the power . On the other hand , network-assisted methods allow for the definition of de novo gene sets by dynamically searching for interconnected subnetworks in the whole interactome and , thus , can effectively alleviate the limitation of the fixed size in pathway analysis . Despite these advantages , there are challenges in the application of network-based approaches to GWAS data . For example , the methods for defining or searching subnetworks vary greatly . While it is impractical to examine all possible subnetworks due to the intensive computing burden , different methods or algorithms may identify substantially different subnetworks [20] , making it difficult to decide in real application . Additionally , network-based analysis could be confounded by nodes with high degrees ( i . e . , the number of interactors of each node in the network ) , although these nodes constitute only a small proportion according to the framework of power-law distribution [21] . One example is TP53 , which interacts with several hundreds of other proteins in the whole PPI network . The existence of such hub nodes with strong interaction in the network may help them more likely to be selected in searching subnetworks and , thus , overwhelm the resultant subnetworks . Appropriate adjustments are needed . In this study , we aim to search for modules that are significantly enriched with association signals in human PPI network weighted by GWAS signals . We take advantage of our recently developed dense module search ( DMS ) algorithm [14] to conduct module searching and construction . Based on this , we introduced statistical evaluations of the modules identified by DMS , including a significance test based on module scores , a weighted resampling method to adjust potential biased in GWAS data ( e . g . , caused by gene length or SNP density ) , a topologically matched randomization process to adjust the bias in network ( e . g . , the high degree nodes ) , and a permutation test to determine the disease association of the modules . In addition , we propose a bi-directional framework to search for consistent association signals from multiple GWAS datasets available for one specific disease or trait . Specifically , two GWAS datasets were analyzed in parallel: one is assigned as a discovery dataset and another as an evaluation dataset , and vice versa . This strategy provides robust results with partial validation — only the modules that were consistently highly scored would be selected for further validation and functional assessment . We demonstrated it in schizophrenia using two major GWAS datasets for module identification , and incorporated a third dataset to independently replicate the results . Finally , we performed a meta-analysis of the markers that were mapped in the module genes . We identified 18 SNPs in 9 module genes that are of particular interests ( Pmeta<1×10−4 ) .
We incorporated two case-control GWAS datasets for schizophrenia in this study for module search: the International Schizophrenia Consortium ( ISC ) study and the Genetic Association Information Network ( GAIN ) dataset . A third dataset , the Molecular Genetics of Schizophrenia ( MGS ) - nonGAIN dataset , was included in the validation stage by bringing independent samples for disease association test . Each of the three datasets was preprocessed and quality controlled , with none observed to have substantial population stratification . As shown in Figure 1 , we started with the GAIN dataset for module discovery , followed by module evaluation using the ISC dataset . In the parallel thread , the ISC dataset was used for constructing modules and the GAIN dataset for evaluation . In both threads , a series of significance tests were performed , each of which aims to build null distributions for different purposes and adjust specific biases . The modules that passed the filtering criteria in both datasets were selected and merged . Module genes were collected and considered as schizophrenia candidate genes , whose association signals were further examined in an independent GWAS dataset , the nonGAIN dataset , as well as , by meta-analysis using three GWAS datasets ( ISC , GAIN , and nonGAIN ) . More specifically , our algorithm for multiple GWAS datasets includes the following steps . Step 1 . Candidate module search in one GWAS dataset . The gene-wise P values from the GWAS results were converted to z-scores and overlaid to the background human interactome ( the whole PPI network ) , with each node being weighted by the z-score of the encoding genes . For each node in the network , DMS is performed to generate a best module , i . e . , with the largest module score , Zm ( see Materials and Methods ) . We performed this module construction step for each GWAS dataset using the R package , dmGWAS , which implements the original DMS algorithm [14] , and the default parameters were used . Step 2 . Module assessment . We provide three types of significance tests to assess the candidate modules: ( 1 ) the significance test based on module scores ( P ( Zm ) ) ; ( 2 ) the evaluation of module scores in the context of various biases ( PGL , PnSNPs , and Ptopo ) ; and ( 3 ) the permutation test by shuffling disease labels in the GWAS datasets ( Pemp ) . Detailed information can be found in the Materials and Methods section . Step 3 . Module selection . In practice , several thousands of modules are likely to be constructed , corresponding to the thousands of genes used as seed; thus , further selection for top modules is needed . In a single GWAS-weighted module search process , we employed the following combinatorial criteria to select modules: ( 1 ) P ( Zm ) <0 . 05; ( 2 ) PGL<0 . 05 , PnSNPs<0 . 05 , and Ptopo<0 . 05; and ( 3 ) Pemp<0 . 05 . When there are two GWAS datasets available for the same disease or trait , we propose to use one dataset serving as discovery ( discov ) and the other as evaluation ( eval ) , and vice versa ( Figure 1 ) . This allows us to select the most reliable modules with enriched association signals from more than one study . For each module generated by the discovery dataset , we also computed the corresponding P ( Zm ( eval ) ) using the same set of genes ( i . e . , in the same module ) with gene weights based on the evaluation GWAS dataset , as well as Pemp ( eval ) by shuffling the case/control labels in the evaluation GWAS dataset . Modules were selected if they have P ( Zm ( eval ) ) <0 . 05 and Pemp ( eval ) <0 . 05 . Using GAIN as the discovery dataset , we identified a total of 8 , 739 modules ( Figure 2A ) . The module size ranged between 5 and 17 , with a median value of 11 ( Figure S2 ) . A total of 935 modules passed the combinatorial criteria , i . e . , ( 1 ) P ( Zm ) <0 . 05; ( 2 ) PGL<0 . 05 , PnSNPs<0 . 05 , and Ptopo<0 . 05; and ( 3 ) Pemp<0 . 05 . Among them , 71 modules were also significant in the ISC evaluation dataset ( P ( Zm ( eval ) ) <0 . 05 ) . Furthermore , 68 out these 71 modules passed the permutation test in the evaluation dataset ( Pemp ( eval ) <0 . 05 ) . They were denoted as the final list of modules . Similarly , using ISC as the discovery dataset , we identified 8 , 899 modules ( Figure 2B ) , with the module size ranging between 5 and 18 and a median value of 11 ( Figure S2 ) . A total of 259 modules passed the combinatorial criteria . However , only one module was significant when adding the GAIN dataset for evaluation , involving 11 genes . We then merged the two lists to build a PPI subnetwork , which consisted of 205 module genes ( Figure S3 ) . A substantial proportion of the 205 module genes had nominally significant P values ( defined as P<0 . 05 without multiple testing correction ) in the corresponding GWAS dataset: 139 module genes ( 67 . 80% ) had PGAIN<0 . 05 , and 125 module genes ( 60 . 98% ) had PISC<0 . 05 . The remaining module genes with non-significant P values were recruited in the top modules due to their physical interactions with the nominally significant genes in the PPI network , as DMS aims to identify joint effects of a set of schizophrenia genes in the context of the PPI network . In summary , 97 of the 205 genes ( 47 . 32% ) were nominally significant in both the GAIN and ISC datasets , and 167 ( 81 . 46% ) were nominally significant in either dataset . Further comparison of these genes with previous association studies in the SZGene database [22] ( as of January 26 , 2011 ) showed that 31 ( 15 . 12% ) of the module genes had been studied for association with schizophrenia . The SZGene database manually curates the association results from previously published association studies as well as recent GWAS findings . Among these 31 genes , 16 had at least one positive association study in previous work . Eighteen of these 31 genes ( 58 . 06% ) were nominally significant ( gene-wise P value<0 . 05 ) in both the GAIN and ISC datasets , while 26 ( 83 . 87% ) had nominal significance in either dataset . These proportions were similar to those evaluated for the whole 205 module genes above . In contrast , the corresponding proportions of nominally significant genes in whole GWAS datasets were much lower ( 16 . 43% genes with nominal significance in both datasets and 55 . 77% in at least one dataset ) , indicating that the identified module genes were closer to genes known to be associated with schizophrenia . We further evaluated the 205 module genes in an independent GWAS dataset , the nonGAIN dataset . First , we assessed whether the module genes contain a proportion of significant genes than randomly expected . This was done through weighted resampling while controlling the potential biases of gene length and SNP density in the nonGAIN dataset . Representing each module gene by the smallest P value among the SNPs located in its gene region , we denote the gene as significant if its nominal P value was less than 0 . 05 . The 205 module genes were pooled together and denoted as one gene set , in which we found 76 genes were observed to be nominally significant in the nonGAIN dataset . We executed the weighted resampling process by 10 , 000 times , and built a null distribution of the number of significant genes given the number of module genes . This process was executed in the same way as the second significance test in module assessment . The details can be found in the Materials and Methods section , as well as in previous study [23] . The empirical P for the module gene set was 0 . 002 when adjusting gene length , and 0 . 003 when adjusting SNP density , indicating that these genes are not expected from random cases . Second , we assessed the module genes in nonGAIN through resampling of SNPs . The 205 module genes had a total of 15 , 548 SNPs in the nonGAIN dataset . In each resample , we randomly selected the same number of SNPs ( i . e . , 15 , 548 SNPs ) out of all the SNPs genotyped in the nonGAIN dataset , and recorded the number of significant SNPs , which were again defined as those whose nominal P values<0 . 05 . We repeated this process by 10 , 000 times and counted the number of resample processes having more significant SNPs than that of the real case . This analysis resulted in an empirical P value of 0 . 022 , indicating that the SNPs harbored in these module genes contained a higher proportion of nominally significant SNPs than randomness . Note that the nonGAIN dataset is independent of the GAIN and ISC datasets we used to discover the module genes . Therefore , these results provide an independent replication of our module genes and showed that they are significantly enriched with association signals to schizophrenia . There were 15 , 252 SNPs in the genomic regions of the 205 module genes that were genotyped in all three GWAS datasets . Using the inverse-variance weighted meta-analysis method and heterogeneity test , we identified a total of 1032 SNPs having nominal significance ( Pmeta<0 . 05 ) after removing substantial heterogeneity ( Pheterogeneity<0 . 05 ) . To determine whether the module genes contain a proportion of significant SNPs higher than expected by chance , we randomly sampled SNP sets with the same number of SNPs mapped to module genes ( i . e . , 15 , 252 ) and computed the proportion of significant SNPs ( defined as those with Pmeta<0 . 05 ) . Repeating the random process by 1000 times , we computed the empirical P value by Pemp = , where K is the number of significant SNPs with Pmeta<0 . 05 in a random set , and k is the number in the real case , i . e . , k = 1032 . This random process showed that the module genes contains a significantly higher proportion of significant SNPs ( Pemp<0 . 001 ) , further proving the enriched signal in the module genes . Among the significant module SNPs by meta-analysis , 18 SNPs in 9 genes were shown to have Pmeta<1×10−4 ( Table 1 ) . The most significant module SNPs were located in the gene HLA-DQA1 , followed by MAD1L1 ( Table 1 , Figure 3 ) . There are two SNPs in HLA-DQA1 with Pmeta<1×10−4: rs9272219 ( Pmeta = 1 . 46×10−6 ) and rs9272535 ( Pmeta = 1 . 58×10−5 ) . Both were in the top list reported in a previous combined analysis of three GWAS datasets for schizophrenia [2] , [3] , [4] , which included all the GWAS datasets we used here plus the SGENE dataset [4] , to which we do not have access currently . The combined P value in the previous work [4] was Pcomb = 6 . 9×10−8 for rs9272219 and Pcomb = 8 . 9×10−8 for rs9272535 . Both SNPs are located in the MHC region chr6: 27 , 155 , 235–32 , 714 , 734 , a region that was reported to harbor a genome-wide significant association signal for schizophrenia [2] . Another gene , MAD1L1 , has six SNPs with small Pmeta values ( 4 . 30×10−6∼6 . 01×10−5 , Table 1 ) . MAD1L1 is a long gene ( ∼417 kb ) and has 70 overlapped SNPs examined in the meta-analysis . We further examined whether these 6 SNPs are located in the same LD block . Using the HapMap3 CEU data ( http://www . hapmap . org/ , release R2 ) , we found that these SNPs were located in 4 blocks , suggesting that they might represent independent association signals . Table 2 summarizes the results of pathway enrichment analysis of the 205 module genes by the Ingenuity Pathway Analysis ( IPA ) . Enrichment results of KEGG [15] pathways were shown in Table S1 . The enriched pathways included Wnt/β-catenin signaling , CREB signaling in neurons , Calcium signaling , Gα12/13 signaling , and synaptic long term depression . Overall , the results are consistent with the neuropathology and immune/inflammation hypotheses in schizophrenia [24] , [25] , suggesting that our DMS-based strategy is effective on detecting joint association signals from multiple GWAS datasets .
We proposed a novel strategy to prioritize candidate genes from multiple GWAS datasets in the context of the human interactome and applied it to schizophrenia . Integration of the PPI network and implementation of our dense module search algorithm greatly improved the coverage of gene annotations , introduced gene set flexibility when searching for candidate genes , and allowed for dynamic identification of putative genes . The bidirectional strategy we proposed here made full use of the discovery and evaluation datasets to avoid potentially incomplete discovery using either one of them separately . The final subnetwork and candidate gene list display the combined results of the two processes , namely GAIN ( discovery ) → ISC ( evaluation ) and ISC ( discovery ) → GAIN ( evaluation ) ; thus , they are comprehensive and cohesive in revealing the signals from both datasets . At the molecular level , the module genes we identified showed substantial overlap with previous studies . We also identified novel genes that had not been studied in schizophrenia , yet could be promising new candidates . The procedure we proposed in this study implemented our previously developed dense module search algorithm . One important improvement is that we introduced P ( Zm ) for module selection , instead of simply relying on the module score , Zm , although the latter is straightforward and has been proved effective in our previous work [14] . In this study , we adopted the Efron et al . [26] method and computed P values based on Zm scores through the estimation of empirical null distribution . Theoretically , Zm and the corresponding P ( Zm ) values are expected to have identical rank , which has also been observed in real data ( Spearman correlation coefficient = 1 ) . In contrast to applying a straightforward cutoff value of Zm to perform module selection , P ( Zm ) examines the overall distribution of all module scores and has the advantage to provide a statistical evaluation . Thus , we replaced Zm by P ( Zm ) for module selection in the current study . Alternatively , using simulated genotyping and phenotype data to estimate the proportions of modules that can capture the most causal variants will help module selection . In such cases , appropriate simulation data for the analyzed disease model is important for both power estimation and module selection , and will be considered in our future work . One limitation of our method is that the dense module searching process is sensitive to the background network . The algorithm of DMS examines all the neighborhood nodes within the distance of d and selects the best node in every step of module expanding . Although this is an advantage to recruit the best node ( s ) in each step , it also makes the DMS algorithm heavily rely on the searching space defined by the background interactions . Currently , our knowledge of human PPI network is far from complete . To reduce the uncertainty of network data , we required our working network only include interactions with experimental evidence while excluding interactions predicted by computational algorithms . However , because our aim is to search for a subnetwork that is significantly enriched with GWAS signals , the background PPI network can be extended to any network that is built under a rational biological hypothesis , e . g . , co-expression network , functional correlated network , or network based on co-occurrence in literature . Using any of these potential datasets , the strategy we proposed here can be easily extended while the aim is always to search for a subnetwork that is significantly enriched with association signals from GWAS data . We performed meta-analysis using three GWAS datasets , two of which have already been used for module construction . In the latter case , the ISC and GAIN datasets were used at the gene and module levels , while in the meta-analysis , the examination of the three GWAS datasets was conducted at the SNP level , including its mutation direction . The results of meta-analysis were intended to provide a complementary view and further examination of association signals of the module genes at the SNP level rather than in any single GWAS dataset . Of note , an ideal way of replication of the module genes is to test them in other datasets that are completely independent of those having already been used in the module construction step; however , there are only limited number of independent GWAS datasets for schizophrenia at the current stage . To partially accomplish this evaluation goal , we examined the module genes in the nonGAIN dataset , an independent dataset from those ( ISC and GAIN ) in module selection . The evaluation results of the nonGAIN dataset thus provide some replication evidence of the module genes . There have been a few previous studies combining network data with GWAS data . A representative method is DAPPLE , which takes the association loci in GWAS datasets as input and tests whether genes located in these loci are significantly connected via PPI . The advantages of DAPPLE include that it does not require the genotyping data of the original GWAS datasets , it provides a comprehensive randomization test to address the high-degree nodes , and it has an online tool for public use . Although DAPPLE and the method we proposed here both use PPI network to analyze GWAS data , they differ substantially in term of the underlying hypothesis . DAPPLE tests whether the associated genes are significantly connected compared to random networks while our method searches for modules that are significantly associated with the disease . Due to this main difference as the starting point , the two methods differ in many aspects in the subsequent analyses , such as the way to build the resultant network and the way to evaluate the results . For example , DAPPLE only takes the associated loci as input , which are typically defined by 5×10−8 and all the other loci , including those with weak to moderate association levels , would be discarded . This might be less efficient in searching association modules , especially for diseases or traits that do not have strong association signals from GWAS . For example , for psychiatric diseases such as schizophrenia , association signal of the markers in any single GWAS dataset failed to reach the genome-wide significance level 5×10−8 . Specifically , if we use DAPPLE to analyze any of the three GWAS datasets used in this study , we would not have any associated loci based on the significance level 5×10−8 . In contrast , DMS considers all the genes genotyped in the GWAS as input ( seeds ) in the network , and searches for the final modules in a weight-guided fashion . Here , the weight is from GWAS P values . Subsequently , many moderately associated genes ( e . g . , those with P values between 0 . 05∼5×10−8 ) might have chance to be included in the final modules for an examination of their joint effects . In practice , depending on the purposes of each study and data availability , investigators may choose appropriate methods for their specific testing . The merged subnetwork ( Figure S3 ) included a number of well-studied candidate genes for schizophrenia , such as DISC1 , DLG2 , DLG3 , DRD5 , GNA12 , GNA13 , and GNAI1 . Many genes have been studied in previous association studies [2] , [3] . Interestingly , GRB2 was present in the merged network . We identified GRB2 as a candidate gene for schizophrenia in our previous study through a network-assisted strategy [24] and then validated it in the Irish Case Control Study of Schizophrenia ( ICCSS ) sample [27] . Here , using an independent strategy and datasets , we again identified this gene , further supporting GRB2 as a candidate gene for schizophrenia . The canonical pathways enriched in the module genes also confirmed the involvement of neuro-related genes and pathways in schizophrenia . In summary , we have performed a comprehensive network-based analysis using our DMS-based approach augmented with IPA software to facilitate interpretation . The outcome of this analysis not only supports previously reported associations with schizophrenia , but also implicates functional components such as the Calcium signaling , Gα12/13 signaling , and the synaptic long term depression pathways in schizophrenia risk . Future work to estimate the power of this network-based strategy through simulation and validation in independent samples will enhance the applications of this method in other diseases or traits .
The Genetic Association Information Network ( GAIN ) dataset for schizophrenia was genotyped using the Affymetrix Genome-Wide Human SNP 6 . 0 array , and our access to it was approved by the GAIN Data Access Committee ( DAC request #4532-2 ) through the NCBI dbGaP . We used the samples of European ancestry . Quality control ( QC ) was executed as follows . For individuals , we excluded those with a high missing genotype rate ( >5% ) , extreme heterozygosity rate ( ±3 s . d . from the mean value of the distribution ) , or problematic gender assignment . We used PLINK [28] to compute the identify-by-state ( IBS ) matrix to pinpoint duplicate or cryptic relationships between individuals , and we retained the sample with the highest call rate for each pair of samples with an identity-by-descent ( IBD ) >0 . 185 . Principle component analysis ( PCA ) was performed using the smartpca program in EIGENSTRAT [29] to detect population structure and to allow removal of outlier individuals . Eight significant PCs with the Tracy Widom test P value<0 . 05 were then used as covariates for logistic regression ( additive model ) . For genotyped SNPs , we removed those with a missing genotype rate>5% , minor allele frequency ( MAF ) <0 . 05 , or departing from Hardy-Weinberg equilibrium ( P<1×10−6 ) . The final analytic dataset included 1 , 158 schizophrenia cases , 1 , 377 controls , and a total of 654 , 271 SNPs with a genomic inflation factor λ = 1 . 04 . The International Schizophrenia Consortium ( ISC ) samples were collected from eight study sites in Europe and the US [2] . The samples were genotyped using Affymetrix Genome-Wide Human SNP 5 . 0 and 6 . 0 arrays , and this data was initially analyzed by ISC [2] . A total of 3 , 322 patients with schizophrenia , 3 , 587 normal controls of European ancestry , and 739 , 995 SNPs were included in our analysis . To account for potential population structure caused by collection sites , we used the Cochran-Mantel-Haenszel test for a single marker association test , following the original report [2] . The Molecular Genetics of Schizophrenia ( MGS ) - nonGAIN dataset ( denoted as “nonGAIN” hereafter ) was genotyped in the same laboratory as GAIN , but in different phases . Access to this dataset was approved by dbGaP ( DAC request #4533-3 ) . Similar QC and PCA as described for GAIN were performed . This process retained 1 , 068 cases and 1 , 268 controls , all of which are of European ancestry , and 623 , 059 SNPs for subsequent analysis . Fifteen significant PCs with the Tracy-Widom test P value<0 . 05 were used as covariates for logistic regression ( additive model ) using PLINK , with λ = 1 . 04 . We mapped SNPs to human protein-coding genes downloaded from NCBI ftp site ( Build 36 ) . A SNP was assigned to a gene if it was located within or 20 kb upstream/downstream of the gene [30] . Each gene was assigned a gene-wise P value using the P value of the gene's most significant SNP . A total of 19 , 739 genes were successfully mapped in the GAIN dataset and 19 , 910 in the ISC dataset . A comprehensive human PPI network was downloaded from the Protein Interaction Network Analysis ( PINA ) platform [31] ( March 4 , 2010 ) , which collects and annotates data from six public PPI databases ( MINT , IntAct , DIP , BioGRID , HPRD , and MIPS/MPact ) . To ensure the reliability of the network , we only kept those interactions having experimental evidence and both interactors are human proteins . Our working network included a total of 10 , 377 nodes ( genes ) and 50 , 109 interactions . Only common genes that were represented in both GWAS and PPI datasets were retained for subsequent analysis . We applied our recently developed dense module search ( DMS ) algorithm [14] with substantial improvement to these schizophrenia GWAS datasets . Details of the DMS algorithm are provided in reference [14] . Briefly , DMS works with a node-weighted PPI network and searches for a best module for each node in a score-guided fashion . A quantitative description of the network includes each node weighted by , where is the inverse normal cumulative density function and P is the P value representing the association signal in the gene region ( which we called the gene-wise P value ) from the GWAS dataset . Each module is scored by , where k is the number of nodes ( genes ) in the module . Given a single GWAS dataset , we first overlay gene-wise P values to the PPI network to generate a GWAS P value-weighted working network . We then took each of the nodes in the network as a seed gene , and searched for a best scored module for it . In each case , starting with the seed ‘module’ formed by the seed node , the DMS algorithm searches for the node with the highest score in the neighborhood within a distance d ( d = 2 ) to the seed module . Then , the module is expanded by adding the highest-scored node if Zm+1>Zm× ( 1+r ) , where Zm+1 is the new module score after adding the node , Zm is the original module score and r is a pre-defined rate . We set r to be 0 . 1 in this study . This module expansion process iterates until none of the neighborhood nodes can satisfy the function Zm+1>Zm× ( 1+r ) . Because this module construction process was conducted taking each node in the network as the seed gene , several thousands of modules are expected corresponding to the thousands of nodes . We provided three procedures to assess the significance of the identified modules , each of which aims to build null distributions for different hypotheses . First , to perform significance test of the identified modules , we calculated P values based on module scores ( Zm ) for each module by empirically estimating the null distribution [26] . According to Efron et al . ( 2010 ) , the null distribution is a normal distribution with mean δ and standard deviation σ , both of which can be empirically estimated using the R package locfdr . Specifically , module scores were first median-centered by subtracting the median value of Zm from each of them , followed by estimation of the parameters of δ and σ for the empirical null distribution using locfdr . The standardized module scores ( ZS ) were then calculated and converted to P values , P ( Zm ) = 1-Φ ( ZS ) , where Φ is the normal cumulative density function . Second , to determine whether the module score is higher than expected by chance , a standard way is to randomly select the same number of genes in a module , i . e . , resample genes in the network regardless of the interactions , and compare the module score in the random gene set with the score in the real case . Specifically to alleviate the biases in GWAS data ( e . g . , gene length or SNP density ) or the network data ( e . g . , high-degree nodes ) , we incorporated weighted resampling which intentionally matches the pattern of biases in each resample to resemble the real case . The gene length bias and the SNP density bias are commonly noticed in GWAS datasets , especially when using the most significant SNP to represent genes [30] . This is because when mapping SNPs to genes , longer genes tend to have more SNPs and in turn have higher chance to be significant . These two types of biases are closely correlated but differ in cases due to different genotyping platforms . For both biases , we first estimate a weight for each gene based on the specific character to be adjusted , and then performed weighted resampling to ensure each of the resample has the similar pattern in term of the adjusted character . This weighted resampling procedure ensures that genes could be selected in a similar pattern of gene length or SNP density as in the real GWAS data . Therefore , the empirical P values for each module built on the bias-matched permutation data could be adjusted by gene length ( PGL ) or the number of SNPs per gene ( PnSNPs ) . A detailed description of this function can be found in previous work [23] . Another type of bias was that , in the PPI network , nodes with many interactors ( high degree ) are more likely to be recruited in module expansion steps . We thus categorized all the nodes in the working PPI network into four categories by their degree values ( degree range 0–22 , 22–24 , 24–26 , and >26 ) ( Figure S1 ) . For each module , a topologically matched random module was generated by randomly sampling the same number of nodes in each of the four node bins . An empirical P value is computed by , where is the score of the random module for the πth resample , and is the observed module score . Third , to assess the disease association of the modules , we performed permutation test by shuffling case/control labels in the GWAS datasets . We generated 1 , 000 permutation datasets using the genotyping data , and computed module scores in each permutation dataset in the same way as for the real case . An empirical P value for each module was computed according to , where Zm ( permutation ) is the module score in the permutation data . A combinatorial set of criteria was defined to select modules: ( 1 ) P ( Zm ) <0 . 05; ( 2 ) PGL<0 . 05 , PnSNPs<0 . 05 , and Ptopo<0 . 05; and ( 3 ) Pemp<0 . 05 . This set of combinatorial criteria is applied whenever one GWAS dataset is used to identify , assess and select modules . When there is an additional GWAS dataset available for evaluation , we included two additional criteria: ( 1 ) P ( Zm ( eval ) ) <0 . 05 and/or ( 2 ) Pemp ( eval ) <0 . 05 . Meta-analysis of module genes was conducted using three major GWAS datasets: ISC , GAIN , and nonGAIN . A quality control step was performed before the meta-analysis to detect whether there is duplication or cryptic relatedness among the samples in the three GWAS datasets . Pairwise IBS was computed using an unrelated list of markers ( generated through the option “–indep-pairwise 50 5 0 . 2” in PLINK [32] ) . No pair was observed with an IBD>0 . 185 , a cutoff value that is halfway between the expected IBD for third- and second-degree relatives . We performed inverse-variance weighted meta-analysis based on the fixed-effects model using the tool meta ( http://www . stats . ox . ac . uk/~jsliu/meta . html ) . This method combines study-specific beta values under the fixed-effects model using the inverse of the corresponding standard errors as weights . Between-study heterogeneity was tested based on I2 and Q statistics . SNPs with evidence of heterogeneity were removed . The three GWAS datasets were genotyped on the same platform; thus , we performed meta-analysis directly on the genotyped SNPs without imputation . Genomic control within each study was conducted in the meta-analysis using the lambda value to adjust the study-specific standard error ( SE ) . We performed pathway enrichment analysis by the IPA system ( http://www . ingenuity . com ) and also using canonical pathways from the KEGG database [9] by the hypergeometric test . The KEGG pathway annotations were downloaded in March 2011 , containing 201 pathways with size ≥10 and ≤250 . For each gene set collection , the results by the hypergeometric test were adjusted by the Bonferroni method for multiple testing correction . To further assess the significance of the identified gene sets , we performed empirical assessment of the significance by resampling 1000 times from the network genes , with each resample containing a random set of 205 genes . For a gene set S , we recorded the number of resamples in which the gene set was significant and computed an empirical P value by .
|
The recent success of genome-wide association studies ( GWAS ) has generated a wealth of genotyping data critical to studies of genetic architectures of many complex diseases . In contrast to traditional single marker analysis , an integrative analysis of multiple genes and the assessment of their joint effects have been particularly promising , especially upon the availability of many GWAS datasets and other high-throughput datasets for numerous complex diseases . In this study , we developed an integrative analysis framework for multiple GWAS datasets and demonstrated it in schizophrenia . We first constructed a GWAS-weighted protein-protein interaction ( PPI ) network and then applied a dense module search algorithm to identify subnetworks with combinatory disease effects . We applied combinatorial criteria for module selection based on permutation tests to determine whether the modules are significantly different from random gene sets and whether the modules are associated with the disease in investigation . Importantly , considering there are many complex diseases with multiple GWAS datasets available , we proposed a discovery-evaluation strategy to search for modules with consistent combined effects from two or more GWAS datasets . This approach can be applied to any diseases or traits that have two or more GWAS datasets available .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"of",
"disease",
"genome",
"analysis",
"tools",
"genomics",
"genetic",
"networks",
"genetic",
"association",
"studies",
"genetics",
"biology",
"computational",
"biology",
"gene",
"networks",
"genetics",
"and",
"genomics",
"human",
"genetics"
] |
2012
|
Network-Assisted Investigation of Combined Causal Signals from Genome-Wide Association Studies in Schizophrenia
|
Apicomplexan parasites rely on a novel form of actin-based motility called gliding , which depends on parasite actin polymerization , to migrate through their hosts and invade cells . However , parasite actins are divergent both in sequence and function and only form short , unstable filaments in contrast to the stability of conventional actin filaments . The molecular basis for parasite actin filament instability and its relationship to gliding motility remain unresolved . We demonstrate that recombinant Toxoplasma ( TgACTI ) and Plasmodium ( PfACTI and PfACTII ) actins polymerized into very short filaments in vitro but were induced to form long , stable filaments by addition of equimolar levels of phalloidin . Parasite actins contain a conserved phalloidin-binding site as determined by molecular modeling and computational docking , yet vary in several residues that are predicted to impact filament stability . In particular , two residues were identified that form intermolecular contacts between different protomers in conventional actin filaments and these residues showed non-conservative differences in apicomplexan parasites . Substitution of divergent residues found in TgACTI with those from mammalian actin resulted in formation of longer , more stable filaments in vitro . Expression of these stabilized actins in T . gondii increased sensitivity to the actin-stabilizing compound jasplakinolide and disrupted normal gliding motility in the absence of treatment . These results identify the molecular basis for short , dynamic filaments in apicomplexan parasites and demonstrate that inherent instability of parasite actin filaments is a critical adaptation for gliding motility .
Actin is an essential protein that is highly conserved in sequence and function in eukaryotic cells . Despite this conservation , parasites within the phylum Apicomplexa encode divergent actins that remain largely in an unpolymerized state in vivo and only form short , unstable filaments in vitro , in contrast to conventional actins from yeast to mammals . Apicomplexan parasites are obligate intracellular protozoan pathogens of animals including humans . Two notable members of this phylum are Toxoplasma gondii [1] , an opportunistic pathogen , and Plasmodium falciparum , the most severe cause of malaria [2] , a devastating global disease . Apicomplexan parasites move by a unique form of gliding motility that is actin-dependent [3] . Initial studies demonstrated that host cell invasion by T . gondii [4] is blocked by cytochalasin D , and it was later shown using a combination of genetic mutants in the host vs . parasite that the primary target of these treatments was parasite actin filaments , which are essential for motility and cell invasion [5] . Gliding motility is considered to be a conserved feature of the phylum [6] and has been described in T . gondii tachyzoites [7] , Plasmodium spp . sporozoites [8] , Cryptosporidium spp . sporozoites [9] , Eimeria sporozoites [10] and the more distantly related gregarines [11] . Gliding motility powers migration through tissues , traversal of biological barriers , and invasion into and egress from host cells [12] . T . gondii contains a single actin gene , TgACTI , which shows 83% amino acid identity with mammalian muscle actin [13] . P . falciparum contains two actin genes , PfACTI , that is closely related to TgACTI , sharing 93% identity at the protein level , and PfACTII , which is more divergent and has only 79% similarity to PfACTI [14] . Transcriptional analysis demonstrates that PfACTI is expressed throughout the parasite life cycle while PfACTII is most highly expressed in gametocytes [15] . Parasite actins have been shown to exist mostly in an unpolymerized state , as defined by sedimentation at 100 , 000g and an absence of staining in fixed cells with fluorescently labeled phalloidin [16] , [17] . In contrast , the majority of actin in mammalian , yeast , and amoeba cells is found in long filamentous networks , or bundled fibers , which are readily stained with phalloidin and sedimented by centrifugation at 100 , 000g [18] . Although very uncommon in apicomplexans , actin filaments have been visualized by freeze fracture electron microscopy beneath the parasite membrane in gliding T . gondii tachyzoites [17] . PfACTI from P . falciparum has been shown to form short filaments in vitro [19] , and similar short filaments of ∼100 nm in length were detected in lysates from asexually propagating stages ( i . e . merozoites ) following sedimentation at 500 , 000g [16] . Apicomplexans are also unusual in having a streamlined set of actin binding proteins consisting of actin depolymerizing factor , cyclase associated protein , profilin , and capping protein [20] , [21] , while they lack Arp2/3 [22] and many other regulatory proteins found in more complex systems . Actin dynamics are controlled in part by an inherent ability of actin monomers to polymerize head-to-tail into parallel helical strands that form filaments [18] . Polymerization is dependent on Mg2+ , salt ( i . e . KCl ) , and ATP-actin and is thermodynamically favored above the so-called critical concentration ( Cc ) [18] . Above the Cc , filaments are typically highly stable , although gradual hydrolysis of ATP and release of Pi increases disassembly and susceptibility to severing [18] . TgACTI also requires high salt and Mg2+ for polymerization , and somewhat surprisingly it initiates polymerization more readily than conventional actins , and yet it only forms short transient filaments in vivo [23] . Consistent with this finding , in vitro polymerization of TgACTI results in formation of short , irregular filaments that rapidly disassemble in the absence of stabilizing compounds such as phalloidin [23] . TgACTI fails to copolymerize with mammalian actin [23]; however , copolymerization of PfACTI with rabbit muscle actin reveals differences in monomer stacking and a larger helical pitch in parasite actin [24] . Consistent with this , previous modeling studies have suggested that instability of parasite actin filaments might arise from structural changes [23] , although this hypothesis has not been directly tested . Highly motile cells often exhibit rapid actin turnover [18] , suggesting that the unusual dynamics of apicomplexan actins may be important in gliding motility . Indirect evidence that actin turnover is important in T . gondii comes from treatment with agents that stabilize actin filaments , such as the heterocyclic compound jasplakinolide ( JAS ) , which is produced by marine sponges and acts to stabilize actin filaments [25] . JAS treatment disrupts motility and cell invasion in T . gondii [17] , [26] , as well as invasion of merozoites [27] , motility of ookinetes [28] , and endocytic trafficking in trophozoites [29] of Plasmodium . Collectively , previous studies indicate that apicomplexan actins spontaneously polymerize into short filaments that are intrinsically unstable; however , the molecular basis and functional significance of these unusual properties are largely unknown . The present study was undertaken to address two questions: 1 ) what intrinsic properties govern actin filament instability in apicomplexans ? , and 2 ) are the unusual dynamic properties of filamentous actin important for efficient motility in apicomplexans ? Here , we demonstrate that two divergent residues partially explain the inherent instability of parasite actin filaments and reveal that this feature is important for efficient gliding motility in T . gondii .
Despite overall conservation in sequence , apicomplexan actins are functionally divergent from actins in yeast , animals , and plants; likely due to molecular differences in parasite actins that affect function ( see supplemental Figure S1 ) . To visualize these shared differences , homology models were created to compare parasite actins: TgACTI and PfACTI are highly similar ( Figure 1A , yellow spheres highlight differences ) while PfACTII is more divergent ( Figure 1B ) . We expressed recombinant actins from T . gondii ( TgACTI ) , P . falciparum ( PfACTI , PfACTII ) , and yeast ( ScACT ) using baculovirus and purified these proteins to study their properties in vitro , as described previously [23] ( Figure 1C ) . Recombinant actins contained an N-terminal His6 tag that was used for purification; previous studies have shown that the presence of this tag does not alter polymerization [23] . The kinetics of actin polymerization was examined by light scattering following addition of filamentation ( F ) buffer . TgACTI only polymerized to a very limited extent ( Figure 1D , red ) , while both PfACTI and PfACTII showed modest levels of polymerization ( Figure 1D , blue and green , respectively ) . In contrast , polymerization of ScACT ( Figure 1D , orange ) was much more efficient , indicating that the inefficiency of parasite actin polymerization was not a consequence of expression in baculovirus , or the N-terminal tag shared by all of the proteins . To test the ability of parasite actins to polymerize under stabilizing conditions , purified actins were incubated with different amounts of phalloidin during polymerization in F buffer . Consistent with previous reports [26] , [30] , filaments were not detected for TgACTI in the presence of low levels of fluorescently labeled phalloidin ( i . e . 0 . 13 µM ) that was added to visualize filamentous actin ( Figure 2A ) . Short , punctate filaments were observed when a slightly higher level of labeled phalloidin ( i . e . 0 . 33 µM ) was added to TgACTI ( Figure 2A , B ) . In contrast , long clusters of filaments were observed when TgACTI was allowed to polymerize in the presence of equimolar levels of unlabeled phalloidin combined with lower levels of labeled phalloidin for visualization ( i . e . 0 . 33 µM ) ( Figure 2A , B ) . A similar dose-response to increasing phalloidin was seen for PfACTI and PfACTII , although these actins also occasionally formed small clusters of short filaments even in low levels of labeled phalloidin ( i . e . 0 . 13 µM ) ( although rare , a representative example is shown in Figure 2A , B ) . Both PfACTI and PfACTII formed more abundant clusters of short filaments in slightly higher levels of labeled phalloidin ( i . e . 0 . 33 µM ) and these were further stabilized by equimolar unlabeled phalloidin ( Figure 2A , B ) . As expected , ScACT formed long , stable filaments regardless of the phalloidin concentration ( Figure 2A ) . Interestingly , the filaments formed by ScACT and PfACTII in the presence of high levels of phalloidin ( equimolar ) were often curved , while those of TgACTI and PfACTI where extremely straight ( Figure 2 ) . Measurement of the sizes of individual filaments formed by these different actins in response to phalloidin confirmed the general patterns seen by microscopy . Both PfACTI and PfACTII formed significantly longer filamentous structures than TgACTI in the presence of low levels of phalloidin used to visualize filaments ( i . e . 0 . 13 or 0 . 33 µM ) , and all three actins showed a shift to longer filaments with with equimolar phalloidin treatment ( Figure 2B ) . Filaments formed by parasite actins were examined by negative staining and electron microscopy to reveal ultrastructural details . Similar to the fluorescent phalloidin assays , EM visualization of abundant parasite actin filaments required incubation in F buffer containing equimolar phalloidin ( Figure 3 ) . In the absence of added phalloidin , the parasite actins were observed to form irregular globular aggregates . Although we did not detect structures by EM that were similar to the small clusters seen by fluorescence staining of PfACTI and PfACTII in low levels of phalloidin ( Figure 2A ) , this may reflect the low frequency of these forms or a requirement for low levels of phalloidin to stabilize them . Enlarged images of the phalloidin stabilized filaments formed by the three parasite actins revealed a spiral pattern of the actin helix and striations along the filament , which are typical characteristics of conventional actin filaments , as observed in ScACT filaments formed under all polymerizing conditions ( Figure 3 ) . Both the parasite actins and yeast actin showed prominent filament bundles , which are also seen in the fluorescence images mentioned above . Collectively , these studies verify that the instability of parasite actin filaments generated from recombinant tagged actins is intrinsic and that polymerization is rescued by high concentrations of phalloidin . To investigate the molecular basis of phalloidin binding , we used structures from our molecular dynamics ( MD ) simulation of the muscle actin filament and performed molecular docking studies with phalloidin ( Figures 4A , 4B ) . Our predicted phalloidin binding site is similar to that reported previously [31] , but also provides more precise information on specific binding contacts that stem from the following improvements: 1 ) unconstrained docking analyses were based on a new higher resolution actin filament model [32]; 2 ) flexible protein conformations were included by choosing multiple snapshots from MD simulations and multiple binding sites were included within each snapshot; 3 ) induced fit was accommodated by simulated annealing . Together these analyses precisely mapped the phalloidin binding site in mammalian actin to the loop formed by residues 196–200 in the lower actin monomer , the 72–74 loop of the middle monomer , and the 285–290 loop of the upper monomer ( Figure 4A ) . These three regions closely coincide with those identified in previous experimental studies as important for phalloidin interactions [31] , [33] , [34] . Importantly , residues including D179 , Y198 , S199 , K284 , I287 and R290 , which were previously observed to be close to the phalloidin binding site [31] , were also within 4 Å of phalloidin in our model . Moreover , our more precise placement of phalloidin predicts maximum interaction between the Cys3-Pro ( OH ) 4-Ala5-Trp6 ring in phalloidin and actin residues , while Leu ( OH ) 7 in phalloidin faces out of the binding pocket and is accessible to solvent . This orientation corresponds well with experimental studies [31] , [33] , [34] showing that derivatives of phalloidin with a fluorophore linked to Leu ( OH ) 7 , bind actin filaments in the same conformation . Homology models for TgACTI and PfACTII were built using the muscle actin filament obtained by simulated annealing . Docking studies were repeated using TgACTI and PfACTII homology models and they yielded very similar conformations although the specific amino acid contacts lying within 4 Å varied slightly between proteins . Residues previously shown by mutational analysis to mediate phalloidin binding in yeast [35] ( i . e . R177 , D179 ) , were conserved in all three models ( Figure 4C ) . Residues R177 and D179 in mammalian actin , corresponding to R178 and D180 in parasite TgACTI , both lie within 4 Å of phalloidin ( Figure 4C ) . Six specific differences in the residues contacting phalloidin in mammalian muscle actin vs . TgACTI and PfACTII were noted ( Figure 4C , see supplemental Figure S1 ) . Together , these differences may mediate the less efficient binding to phalloidin observed for parasite actin filaments . Molecular modeling was also used to identify residues that differ between human muscle actin and TgACTI in regions that are predicted to be critical for stabilizing the actin filament . Divergent residues at positions G200 and K270 in T . gondii were identified as candidates that likely affect monomer-monomer interactions across the filament . However , the previously identified difference R277 in TgACTI , corresponding to glutamate in muscle [23] , no longer made close contact in the new filament model , and consistent with this , no change in polymerization of substituted TgACTI-R277E was observed ( data not shown ) . Instead , our refined model now points to S199 in human muscle actin as forming an important hydrogen bond with D179 of a monomer across the filament ( Figure 5A ) . This hydrogen bonding was observed in a majority of inter-monomer contacts predicted in the MD simulations of the filament . However , at this position TgACTI contains a glycine that would eliminate the hydrogen bond and potentially adversely impact filament stability ( Figure 5B ) . The second residue of interest identified was M269 in muscle actin ( Figure 5A ) that corresponds to K270 in TgACTI ( Figure 5B ) . Mutational studies in yeast have previously demonstrated that loss of hydrophobicity in this loop leads to destabilization of the actin filament [36] , and it has previously been suggested that this natural difference may affect parasite actin stability [23] . All three of the parasite actins studied here contain the alteration K/R270 , whereas G200 is found in both TgACTI and PfACTI , while PfACTII has a threonine at this position ( see supplemental Figure S1 ) . In comparing other actins to those studied here , the alteration in G200 seen in T . gondii is conserved only in ACTI homologues found in a subset of apicomplexans that rely on gliding motility ( Figure 6 ) . In contrast , the substitution of K/R in the hydrophobic plug at residue 269/270 is seen in a wider variety of protozoa including dinoflagelates , ciliates , and apicomplexans ( Figure 6 ) . The distribution of these residues among taxa on the phylogenetic trees suggests a very different ancestry for these two alterations . The presence of a positive charged residue at 269/270 is polyphyletic , being found in a wide diversity of taxa , although not in higher plants or animals . In contrast , the G200 , which was found in combination with K270 , is confined to a subset of apicomplexans , which rely on gliding motility for cell invasion ( Figure 6 ) . These same patterns were confirmed by an independent phylogeny based on maximum likelihood ( see supplemental Figure S2 ) , indicating they are robust to different phylogenetic methods of analysis . We chose T . gondii to test the importance of these two altered residues , since it is more amenable to genetic analyses . TgACTI residues were substituted with the corresponding amino acids from human muscle actin . The substituted proteins TgACTI-G200S ( hydrogen bond substitution ) , TgACTI-K270M ( hydrophobic loop substitution ) and TgACTI-G200S/K270M ( double substitution ) were expressed using baculovirus and purified with Ni-affinity chromatography ( Figure 5C ) . To determine if the substituted TgACTI alleles were more stable , purified proteins were incubated in F buffer and light scattering was used to examine polymerization . Wild type TgACTI underwent only limited polymerization while the TgACTI-K270M substituted protein showed a modest enhancement ( Figure 5D ) . However , TgACTI-G200S and TgACTI-G200S/K270M showed increased polymerization , with both the rate ( slope of the initial phase ) and maximum extent being greater than wild type protein ( Figure 5D ) . The results of the light scattering assays were confirmed using fluorescent phalloidin staining and visualization via fluorescence microscopy . In all cases , filaments were visualized in the presence of low levels of labeled phalloidin ( 0 . 33 µM ) combined with different amounts of unlabeled phalloidin . Wild type TgACTI ( WT ) required addition of 0 . 25 µM of unlabeled phalloidin to form small filaments and long filaments only formed upon addition of 5 µM ( a 1∶1 ratio with actin ) ( Figure 7A , top row ) . The TgACTI-K270M protein showed a slight enhancement in polymerization with small filaments appearing in the presence of low levels of phalloidin and reaching longer lengths with addition of 1 µM molar ratio of unlabeled phalloidin ( Figure 7A , second row ) . Interestingly , the TgACTI-G200S substitution showed much more robust polymerization with short filaments being detected even in the presence of low levels of labeled phalloidin ( 0 . 33 µM ) and longer filaments appearing with addition of 0 . 25 µM of unlabeled phalloidin ( Figure 7A , third row ) . TgACTI-G200S/K270M also formed longer filaments than seen with wild type TgACTI , similar to the TgACTI-G200S single mutant ( Figure 7A , bottom row ) . Quantitation of the filament lengths formed by different TgACTI alleles confirmed that the features seen in microscopy were consistent across different replicate samples . All three of the mutant actins formed significantly longer filaments in low levels of phalloidin needed for visualization ( i . e . 0 . 33 µM ) compared to wild type TgACTI ( Figure 7B ) . At higher levels of phalloidin , all of the actins formed long filaments of approximately equal length ( Figure 7B ) . Taken together , these results show that TgACTI filaments may be inherently unstable due to the presence of only a few differences from conventional actins and that mutations designed to mimic mammalian actin in TgACTI result in formation of more stable actin filaments in vitro . We also investigated the effects of altering yeast actin to mimic residues found in TgACTI ( i . e . converting S199 to G and L269 to K ) : neither mutation alone , nor the combination , showed dramatic change in filament assembly or length ( data not shown ) , indicating there also must be other structural and kinetic differences that explain the extremely stable nature of actin filaments in yeast and likely other conventional actins . However , these data are not inconsistent with the gain of function results seen in mutant forms of TgACTI , where the magnitude of polymerization is still relatively modest when compared to yeast . Despite the relatively small changes in actin stability observed in TgACTI mutants in vitro , we reasoned that such changes might still adversely affect actin-based processes in the parasite , such as gliding motility , which is exquisitely sensitive to actin stabilizing drugs like JAS [17] . To examine the effect of expressing stabilized mutants of TgACTI in T . gondii , we generated transgenic parasites expressing a second copy of TgACTI fused to an N-terminal degradation domain ( DD ) , which allows regulated expression in the presence of Shield-1 [37] . This approach was chosen over allelic replacement , since we reasoned that the mutant alleles might be detrimental , hence compromising attempts to evaluate their functions . The regulated nature of the DD-stabilized proteins also allows the timing of expression to be controlled , thus minimizing the chance for pleomorphic downstream effects or compensatory changes that can occur using conventional dominant negative strategies . Transgenic lines expressing the DD-fusion proteins were tested for regulated expression by Western blot using an antibody against TgACTI ( Figure 8A ) and by immunofluorescence detection of the c-myc tag also present at the N-terminus ( Figure 8B ) . The level of DD-tagged actins was approximately 50% of wild type actin and the patterns of staining was diffuse in the cytosol , similar to the pattern for endogenous actin described previously [13] . The expression of DD-tagged actins was similar at 6 , 12 , 24 , and 40 hr ( the point of natural egress ) ( see supplemental Figure S3 ) . This relatively rapid induction with continued expression maintained over time allowed us to test different biological phenotypes at different time points . Initially , the impact of expressing DD-TgACTI fusions on the life cycle of the parasite was tested under continuous treatment with Shield-1 using a plaque assay , which monitors the normal intracellular growth cycle ( Figure 8C ) . Although parasites expressing DD-G200S or DD-G200S/K270M formed plaques comparable to the controls in the absence of Shield-1 , plaque formation was almost non-existent when parasites were treated with Shield-1 ( Figure 8C ) , demonstrating that expression of stabilized TgACTI disrupts the parasite life cycle . In contrast , expression of the wild type DD-TgACTI ( DD-wild type ) had no effect on plaque formation either in the absence or presence of Shield-1 ( Figure 8C ) . Additionally , growth in Shield-1 did not significantly alter cell division during the first 36 hr ( see supplementary Figure S3 ) , indicating that the expression of wild type or mutant actins does not affect endodyogeny , although we cannot rule out the possible effects on growth from expression of these actins over longer time frames . Based on the observation that culturing parasites in Shield-1 for an entire lytic cycle does not affect replication , we treated cells for 40 hr and harvested parasites to examine the distribution and polymerization state of actin . Parasites expressing DD-TgACTI fusion proteins revealed a diffuse pattern of actin staining with some discrete puncta ( Figure 9A; data not shown ) . The absence of detectable long filaments in cells expressing stabilized actins suggest that they behave somewhat differently in vivo than in vitro , perhaps as a result of other proteins that regulate actin turnover . Actin dynamics are highly sensitive to actin-stabilizing compounds like JAS , which permeates cells and stabilizes actin filaments [38] . Hence , we examined the distribution of actins in parasites expressing DD-TgACTI alleles following treatment with low levels of JAS ( i . e . 0 . 25 µM ) . Filamentous actin structures were revealed emanating from both the apical and posterior poles in parasites expressing the stabilized TgACTI mutants grown in the presence of Shield-1 , whereas staining of wild type DD-TgACTI relocalized to the poles without forming visible filaments ( Figure 9A ) . The actin filaments seen in parasites expressing stabilized mutants of TgACTI formed spiral patterns beneath the surface of the parasite , as visualized in sequential slices of a z-series ( Figure 9B ) . Measurement of the size of actin structures in parasites revealed that larger puncta were found in parasites expressing DD-G200S and DD-G200S/K270M vs . DD-wild type TgACTI in the absence of JAS , and that these mutant actins formed considerably longer filaments in the presence of low levels of JAS ( Figure 9C ) . Co-staining of parasites expressing DD-tagged actins and treated with higher level of JAS ( i . e . 1 µM ) , revealed that the tagged alleles co-localized in actin-filament rich apical projections , which are induced by JAS ( see supplemental Figure S4 ) . To determine if the stabilized DD-TgACTI alleles polymerized more readily in vivo , we examined the proportion of globular and filamentous actin based on sedimentation at 350 , 000g , conditions previously found to be necessary to pellet short filaments that form in parasites [16] . Although no change in sedimented actin was detected in control lysates , treatment with low levels of JAS ( i . e . 0 . 25 µM ) induced much greater polymerization of the DD-G200S and DD-G200S/K270M mutants compared to DD-wild type ( Figure 9D ) . Collectively , these studies demonstrate that stabilized forms of TgACTI were more sensitive to JAS-induced polymerization in vivo . To examine the impact of stabilized TgACTI alleles on parasite motility , we employed video microscopy to analyze the typical circular and helical motions that are characteristic of gliding , as described previously [7] . In contrast to DD-wild type TgACTI expressing parasites that underwent normal circular gliding ( Figure 10A , see supplemental Video S1 ) , a large percentage of the circular movements in parasites expressing DD-G200S and DD-G200S/K270M actins were aberrant ( Figures 10B , 10C ) . For example , DD-G200S and DD-G200S/K270M expressing parasites often stalled , were unable to complete circles , or went off-track during gliding ( Figures 10B , 10C see supplemental Videos S3 , S4 , S5 ) . Quantification of these results indicated that expression of the DD-wild type allele resulted in a higher frequency of circular gliding than helical , relative to untransfected parasites , however these movements were largely normal ( Figure 10D ) . Although mutants expressing DD-G200S and DD-G200S/K270M underwent wild type motility in the absence of Shield-1 , significantly more cells exhibited aberrant forms of gliding motility in the presence of Shield-1 ( Figure 10D ) . Comparison of the radii of tracks made by DD-wild type and DD-G200S and DD-G200S/K270M expressing parasites undergoing circular gliding motility revealed that the mutants traced out partial arcs that were significantly larger than circular tracks formed by wild type parasites ( Figure 10E ) . The larger arc traced by parasites expressing stabilized actins was in part due to a modest change in the shape of the cell , as shown by measuring the curvature of the parasite body during gliding , although this difference was not significant ( Figure 10F ) . Collectively , these results indicate that expression of the mutant DD-G200S and DD-G200S/K270M forms of TgACTI disrupts normal circular gliding motility . Consistent with previous descriptions of normal helical motility [7] , DD-wild type expressing parasites underwent helical gliding at a relatively fast rate and moved through numerous corkscrew motions , noted in the example shown ( Figure 10A see supplemental Video S2 ) . In contrast , DD-G200S and DD-G200S/K270M expressing parasites were delayed in their movements and went through fewer flips and turns ( Figures 10B , 10C , see supplemental Video S6 ) . Parasites expressing the stabilized TgACTI alleles were significantly slower in both helical and circular gliding compared to the untransfected or DD-wild type parasites ( Table 1 ) . Taken together , these findings reveal that expression of stabilized mutants of TgACTI significantly disrupts gliding motility in T . gondii . Since earlier data had shown that DD-tagged alleles are induced to similar levels after 6 hr of treatment , we wanted to test the effects of shorter treatments on gliding motility , in order to rule out possible indirect effects of long-term treatment . Hence , Shield-1 was added to infected monolayers during only the last 6 hr of development , and parasites were harvested following natural egress . As described above , defects in gliding motility were seen in T . gondii parasites expressing either the DD-G200S and DDG200S/K270M mutant actins ( Figure 10G ) . Following Shield-1 treatment , parasites expressing mutant , but not wild type actin , showed increased frequency of aberrant circular trails including stalled , incomplete and off-track patterns , as well as aberrant helical patterns ( Figure 10G ) . The similar phenotypes observed at 6 hr and 40 hr of Shield-1 treatment for the mutant actins , and absence of defects in parasites expressing wild type tagged actin , indicates that the aberrant gliding phenotypes are due to expression of stabilized actin alleles , rather than non-specific effects .
Our studies suggest that the instability of parasite actin filaments is an inherent property that results in part from differences in monomer-monomer interactions that normally stabilize the filament . Although parasite actins only form clusters of small filaments when visualized with low levels of phalloidin , equimolar levels of phalloidin rescued parasite actin filament instability in vitro , resulting in long stable filaments that resembled conventional actin . Reversion of two key residues in T . gondii actin to match those predicted to stabilize mammalian muscle actin , also partially restored filament stability in vitro . Furthermore , in vivo expression of these stabilized actins led to disruption of gliding in T . gondii . These findings provide insight into the molecular basis of parasite actin filament dynamics and reveal formation of short , highly dynamic actin filaments is an important adaptation for parasite motility . Our studies are in agreement with previous work on the polymerization properties of parasite actins and extend these findings by examining the molecular basis for instability of actin filaments . We have previously reported that TgACTI undergoes polymerization in vitro as determined by tryptophan quenching and sedimentation , although the extent of this process was not compared to conventional actins [23] . In the present report , we examined actin polymerization by staining with fluorescently labeled phalloidin , sedimentation , electron microscopy , and light scattering , which provides a convenient method to study dynamics . Our findings indicate that while TgACTI undergoes polymerization , it has a very limited capacity to do so in comparison to yeast actin . There have been previous studies on the polymerization differences between muscle actin and actins from either budding [39] or fission yeast [40] , however the differences observed in these cases are relatively minor compared to what we observe here between parasite and yeast actins . Rather than subtle shifts in the polymerization kinetics or critical concentration , the extent of polymerization with these parasite actins is fundamentally different from yeast or mammalian actins . The inefficient polymerization of TgACTI was rescued by conditions that stabilize the filament including treatment with phalloidin in vitro . Other studies have previously examined PfACTI produced in yeast and concluded that it also polymerizes poorly in vitro [19] . In this prior study , stabilization with phalloidin ( 1∶4 molar ratio ) was used to achieve modest levels of polymerization of PfACTI purified from yeast and copolymerized with bovine ß-actin [19] . In a separate study , the ability of PfACTI purified from merozoites to be stabilized by phalloidin showed a pH dependence , with greater polymerization detected at pH 6 . 0 than 8 . 0 [24] , although the basis of this response is unknown . Our findings with PfACTI and PfACTII demonstrate that these actins , while modestly better at polymerization than TgACTI , also fail to polymerize robustly on their own . The reasons for this apparent instability have not been definitively resolved but could result from a lower capacity for elongation , more rapid disassembly , or a lower capacity to anneal , as described previously [41] . In contrast to yeast and vertebrate actins , our studies show that apicomplexan actins are highly dependent on addition of high levels ( i . e . equimolar ratios ) of phalloidin to form long stable filaments . Because the phalloidin binding site sits at the interface between protomers within the filament , it may overcome inherent instability caused by changes that affect monomer-monomer contacts within parasite actin filaments . Several new F-actin models have been produced in the past few years [32] , [42] and these models have given us new insight into the structural details of protomer interactions within the filament . However , the difficulties with interpreting a single , uniform F-actin structure have also been highlighted [43] , and this is precisely why we make use of molecular modeling studies in our work here . Based on our dynamics simulations , we see that actin filaments are stabilized by interactions across the width ( inter-strand ) of the filament through two key regions including the “hydrophobic plug” encompassing residues 265–270 and a helix from residues 191–199 [32] , [42] . Our studies further suggest that relatively few changes in these critical regions account for the instability of parasite actin filaments . Among these alterations , a change in the hydrophobic plug ( i . e . K270 in T . gondii ) plays a modest role while an alteration in the helix ( i . e . G200 ) has a larger affect on filament stability . The substitution of K270M in TgACTI resulted in filaments that were detected by fluorescent staining at low concentrations of phalloidin , although this change had less effect on actin polymerization as monitored by light scattering assays in the absence of phalloidin . As this residue lies within the phalloidin pocket , it suggests that hydrophobic residues here result in enhanced phalloidin binding . Mutations designed to reduce hydrophobicity in the corresponding residue in yeast actin ( i . e . L269 ) have no affect on polymerization , while those at the other end of the hydrophobic plug are much more severe [44] . Hence , these results indicate that K270 contributes to normally low phalloidin binding of parasite actins , while it likely plays a lesser role in intrinsic filament instability . Modeling predictions also indicate that S199 in muscle actin plays a role in filament stabilization via a hydrogen bond network with R177 and D179 . Consistent with this , mutation of G200S had a larger impact on the in vitro polymerization of TgACTI as shown by increased light scattering , even in the absence of phalloidin . Collectively , the absence of these two stabilizing interactions in TgACTI partially explains the inherent instability of parasite actin filaments . Intriguingly , both PfACTI and PfACTII polymerized slightly better than TgACTI in the absence of stabilizing agents , and the introduction of two alterations in TgACTI ( G200S and K270M ) resulted in polymerization to levels that approximated with wild type levels of the Plasmodium actins ( compare Figures 1 , 2 to 5 , 7 ) . Together , these findings indicate that other sequence and structural differences between these actins must contribute to their inherent differences in polymerization kinetics , which is not surprising in light of findings that even conventional actins such as yeast and muscle differ significantly [39] , [40] . The intrinsic properties of actins may be highly significant in controlling dynamics in apicomplexan parasites since they contain only a streamlined set of actin-binding proteins [20] , [21] . PfACTI contains similar substitutions to those described for TgACTI above , while PfACTII contains a K at 270 and T at 200 instead of S199 in muscle . PfACTI is expressed throughout the Plasmodium life cycle including merozoites , while PfACTII is expressed primarily in sexual stages ( [14] , [15] and EupathDB . org ) . Sporozoites and ookinetes undergo actin-dependent gliding motility on substrates and cells , while merozoites do not show substrate-dependent gliding but rely on a similar actin-dependent process for invading red blood cells [45] . In comparing the two different actin isoforms in Plasmodium , PfACTII was slightly more stable than PfACTI as shown by fluorescent phalloidin staining of filaments , raising the possibility that T200 is capable of partial hydrogen bonding , analogous to the interaction of S199 in muscle actin . Increased actin filament stability may be important in non-motile forms such as gametocytes where PfACTII is highly expressed [15] . It is also possible that the natural variation in actins found in parasite actins , and the specific changes in TgACTI mutants studied here , are influenced by interactions with actin binding proteins . Actin filament instability is evidently an important adaptation since expression of stabilized TgACTI within the parasite had a detrimental effect on gliding motility , while only modestly affecting cell division over the first 24–36 hr . Although we have not directly measured the effects on invasion or egress , these processes also depend on gliding motility and therefore are likely affected by expression of the stabilized mutants of TgACTI . Collectively , these phenotypes likely have an additive effect in the plaque assay , which captures successive rounds of invasion , replication , egress , and motility , thus leading to a more dramatic phenotype . The effects of expressing mutant actins in T . gondii partially mimic the effects of treatment with JAS , supporting the conclusion that they arise by stabilizing actin filaments . Previous studies using actin stabilizing agents such as JAS have revealed that increased polymerization of TgACTI filaments adversely effects motility and host cell invasion [17] , [26] , [30] . In the present study , stabilized mutants of TgACTI were more sensitive than wild type parasites to JAS , as shown by formation of spiral actin filaments and increased sedimentation . The spiral patterns seen here are similar to those reported previously from wild type T . gondii treated with high levels of JAS [17]; however , notably here they occur with low levels of JAS and are only seen in mutants expressing stabilized TgACTI forms . Stabilized DD-TgACTI mutants also had a profound effect on disrupting normal motility in the absence of treatment , revealing that this phenotype is not simply due to enhanced binding to JAS or phalloidin . Intriguingly , parasites expressing stabilized actins formed circles with larger radii , moved more slowly , and stalled in the process of gliding . These larger arcs were in part due to a more relaxed curvature of parasites expressing mutant actins , although this difference was much less pronounced than that seen in the trails . Hence , the increased trail radii likely results from the parasite slipping off its track as it migrates around the circle . Previous studies have also shown that the degree of actin polymerization can influence adhesive strength and hence the gliding behavior of Plasmodium sporozoites [46] . Collectively these data suggest that short , highly dynamic actin filaments are required for parasites to complete the tight arcs and corkscrew turns that are characteristic for circular and helical gliding [47] . The current model for gliding motility predicts that short , highly dynamic actin filaments attached to transmembrane adhesive proteins are translocated along the surface of the parasite by a small myosin [48] . The myosin motor , which is anchored in the inner membrane complex [49] , is also highly nonprocessive [50] , meaning it does not stay attached to a single filament for long periods . Instead , this model predicts that short actin filaments , tethered to transmembrane adhesins , are passed sequentially between motor complexes that operate independently . Consistent with this , where actin filaments have been seen in parasites , they are quite short ( i . e . 50–100 nm ) [16] , [23] . Actin in apicomplexans may be adapted for rapid turnover of short filaments , since long filaments would increase the likelihood of multiple motors being engaged simultaneously , potentially leading to conflicting forces on the same filament . Although we were not able to discern distinct filaments in parasites expressing DD-TgACTI proteins , the observed punctate staining pattern may reflect clusters of short filaments that are below the resolving power of the light microscope ( in theory ∼200 nm , but in practice likely ∼400 nm ) . Nonetheless , we would predict based on their in vitro properties that the G200S and G200S/K270M mutants would form more stable filaments , which could inhibit motility by reducing free monomers needed for new filament assembly , or by physically disrupting productive motor-actin filament complexes . Alternatively , stabilized DD-TgACTI mutants could affect interactions with actin-binding proteins in vivo , including those involved in polymerization or depolymerization . Although apicomplexans lack an Arp2/3 complex [22] , they express several formins that act to increase actin polymerization [51] , [52] and actin depolymerization factor , which acts primarily to sequester monomers and prevent polymerization [53] . Regardless of the exact mechanism , our results indicate that even subtle changes in actin filament stability significantly affect function , underscoring the importance of rapid actin dynamics in apicomplexans . In comparing apicomplexans to other organisms , the G200S mutation is found in a subset of apicomplexans including Toxoplasma , Neospora , Eimeria , and Plasmodium spp . but excluding Theileria , Babesia , Cryptosporidium , and gregarines . Hence it is uncertain if conversion to G200 arose in the common ancestor of coccidians ( monoxenous and tissue cyst forming ) and hematozoa and was subsequently lost by some members , or if arose independently in Plasmodium and the coccidian . Gliding motility has not been described in Thieleria , which enters lymphocytes by a very different process than other apicomplexans [54] . However , Babesia enters red cells by a process very analogous to that seen in Plasmodium [55] , and so likely has a conserved mechanism for actin-based motility . Cryptosporidium and gregarines also move by gliding motility [6] , although the polymerization properties of actins from these organisms have not been examined . Hence it is unclear whether these other apicomplexans rely on more stable actins , or if other divergent residues impart similar properties to those observed in Toxoplasma and Plasmodium . The substitution of K/R in the hydrophobic plug at residue 269/270 is seen in a wider variety of protozoa including dinoflagelates , ciliates , and apicomplexans . Consistent with this , diverse actins from protozoans Leishmania [56] , Giardia [57] , and Tetrahymena [58] have also been reported not to bind well to phalloidin and to display unusual polymerization kinetics or novel actin structures . This pattern further suggests that stable actin filaments are a more recent evolutionary development , found in amoeba , yeast , plants and animals , but not shared by many protozoans . There are some exception to this pattern , such as Giardia , which expresses a very divergent actin that nonetheless forms stable filamentous structures [57] . Although no kinetic measurements have been reported for Giardia actin as of yet , when available they will provide extremely useful comparisons to other systems . Overall these differences in actin filament stability likely reflect adaptations for stable vs . dynamic actin cytoskeletons that are designed for very different life strategies . The importance of dynamic actin turnover in apicomplexans is shown by introduction of stabilizing residues in TgACTI , changes that were sufficient to dramatically slow the speed of gliding and result in aberrant forms of motility . Collectively , these findings demonstrate that actin filament instability and rapid turnover are important adaptations for productive gliding in apicomplexans , and suggest that small molecules designed to selectively stabilize parasite actins may have potential for preventing infection .
Recombinant Toxoplasma , Plasmodium and yeast actins were expressed in baculovirus , as previously described [23] . Recombinant viruses were created by amplification from 3D7 strain of Plasmodium falciparum cDNA or Saccharomyces cerevisiae cDNA using gene-specific primers ( Table S1 ) and the resulting products were cloned into the viral transfer vector pAcHLT-C ( BD Biosciences Pharmingen ) . Recombinant viruses were obtained by cotransfection with linearized baculogold genomic DNA into Sf9 insect cells ( BD Biosciences Pharmingen ) , according to manufacturer's instructions . Recombinant viruses for mutant TgACTI alleles and were created via site-directed mutagenesis using wild type TgACTI as a template and allele-specific primers ( Table S1 ) . Hi5 insect cells were maintained as suspension cultures in Express-Five SFM media ( Invitrogen ) . Hi5 cells were harvested at 2 . 5 days postinfection with recombinant virus and lysed in BD BaculoGold Insect Cell Lysis Buffer ( BD Biosciences Pharmingen ) supplemented with 0 . 2 mM CaCl2 , 0 . 2 mM ATP , 0 . 2 mM NaN3 , and protease inhibitor cocktail ( E64 , 1 µg ml−1 AEBSB , 10 µg ml−1; TLCK , 10 µg ml−1; leupeptin , 1 µg ml−1 ) . His-tagged actins were purified using Ni-NTA agarose ( Invitrogen ) . After binding for 2 hr , the column was washed sequentially with G actin buffer without DTT ( G-DTT buffer ) ( 5 mM Tris-Cl , pH 8 . 0 , 0 . 2 mM CaCl2 , 0 . 2 mM ATP ) , then G-DTT buffer with 10 mM imidazole , G-DTT buffer with 0 . 5 M NaCl and 10 mM imidazole , G-DTT buffer with 0 . 5 M KCl and 10 mM imidazole , and finally G-DTT buffer with 25 mM imidazole . Proteins were eluted with serial washes of G-DTT buffer containing 50 mM , 100 mM , and 200 mM imidazole , pooled together and dialyzed overnight in G-actin buffer containing 0 . 5 mM DTT with 100 µM sucrose . Purified recombinant actins were clarified by centrifugation at 100 , 000g , 4°C , for 30 min using a TL100 rotor and a Beckman Optima TL ultracentrifuge ( Becton Coulter ) to remove aggregates . Purified proteins were resolved on 12% SDS-PAGE gels followed by SYPRO Ruby ( Molecular Probes ) staining , visualized using a FLA-5000 phosphorimager ( Fuji Film Medical Systems ) , and quantified using Image Gauge v4 . 23 . Purified actins were stored at 4°C and used within 2–3 days . Purified recombinant actins were clarified as described above and incubated ( 5 µM ) in F buffer ( 50 mM KCl , 2 mM MgCl2 , 1 mM ATP ) , and treated with different molar ratios of unlabeled phalloidin to actin from 0∶1 to 1∶1 ( Molecular Probes ) . In addition , final concentrations of 0 . 13 µM or 0 . 33 µM Alexa-488 phalloidin ( Molecular Probes ) were added to each sample to visualize filaments . Following polymerization for 1 hr , samples were placed on a slide and viewed with a Zeiss Axioskop ( Carl Zeiss ) microscope using 63× Plan-NeoFluar oil immersion lens ( 1 . 30 NA ) . Images were collected using a Zeiss Axiocam with Axiovision v3 . 1 and processed using linear adjustments in Adobe Photoshop v8 . 0 . Filament lengths were determined using the measurement feature of Axiovision software ( Zeiss ) . For each actin sample , filaments were measured from 8–10 fields ( 63× ) within three biological replicates . Purified recombinant actins were clarified as described above and incubated ( 5 µM ) in G buffer containing 1 mM EGTA and 50 µM MgCl2 for 10 min ( to replace bound Ca2+ with Mg2+ ) . Samples were placed in a submicrocuvette ( Starna Cells ) and following addition of 1/10th volume of 10× F buffer , light scattering was monitored with the PTI Quantmaster spectrofluorometer ( Photon Technology International ) with excitation 310 nm ( 1 nm bandpass ) and emission 310 nm ( 1 nm bandpass ) . Curves were processed by second order smoothing with 15–30 neighbors using Prism ( Graph Pad ) . Homology models for TgACTI , PfACTI , and PfACTII sequences were built on the ADP-actin crystal structure ( 1J6Z ) [59] using Modeller [60] . Homology models were aligned and visualized using VMD [61] . Protein sequences for actins from Homo sapiens ( muscle α-actin ) , gi: 6049633; Saccharomyces cerevisiae , gi: 38372623; Toxoplasma gondii , gi: 606857; Plasmodium falciparum ACTI , gi: 160053; and Plasmodium falciparum ACT2 , gi: 160057; were aligned using DNASTAR Lasergene MegAlign v7 and modified using Adobe Illustrator v10 . An atomic model of phalloidin was derived from the solid state structure of a synthetic derivative [62] , modified to contain dihydroxy-Leu7 using Maestro ( Schrödinger LLC , ) and energy minimized using MacroModel ( Schrödinger LLC , ) with a MMFF94s forcefield . The model was further optimized in continuum solvent using Jaguar ( Schrödinger LLC ) , with DFT level of theory using a hybrid B3LYP functional and 6-31G** basis set . The actin filament model based on X-ray fiber diffraction data [32] was used to create an 8-monomer filament of muscle F-actin . A 50 ns molecular dynamics ( MD ) simulation in explicit water was carried out using NAMD [63] in an NpT ensemble with a pressure of 1 atm and a temperature of 300 K with explicit TIP3P water . CHARMM27 forcefield was used with a 10 Å cut off for van der Waals with a 8 . 5 Å switching distance , and Particle Mesh Ewald for long-range electrostatics . Bonded hydrogens were kept rigid to allow 2 fs time steps . A simulated annealed structure of muscle filament model with phalloidin in the binding site was used as the template for building parasitic actin filament homology models using Modeller [60] . Docking of phalloidin to different sites along the filament was captured using multiple snapshots taken at intervals of 200 ps from the 50 ns simulation . AutoDock [64] was used to perform large scale docking runs with a coarse grid that covered the six binding sites on the filament . To determine the correct orientation of phalloidin in the binding site , higher resolution docking studies were performed on each binding site using both AutoDock and Glide ( Schrödinger LLC ) in independent trials and clustered to derive the most probable docking orientation . For AutoDock , flexible ligand docking was performed using Lamarckian genetic algorithm with a population size of 200 , 10 million energy evaluations , and a local search probability frequency at 0 . 2 . Grid spacings of 0 . 325 Å and 0 . 25 Å were used for coarse and high resolution docking , respectively , and the results were clustered at RMSD of 3 . 0 Å from the lowest docked energy conformer . Gasteiger-Marsili charges were assigned to the ligand using Sybyl ( Tripos Inc . , ) . Default parameters were used for Glide; ligand charges were derived from the quantum optimization calculation and protein charges were derived from the OPLS2001 forcefield . TgACTI alleles were amplified by PCR and inserted into a modified vector pTUB-DD-myc-YFP-CAT-Pst1 [37] at unique Pst1-AvrII sites to generate DD-TgACTI fusions . The resulting plasmids were transfected into tachyzoites of the RH strain of Toxoplasma and parasites were single celled cloned on monolayers of HFF cells and propagated as previously described [65] . For intracellular staining , parasites were allowed to invade HFF monolayers on glass coverslips for 24 hr in the presence or absence of 4 µM Shield-1 . The coverslips were then fixed with 4% formaldehyde and stained with mouse anti-c-myc ( Zymed ) to detect the DD-fusion proteins followed by goat anti-mouse IgG conjugated to AlexaFluor 488 ( Molecular Probes ) and mAb DG52 ( anti-TgSAG1 ) directly conjugated to AlexaFluor 594 to detect the parasite . To examine the pattern of actin following expression of DD-tagged actins , parasites were cultured in Shield-1 for one lytic cycle ( i . e . 40 hr ) and then harvested following natural egress . Freshly harvested parasites were treated ±0 . 25 µM JAS ( Invitrogen ) for 15 min and allowed to glide for 15 min on glass coverslips coated with 50 µg ml−1 BSA . Coverslips were fixed and stained with mouse anti-cmyc ( Zymed ) followed by goat anti-mouse IgG conjugated to AlexaFluor 488 and mAb DG52 labeled with AlexaFluor 594 . Coverslips were mounted in Pro-Long Gold anti-fade reagent ( Invitrogen ) and viewed with a Zeiss Axioskop ( Carl Zeiss ) microscope using 63× Plan-NeoFluar oil immersion lens ( 1 . 30 NA ) . Images were collected using a Zeiss Axiocam and deconvolved using a nearest neighbor algorithm in Axiovision v3 . 1 . Images were processed using linear adjustments in Adobe Photoshop v8 . 0 . To determine the length of actin filament structures , the longest continuously staining patterns ( i . e . puncta , filaments , or spirals ) were determined using the measurement feature of Axiovision software ( Zeiss ) . Measurements were made from 6–8 separate parasites from each of the groups ( i . e . DD-wild type , DD-G200S , and DD-G200S/K270M ) ± JAS treatment . Plaque assays were conducted by adding 300 purified parasites to HFF monolayers in 6-well dishes containing medium+DMSO or medium +3 µM Shield-1 in DMSO and incubated at 37°C with 5% CO2 for 7 days . Plates were then fixed with 70% ethanol and stained with 0 . 01% crystal violet to visualize plaques . Freshly lysed parasites were used to infect HFF monolayers with the addition of 4 µM Shield-1 for 6 , 12 , 24 or 40 hr prior to egress . Following natural egress , parasites were filtered , spun at 400g for 10 min and resuspended in Laemmli sample buffer . Parasite lysates from each time point were resolved on 12% SDS-PAGE gels , Western blotted with anti-TgACTI antibody , visualized using a FLA-5000 phosphorimager ( Fuji Film Medical Systems ) and quantified using Image Gauge v4 . 23 . Analysis of replication of parasites expressing DD-TgACTI alleles were conducted by adding freshly egressed parasites to HFF monolayers on coverslips in 24 well plates containing medium +DMSO or medium+4 µM Shield-1 in DMSO and incubated at 37°C with 5% CO2 for 24 or 36 hr at which time the coverslips were fixed and stained with mAb DG52 labeled with AlexaFluor 488 . Coverslips were mounted in Pro-Long Gold anti-fade reagent ( Invitrogen ) and viewed with a Zeiss Axioskop ( Carl Zeiss ) microscope using 63× Plan-NeoFluar oil immersion lens ( 1 . 30 NA ) . The numbers of parasites per vacuole were counted in triplicate from at least 50 vacuoles per coverslip from three replicate experiments . Parasite strains expressing DD-tagged actins were treated ±0 . 5 µM JAS for 30 min , lysed with Triton-X-100 for 1 hr , centrifuged at 1 , 000g , 4°C for 2 min and supernatants centrifuged at 350 , 000g , 4°C for 1 hr using a TL100 rotor and a Beckman Optima TL ultracentrifuge ( Becton Coulter ) . Supernatant proteins were acetone precipitated and washed with 70% ethanol . All pellets were resuspended in 1× sample buffer , resolved on 12% SDS-PAGE gels , Western blotted with anti-TgACTI antibody , visualized using a FLA-5000 phosphorimager ( Fuji Film Medical Systems ) , and quantified using Image Gauge v4 . 23 . Parasite gliding was monitored by video microscopy as previously described [7] . Parasites were treated with DMSO or 4 µM Shield-1 for 6 or 40 hr , resuspended in Ringer's solution and allowed to glide on uncoated glass coverslips . Images were captured with 50–100 ms exposure times at 1 sec intervals , combined into composites with Openlab v4 . 1 ( Improvision ) , analyzed using ImageJ and saved as QuickTime videos . Cell motility was tracked using the ParticleTracker plug-in to evaluate average speeds from a 3–15 tracks . The percentage of parasites undergoing different forms of motility was quantified from 4 or more separate videos , 60 sec in length and containing 10–40 motile parasites each , using Cell Counter , as described [66] . Radii of circular trail patterns and of the curvature of gliding parasites were determined using the measurement feature of Axiovision software ( Zeiss ) . Measurements of the radii of trails were made from tracks of individual parasites from 4 separate videos containing 10–40 motile parasites each . The curvatures of individual parasites were determined from parasites undergoing circular ( DD-wild type ) vs . off-track and stalled ( DD-G200S , G200S/K270M ) patterns of motility . The curvature of individual parasites was measured independently from 3–5 separate frames from a single motility track taken from representative time-lapse recordings . Sequences of actins for 83 organisms including a variety of protists , plants , fungi and animals , were obtained from GenBank and aligned using Clustal [67] with a gap opening penalty of 30 and extension penalty of 0 . 75 . The alignment ( see supplementary Figure 5 ) was imported as a NEXUS file in the PAUP* [68] and used to generate tress by Neighbor-Joining distance using BioNJ and 1000 bootstrap replicates . Only branches of >50% were retained in building the consensus tree . Unrooted trees were drawn in TreeView [69] . Separately , the alignment file was imported as a nexus file into HyPhy [70] and used to generate a maximum likelihood tree under the HKY85 model with 100 bootstrap replicates . Statistics were calculated in Excel or Prism ( Graph Pad ) using unpaired , two-tailed Student's t-tests for normally distributed data with equal variances , and two-tailed Mann-Whitney analysis for analysis of samples with small samples sizes of unknown distribution . Significant differences were defined as P≤0 . 05 .
|
Cellular movement is key to life and in the case of intracellular parasites , provides a vital mechanism to gain access to the protected niche they require . The parasite Toxoplasma gondii is a model for a group of parasites called apicomplexans , which move by an actin-dependent process referred to as gliding motility . This form of motility is distinct from that used by ciliated or flagellated cells , and from the crawling behavior of amoeba and many mammalian cells . We demonstrate that the normally highly conserved protein actin is divergent in these parasites and that it displays unusual kinetic properties that result in formation of short unstable filaments , in contrast to the highly stable nature of mammalian actin . Our findings reveal that the short dynamic nature of parasite actins is due to a small number of amino acid differences that affect stability of the filament . Moreover , these properties are essential to normal parasite motility since reversion of these residues to match those seen in mammalian cells was detrimental to gliding movement . The dependence of parasites on rapid turnover of highly unstable actins renders them extremely sensitive to toxins that stabilize actin filaments , thus providing a potential target for development of specific intervention .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"biochemistry",
"genetics",
"biology",
"microbiology",
"evolutionary",
"biology",
"biophysics",
"genetics",
"and",
"genomics"
] |
2011
|
Evolutionarily Divergent, Unstable Filamentous Actin Is Essential for Gliding Motility in Apicomplexan Parasites
|
We consider the statistical analysis of population structure using genetic data . We show how the two most widely used approaches to modeling population structure , admixture-based models and principal components analysis ( PCA ) , can be viewed within a single unifying framework of matrix factorization . Specifically , they can both be interpreted as approximating an observed genotype matrix by a product of two lower-rank matrices , but with different constraints or prior distributions on these lower-rank matrices . This opens the door to a large range of possible approaches to analyzing population structure , by considering other constraints or priors . In this paper , we introduce one such novel approach , based on sparse factor analysis ( SFA ) . We investigate the effects of the different types of constraint in several real and simulated data sets . We find that SFA produces similar results to admixture-based models when the samples are descended from a few well-differentiated ancestral populations and can recapitulate the results of PCA when the population structure is more “continuous , ” as in isolation-by-distance models .
In this section , we describe how admixture-based models and PCA can be viewed as factorizing an observed genotype matrix into a product of two low-rank matrices . We assume that contains the genotypes of individuals at SNPs with genotypes coded as copies of a reference allele . Then both admixture-based models and PCA can be framed as models in which: ( 1 ) or , equivalently , ( 2 ) where is a matrix and is a matrix , where is typically small ( Figure 1 ) ( see Table 1 for a complete list of terms and constraints ) . In this framework , the primary difference between the approaches lies in the constraints or prior distributions placed on matrices and as follows .
For simplicity we begin by applying the methods to a small data set of SNPs typed on unrelated HapMap individuals: Europeans , Africans , and Chinese and Japanese ( data from [29] ) . In these data , the three continental groups are well separated , making interpretation of the results relatively straightforward and selection of an appropriate number of factors simple . ( We discuss the issue of selecting an appropriate number of factors later . ) We ran SFA and admixture with three factors; since both of these methods involve a numerical optimization we ran each times , using different random starting points , and in each case the results were effectively identical across runs . Figure 3 compares the loadings from SFA and admixture with the first three PCA loadings . All three methods clearly separate out the three groups , but SFA and admixture produce qualitatively different results from PCA . In particular , in SFA and admixture , each individual has appreciable loading on only one of the three factors; from this we infer that the three corresponding factors each represent the allele frequencies of a single continental group . In contrast , in PCA , each individual has appreciable loading on all three factors , and the factors themselves do not have such a straightforward interpretation . In some ways the different representations obtained by SFA , PCA , and admixture are equivalent: the resulting matrix product , , from each method is essentially identical ( not shown ) . However , in this case we view the results of SFA and admixture as more easily interpretable . Specifically , the three SFA and admixture factors correspond to the Asian , African , and European allele frequencies , respectively . In contrast , the first PCA factor corresponds to the overall mean allele frequency , and subsequent factors correspond to other linear combinations of the allele frequencies in each group . These differences are driven by the different constraints on the and matrices , not by one factorization fitting the data better . Note that , although PCA is forced into using the mean allele frequencies as its first factor by our following the common practice of applying it to the standardized genotype matrix with the genotype means removed , in this case PCA produces almost identical results when applied to the original genotype matrix ( results not shown ) . One consequence of SFA and admixture factors corresponding to individual group frequencies is that their results are more robust to the number of individuals included from each group . For example , when we removed half of the Africans from the sample and reran the methods , the results from SFA and admixture were essentially unchanged , whereas PCA results changed more appreciably ( Figure S1 ) . The intuition here is that , for SFA and admixture , removing some African individuals has only a small effect on the factor corresponding to Africans ( because the sample African allele frequencies change slightly ) and a negligible effect on the factors corresponding to the European and Asian individuals . These small changes in the factors translate into correspondingly small changes in the loadings for each remaining individual . In contrast , removing half of the Africans changes all three PCA factors: the modified sample has a different overall mean allele frequency ( first factor ) , and this has a cascading effect on subsequent factors and their loadings . Indeed , the general lack of robustness of PCA to sampling scheme is well known [30] , [31] . In more complex settings , we have also found SFA and admixture to be more robust than PCA to sampling scheme . We illustrate this using data on SNPs typed in individuals from worldwide populations , including the HapMap individuals considered above plus the Human Genome Diversity Panel [29] . These data contain a much higher proportion of individuals with European or Asian ancestry than the HapMap data alone . Analyzing these data with three factors , SFA and admixture produce loadings for the HapMap individuals that are essentially identical to those obtained from the analysis of the HapMap individuals alone ( Pearson correlation for SFA; for admixture ) . In contrast , the corresponding PCA loadings change more substantially ( correlation ) . We now compare the methods on some simple isolation-by-distance scenarios , involving both one dimensional and two dimensional habitats . For the 1-D habitat we assume demes equally-spaced on a line , and for the 2-D habitat we assume demes arranged uniformly on a by square grid . In each case demes are assumed to exchange migrants in each generation with neighboring demes . We applied PCA , SFA and admixture to data from both 1-D and 2-D simulations . In the 1-D scenario , for each method , two factors suffice to capture the underlying geographical structure ( Figure 4 ) . However , as for the discrete data considered above , the interpretations of the resulting factors differ across methods . In SFA and admixture , the two factors represent , roughly , the allele frequencies near either end of the line ( Figure 5 ) . The genotype of each individual along the line is then naturally approximated by a linear combination of these two factors , with weights determined by their position along the line ( e . g . , individuals near the center of the line have roughly equal weight on the two factors ) . The loadings in SFA seem to capture the underlying structure slightly better near either end of the line than those from admixture , whose loadings effectively saturate at zero on the first and last third of each line . This may partly reflect the constraint that the admixture loadings must sum to one , but may also be exacerbated by the assumption of a binomial distribution , and in particular the assumption of a binomial variance . In contrast , in PCA , the first factor represents the mean allele frequencies and the second represents a difference between the allele frequencies near either end of the line . Thus PCA represents each individual as the mean allele frequency , plus the allele frequency difference weighted according to the location of the individual relative to the center ( the weight being zero for individuals near the center of the line , positive at one end of the line , and negative at the other ) . Again , this behavior is not solely due to our applying PCA to the standardized genotype matrix: it produces almost identical results when applied to the original genotype matrix ( results not shown ) . For the 2-D scenario ( Figure 6 ) , the methods differ more substantially in their results . In particular they differ in the number of factors that they need to model the underlying geographical structure . Due to the convexity constraint , admixture requires four factors , corresponding roughly to the allele frequencies at the four corners of the square habitat . ( This result depends on the shape of the habitat; intuitively , the convexity constraint means that admixture needs a factor for each extreme point of a convex habitat . ) Even then , the 2-D structure is only easy to visualize after the four factor loadings have been mapped into two dimensions ( see Methods ) . As in the 1-D setting , the loadings for individuals near the edges of the grid saturate near zero or one . In contrast , both PCA and SFA can capture the structure using three factors , although again they accomplish this in different ways . PCA uses the mean allele frequencies as the first factor , and then two factors that represent deviations from this mean in two orthogonal directions ( e . g . , the diagonals of the square ) . As a result the PCA loadings on the second and third factors effectively recapitulate the geography of the space , as previously observed [14] , [15] , [30] . The results from SFA are more complicated to describe . All three factors represent linear combinations of the allele frequencies on the grid , where the weights of these allele frequencies vary in a consistent way along a particular direction . For example , in the first row of Figure 6B , the first factor has increasing weight as one moves from the bottom to the top of the grid . The result is that the loadings from any two factors recapitulate a skewed version of the geography . In both of these settings , particularly the 2-D case , the PCA loadings seem to have the simplest interpretation . This is because , after subtracting the genotype mean , the 1-D structure can be captured by a single factor , and the 2-D structure captured by two factors , in each case yielding an attractive geographical interpretation . Thus PCA's use of the mean allele frequency as its first factor , which hinders interpretability in the discrete case , actually aids interpretability in settings with more continuous structure . However , the use of the mean allele frequencies as the first factor need not be limited to PCA . In particular it is straightforward to modify SFA to behave in a similar way , either by applying it to the genotype matrix with the genotype means subtracted , or by modifying the model to include a mean term ( i . e . , a factor for which all individuals have loading one ) . We take the later path here because we think there are advantages to estimating the mean along with the factors , rather than as a preprocessing step . We refer to this approach as SFAm; see Methods for details . Applying SFAm to both the 1-D and 2-D scenarios produces results that are effectively identical to PCA , recapitulating the geographic structure in one or two additional factors respectively ( Figure 4 and Figure 6 ) . In summary , the fact that the first factor in PCA represents the mean allele frequencies is responsible both for the fact that it produces less interpretable factors in the discrete case and more interpretable results in the continuous case . Because SFA provides the flexibility of choice whether or not to include the mean , it can produce interpretable results in both scenarios . Indeed , in the discrete case SFA effectively recapitulates the results of admixture , and in the continuous settings SFAm effectively recapitulates the results of PCA . Up to now we have avoided discussion of automatic selection of an appropriate number of factors , instead relying on intuition and heuristic arguments to guide this selection . In principle one could attempt to formalize this process within a model-selection framework , since SFA has an underlying probabilistic model . However , automatic selection of an appropriate number of factors is difficult , not least because in many practical applications there does not exist a single “correct” number of factors . For example , our 1-D simulations involved discrete populations exchanging migrants locally , so in some sense a “correct” number of factors is , but for realistic-sized data sets reliably identifying factors will not be possible , and analyzing the data with factors is unlikely to yield helpful insights . Note that interpretability of factors does not necessarily correspond with statistical significance: in isolation by distance scenarios many PCA factors may be statistically significant [13] , but usually only the first few are easily interpretable , with additional factors representing mathematical artifacts [30] . For these reasons , in practice it can be helpful to run methods such as admixture and SFA multiple times , with different numbers of factors , to see what different insights may emerge . ( PCA need only be run once , because adding additional factors does not change existing factors . ) To illustrate these issues we applied the methods to a situation that mimics clustered sampling from a continuous habitat; specifically we used samples of twenty individuals from each of five evenly-spaced demes from the 1-D simulation above . These samples can be represented in either a low-dimensional way , as five clusters along a continuum , or a higher-dimensional way , as five distinct populations . Applying SFA to these data ( Figure 8A ) , we obtain qualitatively different results depending on the number of factors used: with two factors the SFA loadings represent the five demes as five points along a line ( so each factor corresponds , roughly , to the allele frequencies near each end of the line ) , whereas , with five factors , the SFA loadings separate the five demes into discrete groups ( so each factor corresponds to the allele frequencies within a single deme ) . Applying admixture to these data ( Figure 8B ) , we obtain similar results as for SFA , except that in the two factor case the five groups are compressed into three groups . Thus , as with the 1-D isolation-by-distance simulations , admixture tends to over-discretize continuous variation . Applying PCA to these data ( Figure 8C ) , the first two factors capture the continuous variation along the line , as in the 1-D simulations . Subsequent factors each distinguish finer-scale structure among the five demes , and the first five PCA factors , jointly , fully capture the structure . However , each factor is individually difficult to interpret . In particular , because computing additional PCA factors does not affect earlier factors , PCA never reaches a representation in which five factors each represent the allele frequencies of a single deme . Applying SFAm to these data , with one factor plus the mean term , produces results almost identical to the first two factors of PCA ( results not shown ) . In summary , this simulation illustrates two important points . First , there is not necessarily a single “correct” number of factors: by applying methods such as SFA and admixture with different numbers of factors , we may obtain qualitatively different results that provide complimentary insights into the underlying structure . Second , SFA seems to be more flexible than either PCA or admixture in its ability to represent both discrete and continuous structure . We now compare the three methods on a set of European individuals , consisting of genotype data on individuals at 200 , 000 SNPs ( after thinning to remove correlated SNPs ) . The collections and methods for the Population Reference Sample ( POPRES ) are described by [32] . Previous analyses of these and similar data using PCA have found that the first two PCA factors recapitulate the geography of Europe ( e . g . , [14] , [15] ) . Based on the results from the 2-D simulations , we chose to apply SFAm ( with two factors plus a mean ) here , rather than SFA . The results from SFAm are strikingly similar to those from PCA ( Figure 9 ) . In a few cases the sparsity-inducing prior we used in SFAm is evident , in that there is a slight tendency for factor loadings near zero to be shrunk closer to zero ( appearing as faint diagonal lines of individuals in the rotated SFAm plot ) . However in general the effect of the sparsity-inducing prior is minimal in these kinds of situations , where the data do not actually exhibit sparsity . Different runs of SFAm produce alternative rotations of this same basic image . As in the 2-D simulations , admixture with four factors is able to capture the geography , but only after these four factors have been mapped to a two-dimensional space ( see Methods ) . As in the 1-D and 2-D simulations , admixture tends to push the data towards the extremes relative to PCA or SFAm , although this effect is substantially less prominent than in the simulations ( perhaps due , in part , to the larger number of SNPs ) . The ability of admixture-based models to capture geography has been noted before [33] . All three methods are computationally tractable for data sets of this size . Of the three methods , PCA was fastest and admixture was slowest , but all three methods took less than a few hours on a modern desktop . Recall that , in settings with discrete structure , the SFA factors , like the admixture factors , correspond to the allele frequencies of each discrete populations . One consequence of this is that in settings involving admixed groups , the SFA loadings are highly correlated with the admixture proportions of each individual . Indeed , in some settings it is possible to translate the SFA loadings into estimates of admixture proportions . Specifically , if an individual has all positive loadings , and the loading on factor is , then is a natural estimate of that individual's admixture proportion from the population represented by factor . However , this estimate assumes implicitly that factors have all been scaled appropriately , which will only be true if the variance of the allele frequencies in the ancestral populations is similar ( something that may well hold in many contexts , but would be difficult to check ) . To compare all three methods on real data that appear to involve admixture , we consider the data from a recent study on individuals from India [2] . These data were sampled from “groups” geographically distributed across India; [2] hypothesized the different groups to be admixed between two ancestral population: ancestral north Indians ( ANI ) and ancestral south Indians ( ASI ) . This is a challenging data set for admixture analysis because the sample contains no individuals representative of either of the two ancestral populations . For this reason , [2] uses a novel tree-based method ( ancestry estimation , described in their supplemental information ) to estimate the ancestry proportions of each group . We applied PCA , SFA with two factors , and admixture with two factors to the genotype data from this study , after imputing the missing genotypes , removing some of the outlier populations as defined in the original study , and removing SNPs with a minor allele frequency less than ( see Methods ) . We encountered problems applying SFA to these data with the low frequency SNPs included; specifically , SFA often converged to a solution where one individual had a very small residual variance term . All three methods produce very similar loadings ( Figure S2 ) that correlate well with the ancestry proportions estimated in [2] ( Pearson correlations of for PCA , for SFA , and for admixture ) ( Figure 10 ) . In one sense , the factor loadings provide more detailed ancestry information than the method , because the loadings are individual-specific rather than group-level . However , in this setting , the loadings provide measures of individual-specific ancestry that are reliable only in a relative sense . That is , they may correctly order the individuals in terms of their degree of ancestry in each ancestral population , but do not necessarily provide accurate ancestry proportions for each individual . For example , the estimated ancestry proportions from admixture range from to , whereas the group-level estimates from the method range from to . This reflects the difficulty of reliably estimating the ancestral population allele frequencies in the absence of any reference individuals from the ancestral populations .
In this paper we have presented a unified view of the two most common methods to analyzing population structure – admixture-based models and PCA – by interpreting both as matrix factorization methods with different constraints on the matrices . This unification provides insights into the different behavior of these methods under various scenarios . For example , viewing admixture-based models as imposing a convexity constraint explains why these models would be expected to need four factors to capture the structure across a square habitat , whereas PCA requires only two factors plus a mean . Viewing these methods as special cases of a much larger class of matrix factorization methods also immediately suggests many possible novel approaches to the analysis of population structure . Here we consider one such method , sparse factor analysis ( SFA ) . We illustrate that SFA bridges the gap between PCA and admixture-based models by effectively recapitulating the results from admixture-based models in discrete population settings , and recapitulating the results from PCA in continuous settings . We also illustrate a scenario involving a mixture of discrete and continuous structure where SFA produces more interpretable results than either admixture-based models or PCA . We have also experimented with two other matrix factorization approaches in the analysis of population structure: sparse principal components ( SPC ) [24] and non-negative matrix factorization [23] . SPC , implemented in the R function SPC in the R package PMA , computes sparse PCs by solving a penalized matrix factorization problem with an penalty ( a penalty on the sum of the absolute values of the factor loadings ) to encourage sparsity . The algorithm is greedy in that it computes the factors one at a time , each time removing the effect of the previous factors from the original matrix . The user can choose whether to require the factors to be orthogonal; in our experiments we did not require orthogonality . SPC has a user-defined tuning parameter that controls the level of sparsity . We found that , with careful choice of this parameter , we were able to get SPC to produce results similar to PCA when the data are continuous , and closer to an admixture-based model when the data are from discrete groups . In particular , the main difference from SFA was on the data from two independent 2-D habitats . where SPC did not model the two habitats in separate factors . ( We were unable to apply SPC to the larger European and Indian data sets , due to limitations of . ) As its name suggests , non-negative matrix factorization ( NMF ) [23] , [34] constrains the factors and loadings to have non-negative values . For data sets considered here , we found that NMF typically produced results similar to SFA . However , NMF is less flexible than SFA in that it effectively requires the input matrix to be non-negative . In the genetic context this is not a big limitation as genotype data are most often encoded as non-negative integers ( , , ) , but even here it makes NMF slightly less flexible . For example , this means that NMF cannot be applied to genotype data that have been mean-centered , and there is no sensible way to include a mean term as in SFAm . As we have seen , in some settings incorporating a mean improves the interpretability of the results . The computational methods used to perform the matrix factorization for PCA , SFA , and admixture ( and also structure ) are practically quite different . In particular , the PCA factorization has a single global optimum that can be obtained analytically , and so multiple runs of PCA produce the same results . In contrast both admixture-based models and the SFA factorizations can have multiple local optima , and the computational algorithms used can produce different results depending on their starting point . In practice , in simple cases ( e . g . , involving a moderate number of discrete populations ) , both algorithms appear to produce consistent results across runs . In more complex situations we have found more variability in the results , particularly when the number of factors is large . In some cases there appear to be identifiability issues: for example , in the European data , multiple runs of SFAm produce loadings that are rotations of one another . Another qualitative difference between the three methods is that PCA produces consistent results as more factors are added , whereas admixture-based methods and SFA may produce qualitatively different results with different numbers of factors . Although consistency may seem a desirable property , there can be benefits to the different perspectives obtained by using different numbers of factors , as we illustrated in the results . To further contrast these two behaviors , consider the application of these methods to data from a continuous 1-D habitat . As noted previously [30] , the first PCA loading ( after removing the mean ) roughly captures position within the habitat , whereas subsequent loadings are sinusoidal functions of increasing frequency . In contrast , when SFA or admixture are run with an increasing number of factors , they redistribute their factors along the line so that each factor represents the average allele frequencies of an increasingly local region . ( If too many factors are used , there is not enough signal in the data to differentiate populations on small neighboring segments , and the results become unreliable . ) Although the additional factors in each case are qualitatively very different , they simply reflect different ways to capture finer-scale structure in the data . Which of these behaviors is preferable may be context-dependent , but understanding these differences is certainly helpful in interpreting the results of a data analysis . Although we have focused on the different constraints imposed by different matrix factorization methods , they also differ in another way: their assumed error distribution . In particular , admixture-based models assume a binomial error , whereas PCA is based on a least-squares criterion , which can be interpreted as a Gaussian error , and our SFA explicitly assumes Gaussian error . The binomial error may be more appropriate for data from an admixed population , but in general it is less flexible than the Gaussian model because the binomial variance is determined by the mean , rather than being a free parameter . It seems possible that this partly explains the convergence problems we observed in admixture for the 2-D habitat , in which case it may be worth adapting the admixture model to assume a Gaussian error . We note that there are several existing approaches to sparse factor analysis besides the novel approach that we introduce here [19]–[21] , [35] . Although these methods have similar motivations , they differ in several respects , and we have found that these differences can substantially impact results ( not shown ) . One advantage of our approach is its computational speed . Another feature of our approach is its lack of manually-tunable parameters ( other than the number of factors ) . This , of course , is a double-edged sword , since on the one hand , it makes the method easy to apply , but on the other hand , reduces flexibility . In practice , as our results show , our approach is sufficiently flexible to deal with a range of contexts involving different levels of sparsity . Our approach to SFA may also be useful in other contexts ( e . g . , gene expression data [22] , [35] or collaborative filtering [36] ) . In some cases , particularly when the data do not exhibit much sparsity , it may be desirable to extend our method in various ways . For example , as we have implemented it here , SFA encourages sparsity only on the loadings , and in some contexts it may be desirable to encourage sparsity on both the factors and the loadings ( as in the general penalized matrix decomposition method [24] ) . This could be achieved by putting an ARD prior on the elements of , and applying an analog of our ECME algorithm . It may also be fruitful to consider ways to increase the sparsity in the loadings , since in some other contexts we have found that the ARD prior we use can be generous in its use of non-zero loadings . Finally , although we have argued that in the context of population structure that applying methods with different numbers of factors may yield more insight than selecting a single “correct” number of factors , this may not be equally true in all contexts . In particular , the population structure case is complicated by the fact that the factors are often highly correlated with one another ( e . g . , because they often represent allele frequencies in closely-related populations ) ; in settings where factors are less correlated it may be more helpful to consider methods for automatically selecting the number factors ( e . g . , [37] ) .
We simulated genotypes from 1-D and 2-D habitats using the program ms [38] , using stepping-stone models similar to [30] . In the 1-D model we assumed demes along a line and allowing a high level of migration ( ) between adjacent demes . This migration rate produced an of between the two demes at either end of the line , which enables the two most extreme demes to be easily separable with SNPs . We sampled one diploid individual ( two independent haplotypes ) from each deme at independent SNPs . For the 2-D simulations , we assumed demes arranged in a by square grid , with migration parameters between neighboring demes . We then sampled one diploid individual from each deme at independent SNPs . For the two 2-D habitat simulations , we simulated two independent sets of demes and sampled a single individual from each deme at independent SNPs . For both the simulated and the real genotype data , we encoded each genotype ( AA , AB , or BB ) as , or . We used the POPRES European data set from [32] , and processed the data as in [14] . The POPRES data set was obtained from dbGaP at http://www . ncbi . nlm . nih . gov/projects/gap/cgi- bin/study . cgi ? study_id=phs000145 . v1 . p1 through dbGaP accession number phs000145 . v1 . p1 . This data included individuals , each of whom identify all four grandparents as being from a particular European country , genotyped at SNPs , and pruned down to SNPs after removing one of any pair of SNPs that had an [14] . Since our SFA method does not currently deal with missing data , we imputed missing genotypes using impute2 [39] . We imputed each chromosome by intervals of Mb , starting at position , with a buffer of size Mb on either side of the interval . We set the number of burn-in iterations to and the number of MCMC iterations to . We set the effective population size of the European sample to be , and we used the combined linkage maps from build , release ( downloaded from the impute website ) . We used these imputed genotypes as input to all three methods to facilitate fair comparisons . We used the Indian genotype data from [2] . The original data includes individuals from groups; we removed the groups that appeared to be genetic outliers as described in the original paper ( Sahariya , Nysha , Aonaga , Siddi , Great Andamanese , Hallaki , Santhal , Kharia , Onge , and Chenchu ) , leaving groups and individuals with genotyped SNPs . We imputed missing genotypes using impute2 as above , but with an effective population size of , and used these imputed genotypes as input to all three methods . After imputation , we pruned the data down to SNPs by removing one of any pair of SNPs that had an , and removing SNPs that had a minor allele frequency less than . Let be the number of individuals in a sample and be the number of genotypes . Represent each allele at a locus as a number ( e . g . , for SNPs from a diploid organism , as in our results above , represent as , as , and as ) . Our factor analysis model with factors can be written as: ( 3 ) or , equivalently , ( 4 ) where is an data matrix , is a -vector of column-specific means , is the matrix of factor loadings , is the matrix of factors , and is an matrix with each element independently distributed . We put a gamma prior on the inverse residual variance that acts as a regularizer: , which has mean and variance . In practice , we set and . This model , with a mean term , is referred to as SFAm in the main text; the SFA model is obtained by fixing the vector at zero . The ECME algorithm for fitting SFAm is described below; the ECME algorithm for fitting SFA is obtained by simply setting throughout . Note that here we have chosen to have column-specific ( i . e . , SNP-specific ) means and row-specific ( i . e . , individual-specific ) variances . It is possible to modify the ECME updates below to allow for different assumptions , for example to allow row-specific means or column-specific variances . In some contexts , including the population structure problem considered here , it might make sense to allow more general assumptions , such as variance terms on both the rows and columns of the matrix; indeed these options are implemented in the SFA software , although not investigated here . To induce sparsity in the factor loadings , we use an automatic relevance determination ( ARD ) prior [40] . Specifically , we assume , where the matrix is a parameter that we estimate , together with the other parameters , using maximum likelihood . If the estimate of , this implies that , thus inducing sparsity . Integrating out , the rows of are conditionally independent given the other parameters , with: ( 5 ) where ( a diagonal matrix with the -vector on the diagonal ) , and . Thus the log marginal likelihood for the parameters is: ( 6 ) ( 7 ) where . We fit this model using an expectation conditional maximization either ( ECME ) algorithm [41] to maximize . This algorithm is similar to an EM algorithm , but each maximization step maximizes either the expected log likelihood , or the marginal log likelihood , for a subset of the parameters conditional on the others . Specifically , the updates to , , and involve maximizing the expected log likelihood ( with the expectation taken over ) , whereas the updates to directly maximize the log marginal likelihood . To compute the expected log likelihood requires the first and second moments of the factor loadings . The data and the loadings are jointly normal ( as in , e . g . , [42] ) : ( 8 ) where is a -vector of zeros . Standard results for joint Gaussian distributions give the conditional expectation for : ( 9 ) where . Similarly , the conditional second moment is given by: ( 10 ) The updates for , , and involve maximizing the expected complete data log likelihood , , which from Equation 4 , and including the prior distribution on , is given by: ( 11 ) where ( 12 ) Taking the derivative of with respect to and setting to , we get the update for : ( 13 ) ( 14 ) In these expressions , and in what follows , we are assuming element-wise multiplication when a scalar multiplies a vector or a matrix . Taking the derivative of with respect to and setting to zero , we get the update for : ( 15 ) Taking the derivative of with respect to and setting to zero , we get the update for : ( 16 ) To update we can use the result from [40] to obtain the values of that maximize the log marginal likelihood with fixed values of , , and : ( 17 ) where and , where and . Note that when and otherwise . This works because , given , the SFA model ( Equation 3 ) is essentially the sparse regression model considered in [40] with playing the role of the covariates . Note that and are non-identifiable in that multiplying the row of by a constant and dividing the column of by will not change the likelihood ( Equation 6 ) . To deal with this we impose an identifiability constraint , for , where . Specifically , after each iteration we divide every element of by its standard deviation , and multiply the column of by . Because we choose not to update the expected values of the loading matrix between the CM steps , monotone convergence of the log marginal likelihood is not guaranteed , although in practice it appears to converge well . We find that convergence is reached for the applications described here after fewer than iterations . For each genotype data set , we run SFA multiple times with random seeds , setting the number of factors as described in the text; results presented in figures are a representative example . A C++ package containing the SFA and SFAm code is available for download at http://stephenslab . uchicago . edu/software . html . For smaller data sets ( all but the European and Indian data ) , we computed principal components by first standardizing the columns of the matrix ( subtracting their mean and dividing by their standard deviation ) and then finding the eigenvectors of the covariance matrix of the individuals in R [43] using the function eigen . In our terminology , these eigenvectors , or principal components ( PCs ) , are the loadings , i . e . , the columns of . For larger data sets , we identify the PCs using the SmartPCA software from the EigenSoft v package [7] , [13] . For both the European genotype data and the Indian genotype data , we set the number of output vectors to , we use the default normalization style , we do not identify outliers , we have no missing data , and we remove all chromosome data . We ran admixture v [11] with multiple random starting points using the -s option . We mapped the four-dimensional admixture proportions into two-dimensions for visualization as follows: the four-dimensional vector maps to the two-dimensional vector .
|
Two different approaches have become widely used in the analysis of population structure: admixture-based models and principal components analysis ( PCA ) . In admixture-based models each individual is assumed to have inherited some proportion of its ancestry from one of several distinct populations . PCA projects the individuals into a low-dimensional subspace . On the face of it , these methods seem to have little in common . Here we show how in fact both of these methods can be viewed within a single unifying framework . This viewpoint should help practitioners to better interpret and contrast the results from these methods in real data applications . It also provides a springboard to the development of novel approaches to this problem . We introduce one such novel approach , based on sparse factor analysis , which has elements in common with both admixture-based models and PCA . As we illustrate here , in some settings sparse factor analysis may provide more interpretable results than either admixture-based models or PCA .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"computational",
"biology/population",
"genetics",
"mathematics/statistics",
"genetics",
"and",
"genomics/population",
"genetics"
] |
2010
|
Analysis of Population Structure: A Unifying Framework and Novel Methods Based on Sparse Factor Analysis
|
Yersinia delivers Yops into numerous types of cultured cells , but predominantly into professional phagocytes and B cells during animal infection . The basis for this cellular tropism during animal infection is not understood . This work demonstrates that efficient and specific Yop translocation into phagocytes by Yersinia pseudotuberculosis ( Yptb ) is a multi-factorial process requiring several adhesins and host complement . When WT Yptb or a multiple adhesin mutant strain , ΔailΔinvΔyadA , colonized tissues to comparable levels , ΔailΔinvΔyadA translocated Yops into significantly fewer cells , demonstrating that these adhesins are critical for translocation into high numbers of cells . However , phagocytes were still selectively targeted for translocation , indicating that other bacterial and/or host factors contribute to this function . Complement depletion showed that complement-restricted infection by ΔailΔinvΔyadA but not WT , indicating that adhesins disarm complement in mice either by prevention of opsonophagocytosis or by suppressing production of pro-inflammatory cytokines . Furthermore , in the absence of the three adhesins and complement , the spectrum of cells targeted for translocation was significantly altered , indicating that Yersinia adhesins and complement direct Yop translocation into neutrophils during animal infection . In summary , these findings demonstrate that in infected tissues , Yersinia uses adhesins both to disarm complement-dependent killing and to efficiently translocate Yops into phagocytes .
Translocation of effectors via a type III secretion system ( TTSS ) is an essential process used by many gram-negative bacterial pathogens to thwart immune defenses during infection [1] . Upon mammalian infection , the three pathogenic Yersinia spp . , Yptb , Y . enterocolitica and Y . pestis deliver 5–6 Yop effectors into cells of the innate immune system [2]–[4] . Most Yops target and disrupt functions of macrophages , neutrophils and dendritic cells [5]–[8] . While Yop delivery is crucial for the virulence of Yersinia spp . [9] , the molecular interactions driving Yop translocation into innate immune cells during tissue infection are not understood . Bacterial attachment to host cells is necessary for TTSS-mediated delivery of effector proteins , and several studies have suggested a role for adhesins in attachment and TTSS mediated delivery by Enteropathogenic Escherichia coli , Pseudomonas and Yersinia spp . into host cells [5] , [10]–[16] . Yersinia expresses numerous adhesins , including Ail , Invasin and YadA , all of which can promote Yop translocation into cultured cells [11] , [16] . Invasin and YadA are expressed by the two enteric Yersiniae , are important for colonization , and contribute to dissemination following oral inoculation [17]–[19] . Ail is expressed by all three pathogenic Yersinia spp . and facilitates Yop delivery by Y . pestis into human epithelial and monocytic cell lines [16] , [20] . Ail and YadA are also implicated in conferring serum resistance [21] , [22] , and Ail , Invasin and YadA promote invasion into cultured cells [23]–[25] . However , while informative , cell culture and in vitro systems do not fully recapitulate the interactions between Yersinia and host cells during the course of infection . For example , while deleting Invasin and YadA is sufficient to abrogate Yop translocation in cell culture models [11] , a ΔinvΔyadA mutant still translocates effectors into freshly isolated splenocytes [3] and is still virulent in murine infection [26] . Thus , while many of the molecular mechanisms of adhesin functions have been well characterized in cell culture , their roles in Yop translocation and serum resistance during animal infection have not been established . The presence of multiple adhesins suggests at least four scenarios for their role in Yptb pathogenesis . First , expression of specific adhesins may be important at distinct stages of infection . It is established that invasin is necessary for survival in the GI tract and in penetrating the Peyer's patches , but is dispensable for establishing systemic infection [27] , [28] . Second , expression of certain adhesins may influence the ability of Yptb to disseminate to distinct tissues . In fact , YadA expression contributes to colonization of the lungs following intravenous ( IV ) infection with Yptb [29] . Third , functional redundancy may exist between adhesins , such that eliminating one adhesin may have minimal or no effect on virulence and/or translocation . For example , in the absence of both Invasin and YadA , other Yersinia factors enable penetration of Yptb across the intestinal epithelium [18] , [30] . Fourth , some adhesins may have roles unrelated to cell binding . For instance Ail and/or YadA may function to resist killing by serum during tissue infection [21] , [22] . Therefore , expression of multiple adhesins may contribute to Yptb survival in distinct host niches; some may direct Yop delivery into cells during infection while others may have different and/or additional roles . Host-encoded factors also play a part in Yop translocation in both cell culture systems and tissue infections . Activation of host signaling pathways , triggered by binding of Invasin or YadA to β1-integrin , promotes efficient Yop translocation into epithelial cell lines [4] , [11] . Antibody or complement opsonization of Yersinia is sufficient to drive Yop delivery into phagocytes grown in culture [5] , [31] . Moreover , serum factors , such as albumin , can activate TTSS-mediated secretion in Yersinia [32] , indicating that crosstalk between Yersinia and the host leads to proper engagement of the TTSS machinery . Finally , Yop delivery is severely reduced during infection of mice depleted of Gr-1+ cells , which include neutrophils and inflammatory monocytes [3] , suggesting that Yptb recognizes and mediates Yop translocation in response to a specific cellular environment . Complement is an integral part of the innate immune system and is composed of more than 30 proteins present in serum , tissue fluids and on cell surfaces [33] . By mediating bacteriolysis , opsonization and inflammation , complement acts as a rapid and efficient surveillance system against invading pathogens . Many bacterial pathogens have evolved mechanisms for evading the action of complement by sequestering or degrading complement proteins [34] . In vitro , Yersinia adhesins Ail and YadA mediate complement resistance by binding complement proteins and rendering them ineffective [21] , [22] , [35] . This binding prevents killing by the membrane attack complex ( MAC ) ; however , mice lack the C5 convertase and thus are unable to kill pathogens via MAC [36] , [37] . Binding of complement proteins to Yptb may also bridge contact of Yersinia with host cells , facilitating Yop translocation . Whether host opsonins participate in Yptb binding to and/or Yop delivery into host cells during animal infection remains to be determined . In this work , we examined the contribution of Yptb adhesins and host serum factors to the specific interactions between Yptb and immune cells leading to Yop translocation . We find the adhesins Ail , Invasin and YadA , function to direct translocation of Yops in infected tissues . We demonstrate that complement plays a role in limiting infection by ΔailΔinvΔyadA indicating that these adhesins also thwart complement-dependent killing . Finally , we show that complement , together with Yptb adhesins , promotes Yop translocation into professional phagocytes during animal infection . This illustrates that bacterial and host factors act in concert for efficient Yop delivery into targeted host cells during Yptb infection .
Three well-characterized Yersinia adhesins , Ail , Invasin and YadA [38]–[40] , were assessed for their contribution to translocation of Yops into isolated splenocytes . Single , double and triple adhesin mutants were constructed in three different Yptb strains: IP2666 , IP32953 and YPIII all of which are virulent in mouse infection models [3] , [7] , [41] , yet differentially express invasin and yadA ( Fig . S1A; [42] , [43] ) . In order to measure Yop translocation all strains were engineered to express a reporter protein , ETEM . ETEM is a recombinant protein consisting of the N-terminus of Yersinia effector YopE , which contains the information necessary for translocation through the TTSS , fused with TEM , a β-lacatamse [44] . TEM cleaves the membrane permeable dye CCF4-AM , changing its fluorescence from green to blue . Using this reporter , TTSS-dependent translocation is measured by quantifying the number of cells that fluoresce blue [3] , [45] . Splenocytes were infected with each ETEM-expressing strain at a multiplicity of infection ( MOI ) of 1∶1 loaded with CCF4-AM , and live cells were analyzed by flow cytometry for Yop translocation by measuring the number of Blue+ cells relative to WT ( Fig . 1A–1C and Fig . S2A–B ) . A ΔyopB strain that is deficient for TTSS-dependent translocation served as a negative control [46]; as expected , it did not translocate Yops into splenocytes ( Fig . 1A–1C and Fig . S2B ) . Deleting all three adhesins , Ail , Invasin and YadA , dramatically reduced translocation into splenocytes by all three Yptb strains ( Fig . 1A–1C and Fig . S2 ) . However , individual strains relied on each adhesin to varying degrees . Notably , translocation by IP2666 was almost exclusively dependent on YadA and Ail , with YadA playing a predominant role; the role of Ail was apparent only in the absence of YadA ( Fig . 1A and Fig . S2C ) . Complementing IP2666 ΔailΔyadA and ΔailΔinvΔyadA strains with either ail or yadA restored translocation to levels observed in the parental strain ( Fig . S2C and Fig . S3A ) . In IP32953 and YPIII , translocation was primarily dependent on Invasin , with YadA playing a modest role in the absence of Invasin ( Fig . 1B–1C ) . Complementation with a plasmid expressing Invasin restored the ability of Δinv and ΔailΔinvΔyadA strains to translocate Yops , in fact to levels greater than WT ( Fig . S3B ) . This may be attributed to greater Invasin expression in the complemented strains ( Fig . S1B ) . In summary , the combination of Ail , Invasin and YadA were necessary for Yop delivery into splenocytes in all three strains , but their individual contribution differed based on the strain background . The translocation defect in a Y . pestis Δail mutant can be overcome by longer incubation with HEp2 cells [20] . This indicates that either productive receptor-adhesin interactions that compensate for Ail require more time for Yop delivery or that other adhesins capable of inducing translocation become expressed as the infection time lengthens . Therefore , we investigated if increasing the time or MOI of infection could overcome the translocation defect in the IP2666 adhesin mutants . Neither infecting splenocytes at an MOI of 1∶1 for 4 hours or at an MOI of 20∶1 for 1 hour led to a relative increase in the translocation efficiency of the ΔailΔinvΔyadA strain ( Fig . 1D–1E ) . In fact , the percentage of Blue+ cells was unchanged for ΔyadA and ΔinvΔyadA strains but increased 5-fold following infection with a higher MOI of WT ( Fig . S2E ) , leading to a net decrease in the relative number of cells translocated by strains lacking YadA compared to WT ( Fig . 1A , 1D–1E ) . Collectively , these results indicate that the factor ( s ) responsible for the low levels of translocation in the ΔailΔinvΔyadA mutant are not enhanced by longer incubation times or higher MOIs . This suggests that the residual levels of translocation by this mutant could be due to adhesin-independent mechanisms [47] . A reduction in the overall number of cells injected with Yops should be reflected in a reduction in translocation into one or more specific splenic cell types . As expected , mutants where the overall levels of translocation were reduced compared to WT generally had a similar reduction in translocation into most cell types ( Fig . S4A–C ) . For example , IP2666ΔailΔyadA , IP32953ΔinvΔyadA and YPIIIΔinvΔyadA strains translocated ETEM into fewer numbers of all cell types analyzed ( Fig . S4B–C ) . A notable exception to this was that the IP32953Δinv , IP32953ΔailΔinv , YPIIIΔinv and YPIIIΔailΔinv strains injected Yops into comparable numbers of Ly6G+ cells as did their isogenic WT strains . However , these strains translocated Yops into fewer F4/80+ , CD11c+ and B220+ cells , which accounted for their overall lower levels of translocation ( Fig . S4B–C ) . Previous work has shown that neutrophils ( Ly6G+ ) , macrophages ( F4/80+ ) and dendritic cells ( CD11c+ ) are enriched in the cell population targeted for Yop delivery in both isolated splenocytes and during murine infection , compared to the abundance of these cell types in the whole splenocyte population [3] . Therefore , we examined whether the mutants translocated Yops into an altered spectrum of splenic cell types , as compared to WT . For this analysis , the percentage of a specific cell type in the Blue+ cell population for each mutant strain was compared to the percentage of that cell type in the Blue+ cell population from the isogenic WT strain ( Fig . 2 and Fig . S5 , grey bars ) . In strains where translocation levels were not significantly higher than ΔyopB strains ( i . e . IP32953ΔailΔinvΔyadA and YPIIIΔailΔinvΔyadA , Fig . 1B–C ) , no cell type analysis was performed . Strikingly , most strains with dramatically reduced translocation levels ( Fig . 1A–C ) exhibited an even greater enrichment of professional phagocytes in the translocated cell population than was observed for WT ( Fig . 2A–C ) . Furthermore , preferential targeting into Ly6G+ , F4/80+ and CD11c+ cells was still observed by most adhesin mutant strains compared to the levels of these cells in the total splenocyte population ( Fig . 2A–C gray bars versus white bar ) . The exceptions were the IP32953ΔyadA and ΔailΔyadA mutants , which did not translocate Yops into a greater percentage of Ly6G+ cells than found in the total splenocyte population ( Fig . 2B ) . In addition , these two mutants consistently targeted fewer CD11c+ cells than did WT IP32953 ( Fig . 2B ) , but this difference was not significant . Combined these results suggest that YadA promotes interactions with neutrophils and possibly dendritic cells for IP32953 . With this exception , however , individual adhesins did not appear to play pivotal roles in directing Yop injection into professional phagocytes . Furthermore , phagocytes were generally over-enriched in the Blue+ population after infection with mutants that translocate Yops infrequently , suggesting that other factors contribute to the selective injection into these cells . Since binding to mammalian cells is a necessary step for Yop translocation [5] , [10] , [11] and as a number of adhesin mutants translocated Yops at low levels , these strains were evaluated for their ability to associate with splenocytes . GFP+ IP2666 Yptb adhesin mutants were incubated with splenocytes and specific cell types associated with bacteria were measured by flow cytometry . Surprisingly , the absence of YadA consistently led to an overall increase in Yptb association with splenocytes ( Fig . 3A ) . Therefore , the defect in translocation of these strains ( Fig . 1A ) is not due to an inability to associate with splenocytes . Next , the percentage of each cell type associated with the adhesin mutants was compared with those associated with WT . Interestingly , strains lacking YadA associated with almost twice as many B220+CD19+ , CD4+TCRβ+ and CD8α+TCRβ+ cells as compared to WT ( Fig . 3B ) . Because CD4+TCRβ+ , CD8α+TCRβ+ and B220+CD19+ cells make up a greater percentage of the total splenocyte population than Ly6G+ , F4/80+ cells and CD11c+ cells ( Fig . S6 ) , this observation accounts for the higher overall association of ΔyadA strains with splenocytes . Fewer F4/80+ cells associated with ΔailΔyadA and ΔailΔinvΔyadA strains ( Fig . 3C ) ; however , since these cells are a minority of the overall splenocyte population ( Fig . S6 ) , this reduction was not great enough to offset the increased association observed with B and T cells ( Fig . 3A–C ) . Combined , these results indicate that the defect in translocation is not a result of a defect in cell association . In fact , the yadA mutants bound to greater numbers of lymphocytes but this binding was not sufficient to drive translocation into a greater number of cells . These results are consistent with previous findings demonstrating that activation of host-cell signal transduction pathways by tight adhesin-receptor binding enhances translocation [11] . Since IP2666 ΔailΔinvΔyadA failed to translocate Yops into isolated splenocytes , we postulated that these adhesins may contribute to efficiency of translocation during animal infection . However , in order to compare the translocation ability of these two strains during animal infection , they must colonize the tissue to similar levels . Due to the deficiency of ΔailΔinvΔyadA in translocating Yops to isolated splenocytes ( Fig . 1A ) and its sensitivity to killing by bovine serum ( data not shown ) , we predicted that ΔailΔinvΔyadA would be attenuated in murine infection . To assess its relative virulence compared to WT , C57BL/6 mice were infected IV with 800 ( 1X ) colony forming units ( CFU ) of WT-ETEM , 800 ( 1X ) CFU of ΔailΔinvΔyadA-ETEM , 30 , 000 ( 37 . 5X ) CFU of ΔailΔinvΔyadA-ETEM or 30 , 000 ( 37 . 5X ) CFU of ΔyopB-ETEM and monitored for 15 days . All mice challenged with WT-ETEM succumbed to infection by the end of day 4 , whereas 90% of mice challenged with 800 CFU of ΔailΔinvΔyadA survived , indicating that ΔailΔinvΔyadA was less virulent than WT ( Fig . 4A ) . However , in contrast to ΔyopB , the ΔailΔinvΔyadA-ETEM strain caused a rapid , lethal infection when mice were challenged with 30 , 000 CFU . The observation that ΔailΔinvΔyadA was more virulent than ΔyopB suggested that the ΔailΔinvΔyadA mutant retained some ability to translocate Yops during animal infection . We next investigated whether ΔailΔinvΔyadA delivered Yops into as many cells as WT when tissues were infected at comparable levels . To achieve comparable CFU in the spleen at 4 days post-infection , mice were infected with 800 CFU of WT-ETEM or 30 , 000 CFU of ΔailΔinvΔyadA-ETEM . Under these conditions , the onset of infection generally occurred in both groups of mice at the same time . To determine whether there was a difference in the levels of Yop translocation into cells , we compared the percentage of Blue+ cells versus the CFU recovered for the two strains ( Fig . 4B ) . Strikingly , the ΔailΔinvΔyadA translocated Yops into significantly fewer cells than WT at comparable CFU at 4 days post-infection ( Fig . 4B ) , as linear regression indicated a significant difference in the Y intercept ( P<0 . 0001 ) . This result demonstrates that one role of these adhesins in animal infections is to direct translocation of Yops into immune cells . Next , the spectrum of host immune cells translocated into by the ΔailΔinvΔyadA mutant was compared to those cells translocated into by WT to determine whether adhesins direct Yops into specific cell types during animal infection . Interestingly , phagocyte enrichment in the Blue+ population was retained by the ΔailΔinvΔyadA-ETEM strain ( Fig . 4C white bars versus gray bars ) . In addition , ΔailΔinvΔyadA injected Yops into relatively more B cells than WT ( Fig . 4C ) . Taken together , these results revealed that although fewer cells are translocated by ΔailΔinvΔyadA-ETEM than by WT , professional phagocytes are still targeted for translocation . Therefore , adhesins contribute to the total numbers of cells injected with Yops , but other factors must contribute to the preferential translocation into professional phagocytes in spleens . Opsonization of Yersinia by antibody or complement is sufficient to mediate binding and Yop delivery to phagocytes [5] , [31] . Therefore , we tested whether bovine serum contributed to the total number or spectrum of cells of isolated splenocytes translocated into by Yptb . Splenocytes were infected at an MOI of 1∶1 with WT-ETEM in the presence of heat-inactivated Fetal Bovine Serum ( HIS ) or in serum free media ( SFM ) for 1 h . The bacteria and splenocytes survived equally well in the presence and absence of serum during the course of this assay ( data not shown ) . Interestingly , infection in SFM led to a significant increase in the number of Blue+ cells ( Fig . 5A ) , suggesting that a heat-resistant factor in serum restricts overall levels of Yop translocation . To probe if serum contributes to the distribution of cell types targeted for Yop translocation , cell type distribution in the Blue+ population was compared between HIS versus SFM infections . Cell type distribution analysis was performed at comparable MOI ( 1∶1 ) for both conditions , as well as at an MOI of 0 . 2∶1 in SFM which results in similar overall numbers of Blue+ cells compared to infection in HIS at an MOI of 1∶1 ( Fig . 5A ) . SFM infections at an MOI of 0 . 2∶1 led to significantly fewer translocated phagocytes , whereas at an MOI of 1∶1 a similar number of translocated phagocytes was obtained as compared to HIS infections at an MOI of 1∶1 ( Fig . 5B ) . In contrast , at an MOI of 1∶1 , significantly more B220+ , CD4+ and CD8α+ cells were translocated into SFM than in HIS . Because CD4+ , CD8α+ and B220+ cells make up a greater percentage of the total splenocyte population ( Fig . S6 ) , this observation accounts for the higher overall translocation levels in SFM at an MOI of 1∶1 ( Fig . 5B ) . There were also changes in the distribution of cell types targeted for translocation among these conditions ( Fig . 5C ) . Specifically , phagocytes were enriched and T cells were underrepresented in the Blue+ population of HIS-infected splenocytes compared to SFM ( Fig . 5C ) . Therefore , bovine serum components curb overall levels of Yop translocation and contribute to the selective targeting of phagocytes by Yptb . Since WT Yptb translocated Yops into more cells in SFM , we examined whether ΔailΔinvΔyadA-ETEM also translocated Yops more efficiently in SFM . However , ΔailΔinvΔyadA-ETEM remained defective for translocation in SFM ( Fig . 5D ) , indicating that Ail , Invasin and/or YadA provide an essential interaction with host cells that drives translocation . Both YadA and Ail bind to several extracellular proteins that facilitate contact with host cells [16] , [25] . Two of these , fibronectin ( Fn ) and bovine serum albumin ( BSA ) , were tested to determine if they modulated the number of and/or cell types translocated with Yops . A comparable number of splenocytes was translocated with Yops in SFM supplemented with Fn or BSA , indicating that neither Fn or BSA serve as the translocation-inhibiting factor in serum ( Fig . 5A ) . Furthermore , infection in SFM supplemented with Fn or BSA led to reduced translocation of professional phagocytes ( data not shown ) , as observed in SFM ( Fig . 5B–5C ) suggesting that these proteins do not alter the interaction of Yptb with phagocytes . The increase in Yptb translocation efficiency in SFM could be driven by an increase in Yptb binding to host cells . To test this hypothesis , splenocytes were incubated with GFP+ WT , ΔailΔinvΔyadA or ΔyopB strains in HIS or in SFM at an MOI of 0 . 5∶1 . The overall association of GFP-expressing Yptb strains with splenocytes was unchanged between HIS versus SFM ( Fig . 5E ) . Furthermore , no difference in association with any of the cell types tested was observed in HIS versus SFM ( Fig . 5F ) . Therefore , the changes in levels and cell types injected with Yops were not reflected in changes in association with splenocytes . Taken together , these results indicate that factors in serum mask interactions between Yptb and T cells that otherwise would result in translocation and enhance interactions between Yptb and professional phagocytes resulting in increased translocation . While heat inactivation of serum eliminates the ability of complement to form MAC and lyse bacteria [48] , complement components still present in serum could bind adhesins and direct Yptb to specific receptors on professional phagocytes . To determine whether complement components play a role in controlling infection and/or directing Yptb to translocate Yops into specific cell types , we depleted complement by injecting cobra venom factor ( CVF ) intraperitoneally into mice [49] . CVF is a complement-activating protein in cobra venom that forms an active convertase complex [50] that is resistant to inactivation by host complement regulatory proteins . As a result of its constitutive activity , the CVF-convertase depletes serum and tissues of complement components . Following complement depletion , mice were infected IV with 800 ( 1X ) CFU of WT-ETEM , 800 ( 1X ) CFU of ΔailΔinvΔyadA-ETEM , or 30 , 000 ( 37 . 5X ) CFU of ΔailΔinvΔyadA-ETEM and monitored for 15 days . Notably , CVF treatment restored the virulence defect of the ΔailΔinvΔyadA strain but had no effect on the virulence of WT ( Fig . 4A and 6A ) . Consistent with this observation , CVF treatment greatly enhanced the number of ΔailΔinvΔyadA bacteria present in spleens but not WT ( Fig . 6B ) . This was also consistent with our observation that in the absence of complement , the ΔailΔinvΔyadA infected mice generally displayed signs of illness , such as a hunched appearance and slower walk , a few hours before mice infected with WT . Combined , these results indicate that complement controls growth of the ΔailΔinvΔyadA strain but not of WT . Therefore Ail , Invasin and/or YadA must be counteracting the action of complement during mouse infection . To investigate whether complement-depletion altered the number of cells injected with Yops , CVF-treated mice were infected IV with 800 CFU of WT-ETEM or 800 CFU of ΔailΔinvΔyadA-ETEM . Four days post infection , the number of injected cells under these two infection conditions was also compared to non-complement depleted mice infected with 800 CFU of WT-ETEM ( Fig . 4B and solid black line in Fig . 6C ) . No difference in the levels of Blue+ cells was observed in mice infected with WT-ETEM , with or without complement depletion , when similar CFUs were recovered from the spleen ( Fig . 6C , solid black line versus solid grey line ) . In contrast , once the ΔailΔinvΔyadA-ETEM strain reached levels of 5 . 5×105 in the absence of complement , it delivered Yops to significantly more cells than did WT ( Fig . 6C dashed lines , versus solid lines ) . However , significantly more ΔailΔinvΔyadA bacteria were recovered in these tissues , which could explain the increased number of cells injected with Yops ( Fig . 6B ) . Therefore , we compared the number of cells injected by ΔailΔinvΔyadA and WT when comparable numbers of bacteria were recovered ( Fig . S7A–B ) . Again , under these conditions , the ΔailΔinvΔyadA strain in CVF-treated mice injected more cells than did WT . In summary , this analysis illustrates that in the absence of complement , ΔailΔinvΔyadA translocates Yops into more cells than WT , indicating that complement components affect both bacterial survival and Yop translocation . Furthermore , the significant increase in growth and efficiency of translocation of the ΔailΔinvΔyadA mutant under complement-depleted conditions suggest that this mutant behaves quite differently than WT and/or is in a distinctly different microenvironment . We next examined whether complement depletion affected the spectrum of cells translocated with Yops by either WT or ΔailΔinvΔyadA . No difference was observed in the cell types targeted for Yop translocation by WT-ETEM in the presence or absence of complement ( Fig . 6D ) . Interestingly , in the absence of complement , the ΔailΔinvΔyadA-ETEM strain exhibited significantly decreased translocation into Ly6G+ cells and increased translocation into B220+ cells , as compared to WT and WT+CVF ( Fig . 6D and Fig . S7C ) . These results combined with those in Fig . 4C suggest that complement components direct translocation of Yops from ΔailΔinvΔyadA to Ly6G+ cells .
In order to mount a successful infection , Yersinia must counteract a multitude of host immune responses that are enlisted to control an invading pathogen . One essential mechanism Yersinia spp . use to thwart immune responses is to deliver Yop effector proteins into host cells via a TTSS [8] . These Yops quickly disrupt signal transduction pathways that are normally geared to respond to invading threats and thus allow Yersinia to successfully colonize and persist in the host [8] , [9] . This work demonstrates that both bacterial adhesins and host factors contribute to the efficiency of translocation and the specificity of immune cell types translocated with Yops . Furthermore , adhesins contribute to resistance of host complement . In mouse infection and in isolated splenocytes , a ΔailΔinvΔyadA strain translocated Yops into significantly fewer cells than WT . These results indicate that the three adhesins are important for the vast majority of translocation that occurs during infection . Thus , one critical feature of Yersinia adhesins is to direct translocation of Yops in infected tissues ( Fig . 7A ) . However , the low yet detectable level of translocation by the ΔailΔinvΔyadA mutant in vivo indicates that other factors can also drive translocation in the absence of these adhesins ( Fig . 7B ) . These factors may include additional adhesins and/or host factors . For example , growth conditions can dictate the expression of major virulence factors [43] , [51] . Growth in human plasma , which may mimic a bloodstream environment , positively regulates levels of YadA but negatively regulates pH 6 antigen in Yptb [51] . Some bacterial factors capable of driving translocation may be expressed poorly or function inefficiently during in vitro culture conditions , but may contribute to translocation during animal infection . Such potential factors include , ifp and invC , which are expressed by Yptb in the mouse intestinal tract but not in vitro [52] . On the other hand , the low level of Yop translocation by ΔailΔinvΔyadA in vivo may be due to host factors that direct translocation by bridging the Yersinia outer surface to receptors on phagocytes . In fact , our results show that host factors also contributed to Yop delivery in isolated splenocytes and during mouse infection . In cell culture systems , serum promotes TTSS-mediated secretion by Pseudomonas , Shigella , Salmonella and Yersinia [32] , [53]–[55] . In addition , BSA , a major component of serum , activates TTSS-mediated secretion by Y . enterocolitica and it has been proposed that serum components such as BSA provide signals for TTSS activation by Yersinia [32] . Thus , one hypothesis is that Yersinia exposed to serum or BSA would translocate effectors at higher levels because it is already primed to secrete Yops . Surprisingly , bovine serum restricted overall translocation levels into isolated splenocytes , but enhanced translocation specifically into neutrophils , indicating that factors in serum modulate the efficiency and specificity of Yop translocation . To further investigate the role of serum components in directing Yop translocation during animal infection , we used CVF to deplete complement . Depletion of complement revealed several important facets of infection . Notably , we demonstrate that the adhesins Ail , Invasin and YadA function to ameliorate the actions of complement in animal infection , as shown by the difference in survival of the ΔailΔinvΔyadA strain in complement depleted versus non-depleted mice . Moreover , complement-depletion enhanced Yop translocation by ΔailΔinvΔyadA but not by WT . To our knowledge , this is the first report of a role for adhesins in counteracting the activities of complement during murine infection by Yersinia . Complement contributes to a rapid immune response during infection by promoting inflammation , mediating bacteriolysis by MAC formation , and providing oponsins for phagocytosis [33] , [34] . A Y . pestis Δail strain is sensitive to the killing action of complement from a wide variety of sources except mice [48] , and protection afforded by Ail is evident in rats where Y . pestis Δail is completely attenuated [56] , [57] . Combined , these observations suggest that the attenuation of the Y . pestis Δail strain in rats is due to killing by complement , although this was not demonstrated directly . Mouse complement lacks bactericidal activity due to a nonfunctional C5 convertase and does not kill a Y . pestis Δail strain [36] , [37] , [48] . However , other functions of complement , such as opsonization and promotion of inflammation , are retained . In mice a Y . pestis Δail mutant exhibits a delayed time to death [56] , suggesting that the mutant is partially sensitive to aspects of murine host defenses . Our data is consistent with the hypothesis that these activities may be mediated by complement . Why does depleting complement affect the growth of ΔailΔinvΔyadA but not WT Yptb ? Complement components may coat the mutant , rendering it susceptible to opsonophagocytosis ( Fig . 7 ) . This , combined with inefficient Yop translocation into neutrophils , may contribute to its elimination . In the absence of complement , innate immune cells may no longer efficiently bind the mutant , allowing for its enhanced growth . WT may be exempt from being opsonophagocytosed as it efficiently delivers Yops into innate immune cells , a number of which block phagocytosis [5] . In addition , WT may inhibit generation of pro-inflammatory chemoattractants C3a and C5a [21] , [22] , whereas these chemoattractants may be produced at higher levels during infection with ΔailΔinvΔyadA , resulting in increased neutrophil recruitment . It has been proposed that Ail stifles neutrophil recruitment by complement during Y . pestis infection [57] . Ail , however , does not inhibit neutrophil recruitment to Yptb because neutrophils surround WT Yptb microcolonies [58] , [59] . Complement may also elicit release of neutrophil extracellular traps ( NETs ) [60] , which may be involved in controlling Yptb infection . NETs are extracellular fibers composed of granule and nuclear constituents that disarm and kill extracellular bacterial pathogens [61] . By trapping bacteria , NET release limits systemic dissemination [60] , [62] . The complement-inhibitory action of Ail and/or YadA may inhibit NET release during WT infection , whereas a ΔailΔinvΔyadA mutant would be unable to inhibit NET release and thus becomes efficiently trapped and killed by innate immune cells . In the absence of complement , NETs are not released [60] , which may account for the growth of ΔailΔinvΔyadA as the mutant may no longer be trapped and killed by NETs . Complement-depletion also enhanced Yop translocation and altered the spectrum of cells targeted for translocation by the ΔailΔinvΔyadA strain but not by WT . In cell culture , opsonization of Yersinia by complement or antibody is sufficient to mediate binding and Yop delivery to macrophages and neutrophils [5] , [31] , but in mouse infection , complement had no effect on translocation by WT . We speculate that with a full repertoire of adhesins , WT Yptb may be fully equipped to bind cells and deliver Yops regardless of complement ( Fig . 7A ) . In the absence of Ail , Invasin and YadA , another Yptb adhesin and/or host factor may promote translocation ( Fig . 7A–B ) ; however , complement can now contain the infection and dampen or hinder the Yptb-host cell interactions required for efficient translocation ( Fig . 7B ) . In the absence of complement , this factor can mediate translocation by ΔailΔinvΔyadA ( Fig . 7C ) and direct injection of Yops into a different subset of immune cells . The observation that ΔailΔinvΔyadA is fully virulent in the absence of complement , despite the fact that it targets an altered spectrum of cells , could be due to several possibilities . First , while neutrophils are not targeted to the same degree as by WT , they are still targeted . This level of targeting combined with a presumable reduction in opsonophagocytosis in the absence of complement may be sufficient to permit growth of the mutant . Second , the absence of both complement and adhesins may permit the bacteria to reside within a privileged niche that it does not normally occupy . For instance , these bacteria may be internalized more frequently by a subset of macrophages than is normally observed for Yptb [63] , [64] and thus may be protected from neutrophils . Finally , in the absence of complement , WT may be vulnerable to other immune responses due to its expression of adhesins . For example , YadA enhances binding to NETs and thus its expression may increase the susceptibility of WT to bactericidal factors released in NETs [60] , [65] . In the absence of both YadA and complement , Yersinia would not become trapped as readily in these NETs . These scenarios are consistent with the idea that depending on the tissue environment , adhesins can either facilitate bacterial survival or elimination by host [65] . In summary , translocation and targeting of Yops into host cells during infection is a multi-factorial process whereby multiple Yersinia adhesins and host factors act together . Bacterial adhesins contribute to the total number of cells targeted for translocation , and Yersinia employs host complement to help drive Yop translocation into specific cell types . Future work will be aimed at identifying the complement component ( s ) required for this function , investigating the relative contribution of each of these adhesins to serum resistance and Yop translocation during infection , and identifying other adhesins that facilitate translocation during animal infection .
Strains ( Table S1 ) , primers ( Table S2 ) , plasmids , and strain construction are described in Methods S1 . Mice were infected as described [7] with the following modifications . 7–8 week old C57BL/6 ( NCI ) mice were infected IV with 100 µl of the strain and dose indicated in the legends for Figures 4 and 6 . Mice were either sacrificed on day 4 or monitored for up to 15 days until they displayed significant signs of illness . CVF ( Quidel Corporation ) was used to deplete complement as described [49] . Briefly , C57BL/6 mice were injected intraperitoneally with two 5 Unit doses of CVF 4 h apart . 24 hours later , mice were infected IV with Yptb . This study was carried out in accordance with the recommendations in the Guide for Care and Use of Laboratory Animals of the National Institutes of Health . The Institutional Animal Care and Use Committee of Tufts University approved all animal procedures . Our approved protocol number is B2012-54 . All efforts were made to minimize suffering; animals were monitored following infection and were euthanized upon exhibiting substantial signs of morbidity by CO2 asphyxiation followed by cervical dislocation . Spleens were harvested aseptically , treated with collagenase D ( Roche ) , and a cell suspension was generated as described [3] . Unless indicated in the figure legends , cell suspensions were resuspended in RPMI supplemented with 5% heat-inactivated fetal bovine serum ( HIS ) . Single cell suspensions were incubated for 30 minutes in the dark in media containing 1 µg/ml CCF4-AM ( Invitrogen ) , 1 . 5 mM probenecid ( Sigma ) and 100 µg/ml gentamicin . 100 µl of cells were aliquoted into a 96-well plate and incubated with 50 µl of FACS buffer ( PBS+1% HIS ) containing a 1∶200 dilution of Mouse BD Fc Block ( BD ) for 10 minutes at 4°C . Cells were incubated in 50 µl of FACS buffer containing fluorescent antibodies Ly6G-PE-Cy7 , B220-PeCy5 , CD19-Cy7 , CD4-PECy5 , CD8α-PECy5 , TCRβ-Cy7 ( BD Pharmingen ) , F4/80-PE-Cy5 and/or CD11c-PECy5 ( eBioscience ) at dilutions of 1∶75 for 30 minutes at 4°C , washed twice in FACS buffer , centrifuged at 340× g , resuspended in 200 µl in FACS buffer and analyzed on an LSRII ( Becton Dickson ) FACS machine . 5×104–4×105 cells were acquired per sample and data was analyzed using FlowJo v4 . 3 software . For strains that injected Yops at levels less than 30% of those observed for WT , 4×105 live cells were collected for analysis . Otherwise , 5×104 live cells were analyzed . Uninfected cells that were not incubated with CCF4-AM and/or antibodies were used as negative controls . Splenocyte adherence assays were performed as described previously [3] with the following modifications . Single cell suspensions were resuspended in RPMI without serum ( SFM ) or media supplemented with 5% HIS as indicated in the figure legends . Cells were infected with the indicated GFP-expressing strains at an MOI of 0 . 5∶1 for 20 minutes at 37°C . Cells were labeled with antibodies and analyzed by flow cytometry . For all figures , except Figures 3 , 4B , 6C and S7B–C , data was graphed and statistical analysis was performed using GraphPad PRISM software by applying One-Way Anova with Tukey's Multiple Comparison Test . For Figure 3 statistical analysis was performed by applying One-Way Anova with Dunnett's Multiple Comparison Test . For Figures 4B and 6C linear regression analysis was performed using SAS9 . 2 analytical software in collaboration with the Tufts Clinical and Translational Science Institute ( CTSI ) . For Figure S7B–C analysis was performed by applying Student's t test .
|
Many bacterial pathogens use a needle-like structure to deliver proteins into host cells to cause disease . Yersinia species use one such structure , called a type III secretion system , to deliver a set of 6–7 proteins , called Yops , into host cells . These Yops act to dismantle host defenses and establish infection . Bacterial adhesins and host factors have been suggested to promote proper delivery of Yops into specific mammalian cells . We identify three Yersinia pseudotuberculosis adhesins that significantly contribute to bacterial survival and efficient Yop delivery into host cells during animal infection . We also demonstrate that host serum factors in combination with Yersinia adhesins contribute to the number of cells that are injected with Yops and to the specific cell types targeted for injection . Our study illustrates that bacterial adhesins and host factors contribute to efficient delivery of effector proteins into targeted host cells during infection .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"infectious",
"diseases",
"genetics",
"immunology",
"biology",
"microbiology"
] |
2013
|
Adhesins and Host Serum Factors Drive Yop Translocation by Yersinia into Professional Phagocytes during Animal Infection
|
To subvert host defenses , Mycobacterium tuberculosis ( Mtb ) avoids being delivered to degradative phagolysosomes in macrophages by arresting the normal host process of phagosome maturation . Phagosome maturation arrest by Mtb involves multiple effectors and much remains unknown about this important aspect of Mtb pathogenesis . The SecA2 dependent protein export system is required for phagosome maturation arrest and consequently growth of Mtb in macrophages . To better understand the role of the SecA2 pathway in phagosome maturation arrest , we identified two effectors exported by SecA2 that contribute to this process: the phosphatase SapM and the kinase PknG . Then , utilizing the secA2 mutant of Mtb as a platform to study effector functions , we identified specific steps in phagosome maturation inhibited by SapM and/or PknG . By identifying a histidine residue that is essential for SapM phosphatase activity , we confirmed for the first time that the phosphatase activity of SapM is required for its effects on phagosome maturation in macrophages . We further demonstrated that SecA2 export of SapM and PknG contributes to the ability of Mtb to replicate in macrophages . Finally , we extended our understanding of the SecA2 pathway , SapM , and PknG by revealing that their contribution goes beyond preventing Mtb delivery to mature phagolysosomes and includes inhibiting Mtb delivery to autophagolysosomes . Together , our results revealed SapM and PknG to be two effectors exported by the SecA2 pathway of Mtb with distinct as well as cumulative effects on phagosome and autophagosome maturation . Our results further reveal that Mtb must have additional mechanisms of limiting acidification of the phagosome , beyond inhibiting recruitment of the V-ATPase proton pump to the phagosome , and they indicate differences between effects of Mtb on phagosome and autophagosome maturation .
In 2015 , 1 . 8 million deaths were attributed to infection with Mycobacterium tuberculosis ( Mtb ) , the causative agent of tuberculosis [1] . Mtb is an intracellular pathogen that subverts multiple antimicrobial mechanisms of the host in order to survive and replicate in macrophages [2] . To avoid trafficking to the antimicrobial environment of acidified phagolysosomes , Mtb blocks the normal series of phagosome maturation events that occurs following phagocytosis [2 , 3] . As a result , Mtb resides in phagosomes that resemble early endosomes in retaining Rab5 on their surface , avoiding host factors that drive downstream maturation events ( e . g . phosphatidylinositol-3-phosphate [PI3P] , Rab7 , and the vacuolar-H+-ATPase [V-ATPase] ) and failing to fuse with degradative lysosomes [4–7] . Notably , Mtb prevents phagosome recruitment and assembly of V-ATPase , a proton pump that acidifies the phagosome , which helps explain the failure of mycobacterial phagosomes to fully acidify [4 , 8 , 9] . Phagosome maturation is a complex multi-step process and there are multiple Mtb protein and lipid effectors that are thought to play a role in arresting phagosome maturation [10] . However , the specific function ( s ) of effectors and the interplay between effectors remains to be determined . It also remains unclear if all the effectors of this process are known . The gaps in our understanding are partly due to redundancy among effectors and the potential for effectors to have functions in other aspects of Mtb pathogenesis or physiology beyond phagosome maturation arrest [8 , 11–16] . These features of effectors make it difficult to study the contribution of individual effectors to phagosome maturation arrest using loss of function mutants . In addition to residing in phagosomes , intracellular Mtb can also localize to double membrane bound compartments known as autophagosomes . Autophagosomes progress through similar maturation stages as phagosomes and culminate in fusion with lysosomes to form degradative autophagolysosomes [17] . As with phagosomes , Mtb is able to arrest autophagosome maturation and prevent fusion with lysosomes [18 , 19] . However , unlike the process of phagosome maturation arrest , there has been very little study of Mtb mechanisms and effectors of autophagosome maturation arrest . Most of the reported effectors of Mtb phagosome maturation arrest are either exported to the bacterial cell wall or fully secreted [20] . In Mtb , the SecA2 protein export pathway is required for phagosome maturation arrest , which indicates that this pathway exports effectors required to inhibit phagosome maturation [21] . Unlike the paralogous SecA1 ATPase , which is responsible for the bulk of housekeeping export and is essential for bacterial viability , SecA2 is a non-essential specialized SecA ATPase required for exporting a relatively small subset of proteins [22–25] . Although not required for growth during in vitro broth culture , SecA2 is required for Mtb replication in macrophages and mice [23 , 26] . Unlike wild type Mtb , during macrophage infection , a secA2 mutant of Mtb is delivered to acidified mature phagosomes [21] . The failure of the secA2 mutant to arrest phagosome maturation is previously shown to be responsible for its intracellular growth defect [21] . We hypothesized that the role of the SecA2 pathway in phagosome maturation arrest is to export multiple effectors of the process . Here , we identify for the first time SapM , a secreted phosphatase previously reported to function in phagosome maturation arrest , as being exported by the Mtb SecA2 pathway [7 , 27] . We further show that the SecA2 dependent export of this protein contributes to both phagosome maturation arrest and intracellular growth of Mtb . By identifying a histidine residue that is essential for SapM phosphatase activity , we confirm that the phosphatase activity of SapM is required for its function . Along with SapM , our data indicates the existence of other SecA2-dependent effectors of phagosome maturation arrest and we identify the Mtb eukaryotic-like serine/threonine protein kinase PknG as one of these additional factors . By restoring export of SapM and PknG individually and in combination to the secA2 mutant , we provide unique insight into specific steps in phagosome maturation arrest that are impacted by one or both of these effectors , as well as extend our understanding of the role of SecA2 , SapM , and PknG to Mtb inhibition of autophagosome maturation . These studies additionally reveal the value of using the secA2 mutant as a platform to study functions of effectors in phagosome maturation arrest .
With the goal of understanding the contribution of SecA2 to phagosome maturation arrest by Mtb , we tested the possibility that the SapM phosphatase is exported by the SecA2 pathway . SapM is a known effector of phagosome maturation arrest , [27] . Immunoblot analysis with SapM antisera was performed on Mtb culture supernatants . Compared to the parental Mtb strain , H37Rv , and a complemented strain , the Mtb secA2 mutant had significantly reduced levels of secreted SapM , although a low residual level of SapM secretion was always observed in the mutant ( Fig 1A ) . The amount of SapM in whole cell lysates was also reduced , albeit more modestly ( Fig 1B ) . These reduced levels of SapM were not due to transcriptional effects in the secA2 mutant , as shown by qRT-PCR measurements of sapM transcript in the secA2 mutant compared to H37Rv in both broth cultures as well as in Mtb infected macrophages ( S1A and S1B Fig ) . Thus , the lower levels of secreted SapM in the secA2 mutant is the likely consequence of a SapM export defect , and the reduced cellular levels may be due to cytoplasmic SapM being unstable in the absence of export . We also examined the contribution of SecA2 to SapM export by quantifying phosphatase activity in culture supernatants using p-nitrophenyl phosphate ( pNPP ) as a substrate . Even though this phosphatase assay is not specific for SapM ( i . e . other phosphatases are detected ) , we observed less phosphatase activity in culture supernatants of the secA2 mutant when compared to H37Rv or the complemented strain ( Fig 1C ) . Importantly , in the presence of sodium molybdate , a known inhibitor of SapM , the secreted phosphatase activity of the secA2 mutant was equivalent to that of H37Rv and complemented strains which is consistent with the secA2 mutant being defective in SapM secretion . ( S1C Fig ) [28] . Together , the immunoblot and activity data provide the first evidence of SapM being secreted by the SecA2 pathway . SapM was previously shown using in vitro approaches to dephosphorylate PI3P , which should limit recruitment of PI3P binding proteins , such as EEA1 , that promote downstream phagosome maturation events [7 , 29 , 30] . Consequently , we hypothesized that SapM secretion by the SecA2 pathway contributes to phagosome maturation arrest by enabling Mtb to avoid EEA1 localization to phagosomes . As a first step to test this possibility , murine bone marrow derived macrophages were infected with the secA2 mutant , H37Rv or the complemented strain and EEA1 localization to Mtb-containing phagosomes was determined using the endogenous auto-fluorescent signal of Mtb and immunostaining with anti-EEA1 antibodies . Compared to phagosomes containing H37Rv or the complemented strain , which avoid EEA1 localization , phagosomes containing the secA2 mutant exhibited significantly higher EEA1 co-localization at both 1hr and 24hrs post infection ( i . e . time following a 4hr period of initial uptake/infection ) ( Fig 1D and 1E , S1D Fig ) . We next set out to determine if the failure of the secA2 mutant to prevent EEA1 recruitment to phagosomes is due to the SapM secretion defect of the mutant . For this purpose , we built a strain of the secA2 mutant with the amount of secreted SapM restored to wild type levels . If SapM is the only SecA2-dependent effector preventing EEA1 recruitment , then restoring SapM secretion to wild type levels in the secA2 mutant background should rescue this step of phagosome maturation arrest . However , if additional SecA2-dependent effectors exist with roles in this step of phagosome maturation arrest , their export will remain compromised and the EEA1 defect will persist . To restore the level of SapM secretion , we introduced a plasmid that overexpressed SapM in the secA2 mutant background ( secA2+SapM ) . In this secA2 mutant strain , the level of secreted SapM was restored , even surpassing the level seen with H37Rv ( Fig 2A ) . While the mechanism of restored secretion is not clear , we suspect the overexpressed SapM is exported by an alternate pathway , as some SapM is observed in culture supernatants of the secA2 mutant ( Fig 1A ) . Importantly , the overexpressed SapM was functional as demonstrated by the increased secreted phosphatase activity of the secA2+SapM strain ( Fig 2B ) . Using this secA2+SapM strain , we tested how restored SapM secretion affects EEA1 recruitment to secA2 mutant containing phagosomes . Restored SapM secretion in the secA2 mutant fully rescued the secA2 mutant defect in preventing EEA1 ( Fig 2F , S2A Fig ) ( i . e . the percent EEA1+ Mtb containing phagosomes was equivalent between secA2+SapM and H37Rv ) . This result indicates that the defect in SapM secretion of the secA2 mutant accounts for the failure to exclude EEA1 from phagosomes . In other words , SecA2 secretion of SapM is required for Mtb to prevent EEA1 recruitment to phagosomes . The effect of overexpressing SapM was specific to the secA2 mutant , as SapM overexpression in H37Rv did not further reduce EEA1 recruitment ( Fig 2F , S2A Fig ) . Past studies lead to a model of SapM functioning to block phagosome maturation by dephosphorylating PI3P [7] . However , there is no direct evidence that the role of SapM in phagosome maturation arrest is through its phosphatase activity . By overexpressing a SapM variant lacking phosphatase activity in the secA2 mutant we tested the significance of SapM phosphatase activity during macrophage infection . Catalytic residues and the active site of SapM have yet to be studied . To create a phosphatase defective SapM , we substituted an alanine for histidine 204 , which aligns with a catalytically important residue in fungal acid phosphatases ( Fig 2C ) [31] . When plasmids overexpressing SapM or SapM H204A were introduced in the secA2 mutant , the level of secreted SapM was comparable , as measured by immunoblot ( Fig 2D ) . However , unlike overexpressed wild-type SapM , when SapM H204A was overexpressed there was no increase in secreted phosphatase activity , indicating H204 is essential for SapM phosphatase activity ( Fig 2E ) . Using SapM H204A , we then tested the importance of phosphatase activity to the role of SapM in preventing EEA1 recruitment to Mtb containing phagosomes . Unlike overexpressed SapM ( secA2+SapM ) , SapMH204A ( secA2+SapMH204A ) was unable to rescue the defect of the secA2 mutant in preventing EEA1 recruitment ( Fig 2G , S2B Fig ) . This result proves that the phosphatase activity of SapM is essential for SapM to exclude EEA1 from Mtb containing phagosomes . During the normal process of phagosome maturation , Rab5 is recruited to early phagosomes and is then exchanged for Rab7 as phagosomes mature . However , Mtb has the effect of retaining Rab5 and excluding Rab7 from phagosomes [5] Using immunofluorescence microscopy , we measured percent co-localization of Rab5 and Rab7 with secA2 mutant containing phagosomes . In contrast to H37Rv-containing phagosomes , secA2 mutant-containing phagosomes retained less Rab5 and recruited more Rab7 , confirming the secA2 mutant is defective for phagosome maturation arrest ( Fig 3A and 3B ) . Taking advantage of the secA2+SapM strain , we tested if SapM additionally impacts Rab5-Rab7 exchange . When secreted SapM was restored to the secA2 mutant , a partial , but significant , rescue of Mtb inhibition of Rab5-Rab7 exchange on phagosomes was observed ( i . e . restoring secreted SapM significantly increased Rab5 retention and reduced Rab7 recruitment ) ( Fig 3A and 3B ) . Furthermore , as shown with the phosphatase defective SapMH204A , the phosphatase activity of SapM is required for its function in inhibiting Rab5-Rab7 exchange ( Fig 3C and 3D ) . However , because the secA2+SapM strain did not restore the block in Rab5-Rab7 exchange to levels seen with H37Rv infected macrophages , this data argues for the existence of additional Mtb effectors exported by the SecA2 pathway impacting this step of phagosome maturation . It is noteworthy that the effect of the secA2+SapM strain on Rab5 retention and Rab7 exclusion waned as infection progressed ( 1 hr versus 24 hrs post infection ) ( Fig 3A and 3B ) . Avoiding phagosome acidification is another feature of Mtb phagosome maturation arrest that is impaired in secA2 mutant containing phagosomes [21] . Using LysoTracker , an acidotropic dye that accumulates in acidified compartments , we examined if restoring secreted SapM to the secA2 mutant rescues the ability of the mutant to avoid acidified phagosomes . The secA2+SapM strain was associated with a significant reduction in the percent LysoTracker co-localization ( acidification ) when compared to the secA2 mutant , indicating that SapM secretion by the SecA2 pathway contributes to Mtb inhibition of phagosome acidification ( Fig 3E , S3A Fig ) . However , the percentage of LysoTracker co-localization observed for the secA2+SapM strain was still significantly higher than that observed for H37Rv-containing phagosomes . This partial rescue reinforces the above conclusion that SapM is not the only SecA2-dependent effector of phagosome maturation arrest . The phosphatase activity of SapM is also required to prevent phagosome acidification as shown with SapMH204A ( Fig 3F , S3B Fig ) . We next examined the effect of restoring secreted SapM to the secA2 mutant on the ability to inhibit V-ATPase , the proton pump complex that acidifies the phagosome [4] . We previously showed that V-ATPase is excluded from Mtb containing phagosomes but has a significantly higher association with secA2 mutant-containing phagosomes [21] . In stark contrast to the effect restoring SapM secretion to the secA2 mutant had on phagosome acidification , no effect was observed on recruitment of the V-ATPase V1B1/B2 subunits ( Fig 3G , S3C Fig ) . To validate this result , we repeated the immunostaining utilizing antibodies that recognize a different component of the V-ATPase complex ( V0a1 ) . Again , the result revealed no effect of SapM on V-ATPase recruitment ( S4A and S4B Fig ) . These results are significant in revealing a role of SapM in preventing phagosome acidification that is independent from inhibiting recruitment of V-ATPase to phagosomes . Having previously linked the failure of the secA2 mutant to arrest phagosome maturation with the intracellular growth defect of the mutant , we tested the effect of restoring secreted SapM to the secA2 mutant on growth in macrophages . Intracellular growth was monitored over time by plating macrophage lysates for viable bacilli . While there was no difference in bacterial burden 24 hrs post infection , significantly fewer secA2 mutant bacilli were recovered after three and five days of infection compared to H37Rv ( Fig 3H ) . When secreted SapM was added back to the secA2 mutant , intracellular growth of the mutant significantly improved ( Fig 3H ) . The improvement in intracellular growth was dependent on phosphatase activity of SapM , as SapMH204A had no effect on intracellular growth of the secA2 mutant ( Fig 3I ) . However , intracellular growth of the secA2+SapM strain was not restored to the level exhibited by H37Rv , which reinforces the above conclusions that additional SecA2 exported effectors must exist . There was no effect on intracellular growth with SapM overexpression in H37Rv . Recent studies identified the PknG kinase , a protein with functions in Mtb physiology as well as phagosome maturation arrest , as being exported by the SecA2 pathway to the cell wall of Mtb and Mycobacterium marinum [13 , 24 , 32 , 33] . To elucidate the role of SecA2 export of PknG in phagosome maturation arrest , we took the same approach as used with SapM of testing the effect of restoring PknG export to the secA2 mutant . By overexpressing pknG in the secA2 mutant ( secA2+PknG ) we were able to restore export of PknG to greater than wild type levels ( Fig 4A ) . In contrast to the full rescue in EEA1 inhibition observed with SapM restoration in the secA2 mutant , restoring PknG export to the secA2 mutant had no effect on EEA1 ( Fig 4B , S5A Fig ) . However , restoring exported PknG to the secA2 mutant significantly increased the ability of the secA2 mutant to retain Rab5 and exclude Rab7 on phagosomes ( Fig 4C and 4D ) . This result reveals a role for PknG in preventing Rab5-Rab7 exchange that has not been described previously . However , like the secA2+SapM strain , the secA2+PknG strain was not as effective as H37Rv in inhibiting Rab5-Rab7 exchange , indicating it is not the only SecA2 exported effector involved in inhibiting this step of phagosome maturation . Intriguingly , unlike SapM , the effect seen with PknG restoration was consistent at both 1hr and 24hrs post infection ( Fig 4C and 4D ) . When we examined phagosome acidification using LysoTracker , restored PknG export in the secA2 mutant partially rescued the ability of the mutant to inhibit phagosome acidification ( Fig 4E , S5B Fig ) . However , as with restoring SapM secretion , recruitment of V-ATPase subunits V1B1/B2 and V0a1 were both unaffected by restoring export of PknG in the secA2 mutant ( Fig 4F , S4A , S4B and S5C Figs ) . Finally , we tested the effect of restored levels of exported PknG on intracellular growth of the secA2 mutant . The secA2+PknG strain grew significantly better than the secA2 mutant in macrophages , indicating SecA2 export of PknG contributes to intracellular growth of Mtb but , again , additional SecA2 exported proteins are also required , as growth was not restored to the levels seen with H37Rv ( Fig 4G ) . We next tested the effect of restoring export of SapM and PknG in combination to determine if these effectors have cumulative effects and if together they are sufficient to fully rescue the defects of a secA2 mutant . We simultaneously overexpressed sapM and pknG to restore export of both effectors in the secA2 mutant ( secA2+SapM+PknG ) . As expected , the secA2+SapM+PknG strain fully inhibited EEA1 recruitment , like the secA2+SapM strain ( Figs 2F and 5A , S6A Fig ) . When we examined Rab5 and Rab7 localization on phagosomes , simultaneous restoration of exported SapM and PknG to the secA2 mutant inhibited Rab5-Rab7 exchange significantly more than restoration of either effector individually ( Fig 5B and 5C ) . In fact , when compared to H37Rv at 1hr post infection , full rescue of the Rab5-Rab7 exchange inhibition was observed for the secA2+SapM+PknG strain . However , at 24hrs post infection the effect waned , which is reminiscent of what was observed with the secA2+SapM strain ( Fig 3A and 3B ) . In regard to phagosome acidification , the secA2+SapM+PknG strain had a greater effect on inhibiting phagosome acidification ( LysoTracker ) than observed with restoration of either effector individually ( Fig 5D , S6B Fig ) . However , phagosome acidification was still not inhibited to wild-type levels by the secA2+SapM+PknG strain ( Fig 5D , S6B Fig ) . Furthermore , even when export of both effectors was restored , exclusion of the V1B1/B2 or V0a1 subunits of V-ATPase was not rescued ( Fig 5E , S4A , S4B and S6C Figs ) . Finally , we tested the effect of restoring export of both effectors on growth of the secA2 mutant in macrophages . The secA2+SapM+PknG strain grew significantly better than the secA2 mutant with each effector restored individually ( Fig 5F , S6D Fig ) . However , as seen with phagosome maturation arrest , the secA2+SapM+PknG strain was not fully rescued in its ability to grow intracellularly ( Fig FG , S6D Fig ) . Thus , the secA2+SapM+PknG strain revealed a cumulative effect of adding back exported SapM and PknG on Rab5-Rab7 exchange , acidification and intracellular growth . However , in nearly all cases adding back these two effectors was insufficient to restore phenotypes to the level seen with H37Rv , which indicates the existence of even more SecA2-dependent effectors of phagosome maturation arrest . The discrepancy between the effects of SapM and PknG on phagosome acidification and V-ATPase localization suggests that Mtb has another mechanism of blocking acidification in addition to inhibiting V-ATPase recruitment . Hv1 is a voltage gated proton channel that was recently shown to localize to the phagosomal membrane of macrophages and to contribute to phagosome acidification along with V-ATPase [34 , 35] . Using immunofluorescence microscopy , we measured percent co-localization of Hv1 with H37Rv containing phagosomes . Only a small percentage ( 9 . 8% ) of H37Rv localized to Hv1-positive phagosomes ( Fig 6 ) . In contrast to live H37Rv , the level of Hv1 co-localization was significantly higher in macrophages infected with either heat-killed H37Rv or non-pathogenic Mycobacterium smegmatis ( 21 . 4% and 22 . 5% respectively ) ( Fig 6 ) . Thus , live Mtb has a lower association with Hv1-containing phagosomes than dead Mtb or non-pathogenic mycobacteria . However , there was no significant difference in Hv1 association in H37Rv versus secA2 mutant infected macrophages ( Fig 6 ) . Therefore , while limiting Hv1 recruitment may be another mechanism by which live Mtb limits acidification of phagosomes , it does not account for the role of the SecA2 exported effectors SapM and PknG in this process . In addition to phagosome maturation arrest , Mtb inhibits the maturation of autophagosomes to autophagolysosomes which is sometimes referred to as autophagy flux [18 , 19] . To determine if the SecA2 pathway is additionally required for autophagosome maturation arrest , we used LC3-II , the lipid modified form of LC3 that is associated with autophagosomes , to monitor autophagy in H37Rv and secA2 mutant infected RAW 264 . 7 macrophages [36] . Lower levels of LC3-II were observed in secA2 mutant infected macrophages when compared to H37Rv infected macrophages both immediately after the 4hr infection and 24 hrs post infection . ( Fig 7A ) . The lower LC3-II levels were not due to a reduced bacterial burden in secA2 mutant infected RAW cells as there was no difference in intracellular burden of H37Rv or the secA2 mutant at these time points ( S7E Fig ) . The lower levels of LC3-II in secA2 mutant infected macrophages could indicate a defect of the secA2 mutant in arresting autophagosome maturation such that there are more mature autophagosomes resulting in more LC3-II degradation . To test this possibility Mtb infected cells were treated with Bafilomycin A1 , which blocks autophagosome acidification , maturation and the associated degradation of LC3-II . With Bafilomycin A1 treatment , the levels of LC3-II were comparable in secA2 mutant and H37Rv infected macrophages . This result is consistent with the secA2 mutant being defective in the ability to arrest autophagosome maturation ( Fig 7A ) . To further examine the requirement for the SecA2 pathway in autophagosome maturation arrest we utilized RAW 264 . 7 macrophages expressing a dual RFP::GFP::LC3 fusion protein ( RAW-Difluo mLC3 cells ) . While RFP is resistant to the acidic environment of the autophagolysosome , GFP is acid sensitive and the signal is quenched in autophagolysosomes . By quantifying the number of RFP+ and GFP+/- autophagosomes , this cell line can report on autophagosome maturation . When we infected cells expressing RFP::GFP::LC3 with the secA2 mutant , H37Rv or the complemented strain , there was no difference in the percent of Mtb that co-localized with any LC3+ compartments ( RFP+ , GFP+/- ) ( Fig 7B and S7A Fig ) . However , when we specifically examined the association of Mtb with mature autophagosomes by quantifying the percentage of Mtb that localize to autophagolysosomes ( RFP+ , GFP- ) , we found a significantly higher association of the secA2 mutant with autophagolysosomes than either H37Rv or the complemented stain . Together , the LC3-II immunoblots and the RFP::GFP::LC3 reporter indicate an additional role of the SecA2 pathway in autophagosome maturation arrest ( Fig 7C , S7B Fig ) . Using the secA2 mutant strains with restored export of SapM and/or PknG , we next set out to determine if SecA2 export of SapM and PknG contributes to the function of SecA2 in autophagosome maturation arrest . Restored export of either SapM or PknG reduced localization of the secA2 mutant in autophagolysosomes indicating both SapM and PknG contribute to Mtb inhibition of autophagosome maturation ( Fig 7D , S8A Fig ) . Notably , restored SapM secretion resulted in a more significant reduction in secA2 localization to autophagolysosomes than PknG ( Fig 7D , S8A Fig ) . Simultaneous restoration of SapM and PknG export was more effective than restoration of either effector individually ( Fig 7D , S8A Fig ) . In fact , when compared to H37Rv , full rescue of autophagosome maturation arrest was observed for the secA2+SapM+PknG strain ( Fig 7D , S8A Fig ) . A benefit of utilizing the RFP::GFP::LC3 expressing cell line is the ability to simultaneously examine autophagosome and phagosome maturation in the same cells . To monitor phagosome maturation , we quantified co-localization of LC3 negative ( LC3- ) Mtb with LysoTracker . In LC3- phagosomes , the secA2 mutant localized more frequently to mature LysoTracker positive phagosomes than H37Rv ( Fig 7E , S8B Fig ) . Moreover , the secA2+SapM and secA2+PknG strains exhibited significantly reduced localization to mature LC3- phagosomes when compared to the secA2 mutant ( Fig 7E , S8B Fig ) . Interestingly , unlike with autophagosome maturation arrest , the effect on phagosome maturation of adding back exported SapM to the secA2 mutant was significantly less than that of PknG ( Fig 7E , S8B Fig ) . The secA2+SapM+PknG strain exhibited even greater rescue of phagosome maturation arrest than observed with restoration of either effector individually ( Fig 7E , S8B Fig ) . However , in contrast to autophagosome maturation , the secA2+SapM+PknG strain was not fully rescued in its ability to arrest phagosome maturation ( Fig 7E , S8B Fig ) . The function of SapM in both autophagosome and phagosome maturation arrest depends on SapM phosphatase activity , as shown by the secA2+SapMH204A strain remaining defective in both processes ( S8C and S8D Fig ) . Together , these results demonstrate that both phagosome and autophagosome maturation arrest depend on the SecA2 pathway , SapM , and PknG . However , these experiments also reveal differences in the contribution individual effectors make to each process and expose the existence of additional SecA2-dependent effectors that are required for phagosome maturation arrest but not necessarily for autophagosome maturation arrest .
Phagosome maturation arrest by Mtb is complex and much remains to be learned about the effectors involved in the process and how they work together . We showed previously that the SecA2 pathway is required for Mtb to inhibit phagosome maturation; however , the SecA2-dependent effectors of phagosome maturation arrest remained unknown [21] . Here , we identified two SecA2 exported effectors of Mtb phagosome maturation arrest as the phosphatase SapM and the kinase PknG . Then , using a strategy of adding back export of SapM and PknG to the secA2 mutant , we not only established the significance of the role of the SecA2 pathway in exporting these proteins but we identified steps in phagosome maturation that are impacted by these factors individually and in combination . Moreover , we revealed that the SecA2 pathway , SapM , and PknG also function in inhibiting autophagosome maturation . Prior to this study , SapM was not known to be secreted by the SecA2 export pathway . By testing the requirement for SecA2 in the export of a set of known effectors of phagosome maturation arrest ( SapM , LpdC , Ndk ) we identified SapM as a SecA2-exported protein [15 , 37] . The SecA2 pathway did not contribute to LpdC or Ndk secretion , and these effectors were not studied further ( S9 Fig ) . PknG was identified as being exported by the SecA2 pathways of Mtb and M . marinum in recent proteomic studies [24 , 32] . While SapM and PknG are both known to function in phagosome maturation arrest and there are reports of Mtb mutants lacking these effectors having defects in phagosome maturation arrest , our understanding of their roles in inhibiting specific steps of phagosome maturation is far from complete [7 , 12 , 13 , 27] . We established the significance of SecA2 export of SapM and PknG to phagosome maturation arrest and intracellular growth , using the strategy of adding back export of these proteins to the secA2 mutant . To create the necessary strains , we reasoned that overexpressing SecA2-dependent proteins in the secA2 mutant could boost their export through the alternate mechanism , possibly the SecA1-dependent pathway , that accounts for the residual export in the secA2 mutant of SapM , PknG and all SecA2 exported proteins identified to date [23 , 24 , 32 , 38] . Using overexpression , we achieved our goal of producing a secA2 mutant strain that has at least as much exported SapM and/or PknG as detected in the wild type H37Rv strain . Notably , even when overexpressed , the secA2 mutant exported less SapM and PknG than the corresponding H37Rv overexpression strain , confirming the dependency of these effectors on SecA2 for export . The effects of SapM and/or PknG overexpression were specific to the secA2 mutant and specific to the overexpressed proteins . H37Rv was unaffected by increased levels of these proteins and equivalent levels of overexpressed SapM H204A in the secA2 mutant had no effects . We also repeated the experiments using a single-copy vector with reduced , though still higher than wild-type levels , of secreted SapM and saw comparable restoration of phagosome maturation arrest and intracellular growth ( S11 Fig ) . The secA2 mutant strains with restored levels of exported SapM and/or PknG allowed us to investigate steps in phagosome maturation affected by these effectors . Past studies demonstrate purified SapM can dephosphorylate PI3P in vitro [7] . This data led to a model for secreted SapM dephosphorylating PI3P and inhibiting recruitment of PI3P binding proteins , such as EEA1 , to the phagosome to arrest phagosome maturation . However , critical details of this model have not been confirmed , including an effect of SapM on EEA1 and proof that SapM functions as a phosphatase to arrest phagosome maturation . Thus , our demonstration that restored levels of exported wild type SapM , but not the phosphatase defective SapM H204A , inhibits EEA1 localization to phagosomes provides important validation of the model . Adding back wild type SapM also partially restored inhibition of Rab5-Rab7 exchange and this again depends on SapM phosphatase activity . This effect of SapM on Rab5-Rab7 exchange was not previously noted , but it is consistent with the role of PI3P in Rab5-Rab7 exchange and is intriguing given a report of SapM binding to Rab7 [39–41] . The SapM effect on Rab5-Rab7 exchange was reproducibly more extreme at 1hr versus 24hrs post infection ( Fig 3A and 3B ) . The temporal nature of effector functions revealed by this data reveals another layer of complexity to phagosome maturation arrest by Mtb . The strategy of overexpressing PknG to restore exported levels to the secA2 mutant was used previously in M . marinum [32] . Similar to what we observed , adding back exported PknG to the secA2 mutant of M . marinum results in a partial restoration of phagosome maturation arrest . However , in the M . marinum study , the effect of PknG was only assessed on co-localization with the late lysosomal-associated membrane protein ( LAMP1 ) [32] . Since the function ( s ) of PknG that impact phagosome maturation is unknown , we took advantage of the secA2+PknG strain to reveal effects on individual steps of phagosome maturation . PknG had no effect on EEA1 recruitment to phagosomes , but it did partially restore inhibition of Rab5-Rab7 exchange , revealing for the first time a function of PknG in inhibiting Rab5-Rab7 exchange . When exported SapM and PknG were added back simultaneously , a combined effect was observed that resulted in complete inhibition of Rab5-Rab7 exchange at 1hr post infection but waned as infection progressed . Thus , multiple SecA2-dependent effectors act on the same step of phagosome maturation and the combinatorial effects of these effectors suggests SapM and PknG work through different but complementary mechanisms . Studies of PknG function in phagosome maturation arrest are complicated by the fact that , in addition to being exported , PknG functions in the bacterial cytoplasm in glutamate metabolism , regulation of the TCA cycle , and in a redox homeostatic system ( RHOCS ) that contributes to resistance to oxidative stress [11 , 12 , 33 , 42] . Because Mtb mutants with RHOCS defects are delivered to mature phagosomes and have intracellular growth defects , it raised the possibility that the redox function of PknG explains its role in phagosome maturation arrest [12] . However , this does not appear to be the case as the secA2 mutant did not exhibit a RHOCS defect , as assessed by sensitivity to redox stress , and overexpression of PknG did not increase resistance to redox stress ( S10 Fig ) . The ability of Mtb to exclude V-ATPase from the phagosome is generally assumed to account for the lack of acidification of Mtb phagosomes [4 , 43] . Therefore , our demonstration that restoration of SapM and/or PknG export to the secA2 mutant partially rescued the acidification defect of the secA2 mutant but had no effect on recruitment of V-ATPase was a surprise . These results indicate the existence of additional mechanism ( s ) for Mtb to prevent phagosome acidification that were not previously appreciated . One possibility is that Mtb affects an additional proton transporter that contributes to phagosome acidification in conjunction with V-ATPase , such as Hv1 [34 , 35] . Unlike the V-ATPase , nothing is known about how Hv1 interfaces with bacterial pathogens . Because Hv1 also impacts NADPH oxidase-dependent generation of reactive oxygen species ( ROS ) by providing a compensating charge for electrons transferred to superoxide , an effect on Hv1 phagosomal levels could potentially not only affect phagosome acidification but also ROS production [35 , 44 , 45] . Our data suggests that Mtb actively limits Hv1 recruitment to macrophages , but in a manner that is independent of SecA2 , SapM and PknG . Alternate explanations for how Mtb affects phagosome acidification independent of V-ATPase exclusion include inhibiting the counter ion flux that is required for acidification to occur or direct inhibition of the V-ATPase pump [46] . In terms of the latter possibility , phosphorylation of the ATPase subunit of the V-ATPase is known to regulate activity of the proton pump , which raises the possibility of SapM and/or PknG affecting acidification by impacting phosphorylation of the V-ATPase [47] . Finally , because adding back SapM and PknG failed to rescue the mutant defect in excluding V-ATPase from the phagosome , there is at least one additional SecA2-dependent effector involved in this step of phagosome maturation arrest . PtpA , which binds subunit H of V-ATPase and thereby excludes the proton pump from phagosomes , is a candidate for this missing SecA2-exported effector [8] . Unfortunately , our inability to detect secreted PtpA in Mtb cultures prevented us from determining if PtpA is secreted by the SecA2 pathway . Compared to phagosome maturation arrest , even less is known about autophagosome maturation arrest by Mtb . Using the RFP::GFP::LC3 reporter , we were able to reveal a role for SecA2 export in the maturation arrest of both LC3- phagosomes and LC3+ autophagosomes . It is important to note in this study , we are unable to distinguish between autophagosomes and other LC3+ compartments , including LC3 associated phagosomes ( LAP ) so we cannot exclude a function for SecA2 in those processes . Using the secA2+SapM+PknG strain we were able to demonstrate a function for both SapM and PknG in autophagosome maturation arrest by Mtb . Our data confirms a recent study using sapM transfected cells that suggests a role for SapM in autophagosome maturation arrest but this is the first evidence indicating a function for PknG in this process [39] . Intriguingly , the SecA2 exported effectors do not affect both autophagosome and phagosome maturation equally . SapM seems to have more of an effect on autophagosomes while PknG is more impactful on altering phagosomes . Further , while simultaneous restoration of both effectors was able to fully rescue the defect of the secA2 mutant in autophagosome maturation , it was not sufficient to rescue the defect in phagosome maturation . Our results highlight the overlap in Mtb factors involved in phagosome and autophagosome maturation but at the same time reveal differences in specificity of Mtb effectors for both processes . It is important to note that by focusing on effectors exported by the SecA2 pathway , our study does not rule out or diminish the significance of effectors that are exported by other routes . However , at the same time , the fact that the secA2 mutant exhibits phagosome and autophagosome maturation arrest defects indicates that SecA2-independent effectors are not sufficient on their own to block these critical macrophage responses . When we investigated the effect of SecA2 export of SapM and PknG on Mtb growth in macrophages , we found adding back either effector individually improved intracellular growth of the secA2 mutant while restoring export of both effectors simultaneously resulted in a further improvement . These results reinforce prior studies indicating that the role of SecA2 in inhibiting Mtb delivery to mature phagosomes is required for intracellular growth [21] . Furthermore , the effect of SecA2 , SapM and PknG on phagosome maturation arrest will likely extend beyond promoting replication in macrophages . By arresting phagosome maturation , Mtb also limits the presentation of antigenic peptides to the immune system [48] . In summary , our studies demonstrate that multiple effectors require the SecA2 pathway for their export and function in phagosome maturation arrest and they provide unique insight into how Mtb effectors work in concert to inhibit phagosome and autophagosome maturation . Our studies also revealed the advantages of using of the secA2 mutant as a platform to study the function of effectors individually or in combination . This approach provides an alternative to studying effectors through deletion analysis , which can be problematic for effectors that share redundant functions or for effectors that have additional unrelated functions in Mtb ( such as PknG ) . In this study , we discovered new layers of complexity in how Mtb arrests phagosome maturation ( multiple means of inhibiting acidification , temporal effects ) , exposed a new host factor inhibited by Mtb ( Hv1 ) , uncovered distinct and cumulative effects of a pair of effectors , and revealed a broad role of the SecA2 pathway in phagosome and autophagosome maturation arrest that involves SapM , PknG and additional effectors that await identification .
This study included the use of mice and followed recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . The protocol was approved by the International Animal Care and Use Committee at the University of North Carolina at Chapel Hill ( protocol: 15–018 . 0 ) . In this study we used the Mycobacterium tuberculosis wild type strain H37Rv , and the ΔsecA2 mutant ( mc23112 ) generated in the H37Rv background as well as the wild-type Mycobacterium smegmatis strain MC2155 . [23] . The plasmids and strains over-expressing sapM and/or pknG constructed for this study are listed in S1 and S2 Tables respectively . Mtb strains were cultured in either liquid Middlebrook 7H9 ( BD ) or on solid Middlebrook 7H10 ( BD ) or 7H11 ( Sigma ) media supplemented with 0 . 05% Tween 80 , 0 . 5% glycerol , 1x albumin dextrose saline ( ADS ) and kanamycin ( 20μg/ml ) or hygromycin ( 50μg/ml ) when appropriate . Sauton media was used for preparation of culture supernatants containing 30mM DL-asparagine , 7mM sodium citrate , 3mM potassium phosphate dibasic , 4mM magnesium sulfate , 0 . 2mM ferric ammonium citrate and 4 . 8% glycerol adjusted to a pH of 7 . 4 . For cell wall isolation , we utilized a modified Middleboook 7H9 based media containing 0 . 1% glycerol , 1mM proprionic acid , 0 . 1% tyloxapol , 0 . 1M MES ( buffer ) , 0 . 5% BSA , and adjusted to a pH of 6 . 5 [24] . The Histidine at position 204 of SapM was changed to an Alanine using site directed mutagenesis to generate SapMH204A . The sapM expression plasmid pJTS130 was used as a template . Primer sequences are as follows: 5'-cgatcgagccgtcggccatgtcgttgtcgg-3' and 5'-ccgacaacgacatggccgacggctcgatcg-3' . Dpn1 ( NEB ) was added to degrade the methylated template . Mutation was confirmed by sequencing . For culture supernatant isolation , cultures were first grown to log-phase in Middlebrook 7H9 with 0 . 05% tyloxapol . Cultures were then washed in Sauton media and grown in Sauton media with 0 . 05% tyloxapol for 5 days after which cultures were washed again to remove detergent and sub-cultured into 100ml Sauton media ( no detergent ) at a starting OD600 of 0 . 25 for 2 days . Then the entire 100 ml culture was centrifuged at 3500 rpm and supernatants were collected and double filtered with a 0 . 2μm filter to remove cells . Culture supernatants were concentrated 100-fold using a 15 ml capacity 10 , 000 MW cut off centrifugal filter ( Millipore ) . For Immunoblots , proteins were isolated by precipitation using 10% trichloroacetic acid overnight . To confirm that the supernatants were free of cytosolic contamination due to cell lysis , samples were examined by Immunoblot for absence of the cytoplasmic mycobacterial proteins SigA , GroEL , and SecA1 . For cell wall isolation , Mtb was first grown in 7H9 0 . 05% Tyloxapol to mid-log phase and then sub-cultured into the modified 7H9 media at a starting OD600 of 0 . 125 . Cultures were harvested when they reached an OD600 of 1 . 0 and were then sterilized by gamma-irradiation in a JL Shephard Mark I 137Cs irradiator ( Dept . of Radiobiology , University of North Carolina at Chapel Hill ) prior to removal from BSL-3 containment . Subcellular fractions were isolated as previously described [49] . Briefly , irradiated cells were suspended in 1X PBS containing protease inhibitors and lysed by passage four times through a French pressure cell . Unlysed cells were removed by centrifugation at 3500 rpm to generate clarified whole cell lysates ( WCLs ) , which were then spun at 25 , 000 rpm for 30 minutes to pellet the cell wall fraction . Protein concentrations were determined by Bicinchoninic acid assay ( Pierce ) . Equal protein for whole cell lysates , cell wall fractions , or concentrated culture supernatants was run on a SDS-PAGE gel , and then transferred to nitrocellulose membranes . After transfer , the membranes were blocked for one hour and then probed with primary antibodies . Antibodies to Mtb proteins were kind gifts of Vojo Deretic , University of New Mexico ( SapM ) , Zakaria Hmama , University of British Columbia ( LpdC and NdkA ) , Yossef Av-Gay , University of British Columbia ( PknG ) and Douglas Young , Imperial College ( 19kDa ) . LC3 and Actin antibodies were acquired from Cell Signaling Technologies . Antibodies were used at the following dilutions ( SapM 1:5 , 000 , LpdC 1:2 , 000 , NdkA 1:2 , 000 , PknG 1:5 , 000 , 19kDa 1:20 , 000 , LC3 1:500 , and Actin 1:1000 ) Secondary antibodies were conjugated to horseradish peroxidase ( BioRad ) and signal was detected using chemiluminescence ( Western Lighting Perkin Elmer ) . SapM phosphatase activity was assayed as described previously [28] . The phosphatase activity of 5 μg of culture supernatants was assessed for triplicate samples . Each reaction contained 0 . 1mM Tris base pH 6 . 8 and 50mM p-nitrophenyl phosphate ( pNPP ) ( New England Biolabs ) with either 2 mM sodium tartrate to inhibit background phosphatase activity or 1 mM sodium molybdate to inhibit SapM activity . Samples were incubated at 37°C and the absorbance at 405nm was measured every minute for two hours . We then calculated the rate of pNPP conversion and normalized the data to H37Rv . To isolate RNA from Mtb grown in vitro , triplicate Mtb cultures were grown in modified 7H9 medium to an OD600 of 1 . 0 and pelleted by centrifugation at 3 , 000 rpm for 10 min . Bacteria were lysed in 1 ml 3:1 chloroform-methanol , then vortexed with 5 ml TRIzol ( Invitrogen ) and incubated for 10 min at room temperature . Phases were separated by centrifugation at 3 , 000 rpm for 15 min at 4°C , and RNA was precipitated from the upper phase using 1X volume of isopropanol . RNA was pelleted by centrifugation at 12 , 000 rpm for 30 min at 4°C , washed twice with cold 70% ethanol , and resuspended in RNase-free water . Mycobacterial RNA was isolated from Mtb infected macrophages as previously described [50 , 51] . Triplicate plates containing 2x107 RAW 264 . 7 cells were infected at an MOI of 10 for 4 hrs . After 24hrs of infection , cells were washed with PBS and lysed using a guanidine thiocyanate buffer as previously described [47] . Mtb was pelleted by centrifugation and resuspended in 65°C TRIzol . Glass beads were added to the samples and the samples were vortexed to maximize lysis . Chloroform was added for a final concentration of 20% . Phases were separated by centrifugation at 3 , 000 rpm for 15 min at 4°C , and RNA was precipitated from the upper phase using 1X volume of isopropanol . RNA was pelleted by centrifugation at 12 , 000 rpm for 30 min at 4°C , washed twice with cold 70% ethanol , and resuspended in RNase-free water . RNA samples were treated with DNase ( Promega ) and then column purified ( Zymo RNA clean and concentrator kit ) . Following RNA isolation , cDNA was synthesized with random primers using the iScript cDNA Synthesis Kit ( BioRad ) . Real-time PCR was completed using 25ng of cDNA template in triplicate technical replicates using the SensiMix SYBR and fluorescein kit ( Bioline ) . Transcripts were normalized to the housekeeping gene sigA . Primer sequences are for sapM ( ATCGTTGCTGGCCTCATGG and AGGGAGCCGACTTGTTACC ) and sigA ( GAGATCGGCCAGGTCTACGGCGTG and CTGACATGGGGGCCCGCTACGTTG ) . For bone marrow-derived macrophages ( BMDM ) , femurs were isolated from C57/Bl6 ( Jackson Labs ) mice and flushed with complete DMEM ( DMEM [Sigma] supplemented with 10% Heat inactivated fetal bovine serum [FBS] 5mM non-essential amino acids and 5mM L-glutamine ) . Bone marrow cells were washed , re-suspended and plated in complete DMEM containing 20% L-929 cell conditioned media ( LCM ) . After six days , the cells were lifted off the plates using cold 5mM EDTA in PBS . Macrophages were seeded at 2 × 105 macrophages/well in complete DMEM containing 20% LCM using either eight-well chambered slides to monitor growth of Mtb or chambered cover slips for microscopy experiments . After resting 24 hours the macrophages were infected with Mtb grown to log-phase , and washed twice with PBS containing 0 . 05% Tween 80 and diluted in warm complete DMEM . BMDM were infected at an MOI of 1 . 0 for four hours . Infected macrophages were then washed three times with pre-warmed complete DMEM to remove extracellular bacteria . Macrophages were lysed using 0 . 1% Triton X-100 at various time points and lysates were plated for cfu determination or slides were fixed in 4% paraformaldehyde ( PFA ) for immunofluorescence staining The 1 hour ( hr ) and 24 hr post infection time points reflect time following 4 hours of initial uptake/infection . At both the 1 and 24hr time points there were no differences in intracellular viability for any of the strains used in this study ( Figs 3 and 4 , S6 Fig ) . M . smegmatis infections followed the same procedure as Mtb . To prepare heat-killed H37Rv , H37Rv was prepared for infection as described above but heated to 80°C for 1hr prior to dilution in complete DMEM . RAW 264 . 7 cells were cultured in DMEM supplemented with 10% FBS . Cells were seeded at 1*106 macrophages/well ( 6-well plate ) or 1*105 macrophages/well ( 8-well chamber slide ) . For immunoblots , RAW cells were infected at an MOI of 10 for 3 hrs and washed three times with pre-warmed DMEM to remove extracellular bacteria . Bafilomycin A1 ( Sigma ) was utilized at a concentration of 10nM and maintained throughout the course of the experiment . Cells were lysed using RIPA buffer ( 50mM Tris-HCL pH 7 . 4 , 1% NP-40 , 0 . 25% Sodium deoxycholate , 150mM NaCl , and protease inhibitors ) . For cfu determination RAW cells were infected at an MOI of 1 for 4 hrs and washed three times with pre-warmed DMEM to remove extracellular bacteria . Cells were lysed using 0 . 1% Triton X-100 at various time points and lysates were plated for cfu determination . RAW-Difluo mLC3 Cells expressing RFP::GFP::LC3 ( InvivoGen ) were cultured in DMEM supplemented with 10% FBS and zeocin . Cells were seeded at 1*105 macrophages/well without zeocin and infected at an MOI of 1 for 4 hrs and washed three times with pre-warmed DMEM to remove extracellular bacteria . At 1hr or 24hrs post infection cells were fixed in 4% PFA . For LysoTracker staining , media on Mtb infected BMDM was replaced with prewarmed DMEM containing 100nM LysoTracker Red DND99 ( Invitrogen ) for BMDM and 100nM LysoTracker Deep Red ( Invitrogen ) for RAW 264 . 7 cells and incubated for one hour . After which , media was removed and the slides fixed in 4% PFA . For immunofluorescence , media was removed from Mtb infected macrophages and the slides were submerged in 4% PFA . The fixed slides were submerged in PBS to remove residual PFA and then cells were permeabilized with 0 . 1% Triton-X 100 in PBS for 5 minutes at room temperature , washed in PBS and blocked in PBS containing 10% donkey serum . Antibodies to the mammalian markers Rab5 ( S-19 ) [52] , Rab7 ( H-50 ) [52] , V-ATPase B1/B2 ( H-180 ) [8] and V-ATPase A1 ( D-20 ) and Texas Red conjugated donkey anti-rabbit or donkey anti-goat secondary antibodies were acquired from Santa Cruz Biotechnology . Antibodies to EEA1 were acquired from Abcam ( ab2900 ) [53] . Antibodies to Hv1 were acquired from Invitrogen ( PA5-21008 ) . Primary antibodies were used at a 1:50 dilution in PBS with 3% serum and incubated overnight at 4°C . After which slides were washed in PBS , and secondary antibodies conjugated to TR fluorophores were used at 1:100 dilution in PBS with 3% serum and incubated at room temperature for one hour . Slides were washed to remove the secondary antibody and Fluormount-G ( Southern Biotech ) was added to each well to protect the fluorescent signal . As controls we included uninfected cells and single or no antibody controls . Widefield fluorescence microscopy was performed using an Olympus IX-81 controlled by the Volocity software package . All images were taken using a 60X oil-immersion objective . To visualize Mtb in infected macrophages we used the endogenous autofluorescence of the bacteria . Mycobacterial autofluorescence was visualized using a CFP filter cube ( Semroc ) with an excitation band of 426-450nm and emission band of 467-600nm [21] . As the autofluorescent signal quenches quickly , samples were prepared in the dark and the CFP channel was the first visualized on the microscope . A minimum of ten fields per well were captured and a minimum of 250 bacteria per well were scored for phagosomal markers . For each experimental group four replicate wells ( i . e . ≥1000 bacteria per infection condition ) were analyzed per experiment and data represents a minimum of two independent experiments . Representative images for each phagosomal marker are shown in Supplemental S12–S19 Figs . To assess sensitivity to oxidative stress Mtb cultures were exposed to 5mM H2O2 in 7H9+ADS 0 . 05% Tween 80 for 24 and 48 hours . Survival was assessed by plating for viable CFU . Cultures without H2O2 were included as controls . Strains tested in this manner include H37Rv and the secA2 mutant with and without PknG overexpression . A pstA1::tn mutant ( generous gift of Jyothi Rengarajan , Emory University ) which is extremely sensitive to oxidative stress was included as a control [54 , 55] .
|
Mycobacterium tuberculosis ( Mtb ) is the infectious agent of the disease tuberculosis . Inside the host , Mtb replicates primarily within the phagosome of macrophages . To replicate within macrophages , Mtb modifies the phagosome by inhibiting the normal host process of phagosomes maturing into acidified degradative phagolysosomes . In order to arrest this process of phagosome maturation , Mtb exports multiple effectors to the host-pathogen interface . Here we found that the specialized SecA2 protein export pathway of Mtb exports two such effectors: SapM and PknG . We discovered that SapM and PknG play non-redundant functions in phagosome maturation arrest by Mtb . We further demonstrated that SecA2 export of both SapM and PknG contributes to the ability of Mtb to replicate in macrophages . We also identified a role for the SecA2 pathway , SapM and PknG in arresting the host process of autophagosome maturation . Our research highlights how two effectors , SapM and PknG , work in concert but also have distinct roles in phagosome and autophagosome maturation arrest by Mtb .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"vesicles",
"immune",
"cells",
"immunofluorescence",
"staining",
"enzymes",
"immunology",
"enzymology",
"phosphatases",
"physiological",
"processes",
"phagosomes",
"immunologic",
"techniques",
"cellular",
"structures",
"and",
"organelles",
"bacteria",
"research",
"and",
"analysis",
"methods",
"specimen",
"preparation",
"and",
"treatment",
"staining",
"white",
"blood",
"cells",
"animal",
"cells",
"proteins",
"immunoassays",
"actinobacteria",
"immunofluorescence",
"biochemistry",
"cell",
"staining",
"cell",
"biology",
"mycobacterium",
"tuberculosis",
"physiology",
"secretion",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"macrophages",
"organisms"
] |
2018
|
The SecA2 pathway of Mycobacterium tuberculosis exports effectors that work in concert to arrest phagosome and autophagosome maturation
|
Bacteria , yeast and human cancer cells possess mechanisms of mutagenesis upregulated by stress responses . Stress-inducible mutagenesis potentially accelerates adaptation , and may provide important models for mutagenesis that drives cancers , host pathogen interactions , antibiotic resistance and possibly much of evolution generally . In Escherichia coli repair of double-strand breaks ( DSBs ) becomes mutagenic , using low-fidelity DNA polymerases under the control of the SOS DNA-damage response and RpoS general stress response , which upregulate and allow the action of error-prone DNA polymerases IV ( DinB ) , II and V to make mutations during repair . Pol IV is implied to compete with and replace high-fidelity DNA polymerases at the DSB-repair replisome , causing mutagenesis . We report that up-regulated Pol IV is not sufficient for mutagenic break repair ( MBR ) ; damaged bases in the DNA are also required , and that in starvation-stressed cells , these are caused by reactive-oxygen species ( ROS ) . First , MBR is reduced by either ROS-scavenging agents or constitutive activation of oxidative-damage responses , both of which reduce cellular ROS levels . The ROS promote MBR other than by causing DSBs , saturating mismatch repair , oxidizing proteins , or inducing the SOS response or the general stress response . We find that ROS drive MBR through oxidized guanines ( 8-oxo-dG ) in DNA , in that overproduction of a glycosylase that removes 8-oxo-dG from DNA prevents MBR . Further , other damaged DNA bases can substitute for 8-oxo-dG because ROS-scavenged cells resume MBR if either DNA pyrimidine dimers or alkylated bases are induced . We hypothesize that damaged bases in DNA pause the replisome and allow the critical switch from high fidelity to error-prone DNA polymerases in the DSB-repair replisome , thus allowing MBR . The data imply that in addition to the indirect stress-response controlled switch to MBR , a direct cis-acting switch to MBR occurs independently of DNA breakage , caused by ROS oxidation of DNA potentially regulated by ROS regulators .
Spontaneous mutations drive development of most cancers and their resistance to therapy , aging , pathogen escape from the immune response and antibiotics , and evolution generally . Although central in all of biology and many aspects of human health , the causes of spontaneous mutagenesis have long been elusive , and generally difficult to assign with confidence . Whereas many ways to increase or induce mutagenesis are known , the origins of spontaneous mutations in most organisms remain speculative [1 , 2] . Spontaneous mutations could potentially form by many different mechanisms [1 , 2] . An early proposal [3] was that most spontaneous mutations result from spontaneous DNA damage—that repair of damaged DNA is more mutagenic than standard DNA synthesis . The mutagenicity of DNA repair has been borne out , for example , by demonstrations that repair of DNA double-strand breaks ( DSBs ) is mutagenic in bacteria [4–8] , yeast [9] , and human cancer [10–12] ( reviewed [13–16] ) . That DNA damage underlies much of spontaneous mutagenesis was supported by discoveries that yeast antimutator mutants , i . e . , mutants with lower-than-normal spontaneous mutation rate , carry mutations in DNA-damage-survival genes [17] . These genes encode alternative error-prone DNA polymerases or proteins that assist them , allowing survival of DNA damage by replication over or extension from otherwise-replication-inhibiting damaged DNA bases [18] , implying that most spontaneous mutagenesis in yeast results from use of error-prone DNA polymerases during DNA-damage survival . To our knowledge , the only mechanistic detail available on spontaneous mutagenesis mechanisms is that about half of all spontaneous base-substitutions and small insertion/deletions ( indels ) in starving Escherichia coli form dependently on the proteins used in stress-inducible mutagenic DNA break repair ( MBR ) [7] . In MBR , alternative error-prone DNA polymerases appear to be switched into the DSB-repair replisome under the control of stress responses [6–8] ( reviewed [13 , 15 , 16] ) . The central features of this mechanism have been recapitulated biochemically with purified proteins [19] . Repair of DSBs by homologous recombination requires high-fidelity replicative DNA polymerase III in unstressed cells [20] . In MBR , DSB repair uses alternative error-prone DNA polymerases , principally Pol IV ( DinB ) [21] , but also Pols II [22] and V [7 , 23] , under the control of the SOS DNA-damage response and the general/starvation stress response controlled by the RpoS transcriptional activator . The sequence signature of RpoS-dependent MBR is evident across bacterial genomes [24] , implicating this mechanism or similar mechanisms in most bacterial evolution . Moreover many stress-induced mutation mechanisms from bacteria to human cancer cells show similarities to E . coli MBR [13] . From these previous studies , it seemed probable that the critical step leading to MBR , and perhaps spontaneous mutagenesis generally , could be the switch from high-fidelity replicative DNA polymerases to error-prone DNA polymerases , resulting from DNA-polymerase competition promoted by upregulation of those DNA polymerases , in E . coli , by the SOS and general stress responses [13 , 15 , 16] . Here , we show that , for MBR , it is not sufficient to upregulate mutagenic DNA polymerases and repair DSBs; damaged DNA bases are also required . Reactive oxygen species ( ROS ) are among the most common DNA-damaging agents , and are known to promote mutagenesis when in excess [25–27] . ROS attack essentially all macromolecules in cells , including DNA , RNA , proteins , and lipids [28–30] . ROS include superoxide ( O−2· ) and its radical derivatives , which are detoxified by superoxide dismutases upregulated in E . coli by the SoxRS oxidative stress response [31] . ROS also include hydrogen peroxide ( H2O2 ) and its radical derivatives including hydroxyl radicals ( OH• ) , which are detoxified by catalases and alkylhydrogenperoxidases upregulated by the E . coli OxyRS oxidative stress response [30–32] . Oxidative damage to proteins is characterized by carbonylation ( reviewed by [33] ) , whereas ROS damage to DNA includes mainly base modifications 7 , 8-dihydro-8-oxo-deoxyguanosine ( 8-oxo-dG ) and thymine glycol , in addition to other modified bases and nicks [28 , 29] . The presence of 8-oxo-dG in DNA results both from incorporation of oxidized dGTP from the deoxynucleotide triphosphate ( dNTP ) pool and from in situ oxidation of guanine ( G ) in DNA [34] . 8-oxo-dG is often incorporated opposite to , and templates incorporation of , an adenine ( A ) leading to A to C and G to T transversion mutations ( e . g . , [35 , 36] ) . If 8-oxo-dG is incorporated into DNA , it can persist or be excised by base excision repair ( BER ) . Although replication bypass of template 8-oxo-dG has been shown to occur with high efficiency [35 , 37] , it is found that the eukaryotic replicative polymerase pol δ is transiently inhibited at 8-oxo-G [38] . Pol δ is able to extend from the A:8-oxo-dG base pair , but not from C:8-oxo-dG . A switch to an alternative polymerase , often pol λ , allows extension from C:8-oxo-dG with no substitution mutation [38] . In E . coli , of the alternative DNA polymerases , Pol IV is responsible for most MBR mutations [21] , and makes base substitutions and indel mutations , mostly 1 basepair deletions in mononucleotide repeats [39] . Pol IV is reported to make mutations in the absence of induced DNA damage when overproduced [39 , 40] , and SOS-upregulated levels of Pol IV are required for MBR [21 , 41] . Here , we report that SOS- and general-stress-response induced overproduction of Pol IV is not sufficient for Pol IV-dependent MBR . Damaged bases in the DNA must also be present . We hypothesize that damaged bases allow the switch to use of Pol IV during MBR by pausing the replicative polymerase to allow polymerase exchange in the replisome . The findings suggest that spontaneous mutation by MBR is regulated by regulation of the intracellular level of ROS .
The E . coli Lac assay [42] for MBR quantifies both indels and gross chromosomal rearrangements ( GCRs ) as reversions of a lacI-Z+1bp frameshift allele in an F’ conjugative plasmid . Revertants are scored as Lac+ colonies formed over days of starvation on solid lactose minimal medium ( e . g . , Fig 1A , WT ) . The leaky lac allele reverts either by compensatory frameshift mutations ( indels ) or by amplification ( GCRs ) [43] . When placed under general/starvation-stress-response inducing conditions , or if RpoS is upregulated artificially [6 , 7] , cells switch from faithful repair of DSBs by homologous recombination ( HR ) to an error-prone HR mechanism that requires the translesion DNA Pol IV , encoded by dinB [6 , 7 , 21 , 44] , leading to indel Lac+ revertants . The prevailing hypothesis for the GCR mechanism is that initial duplications of the lacI-Z+1bp allele are formed by a micro-homologous DSB-repair-instigated recombination event , followed by unequal crossing-over to give tandem arrays of 20 or more copies of the leaky lac allele [45 , 46] . Both formation of GCRs and single-nucleotide alterations ( SNAs , including base substitutions and indels ) via MBR require DSBs [4 , 6 , 7 , 47] , HR DSB-repair proteins RecA , RuvABC [48 , 49] , and RecBC [4] , activation of the general stress response [50 , 51] and , at some genomic loci including lac , the RpoE ( σE ) unfolded membrane protein response [52] . Formation of SNAs also requires the SOS response , which promotes SNA MBR by its 10-fold transcriptional upregulation of the DinB error-prone translesion polymerase [41] . SOS and DinB play no role in GCR formation [21] , which instead requires DNA polymerase I [45 , 53] . We reduced ROS levels in cells undergoing MBR using 2 , 2’-bipyridine ( bipyridine ) and thiourea ( TU ) , chemical agents commonly used to reduce ROS levels in living cells . Both agents enter cells and prevent ROS formation ( bipyridine ) or quench ROS ( TU ) [54] . Bipyridine chelates ferrous iron and prevents it from catalyzing ROS-forming Fenton reactions [55] . TU quenches hydroxyl radicals after formation , and prevents them from damaging macromolecules [56] . We added varying amounts of bipyridine ( 0 . 1 uM and 0 . 2 μM ) or TU ( 25 , 50 and 100 mM ) to solid minimal lactose medium on which MBR occurs , and found that Lac+ mutation rates were reduced dose-dependently by either ROS reducer , with a 90% ± 4 . 3% reduction at 100 mM TU ( Fig 1A and 1B ) ( p = 5 . 2x10-5 , Student’s 2-tailed t-test ) . TU also prevented the previously reported increase in MBR in cells lacking Dps [57] , a stationary-phase nucleoid-compaction protein that protects DNA against ROS , again in a dose-dependent manner ( Fig 1C ) . There is a 40% ± 7 . 8% reduction caused by 0 . 2 μM bipyridine ( Fig 1D and 1E ) ( p = 0 . 02 , Student’s 2-tailed t-test ) . Neither bipyridine nor TU affected viability of Lac- mutation-reporter cells over the duration of incubation , nor did they affect the time required for formation of Lac+ colonies under precise reconstructions of experimental conditions ( S1 Fig ) , indicating that mutation rate , not ability of treated cells to form colonies , was reduced by lowering ROS levels . All reductions in mutagenesis affected SNAs and amplification proportionately ( an example is shown in Fig 2A ) . These data imply that ROS are required for MBR and suggest that Dps might inhibit mutagenesis by preventing ROS damage to DNA [57] . We reduced ROS levels in cells undergoing MBR by constitutive overexpression of either of two E . coli oxidative stress responses: the SoxRS response and the OxyRS response [31 , 32] . Whereas each regulon responds to the presence of ROS in the cell , they employ different enzymes that detoxify different radical species . The SoxR response upregulates superoxide dismutases that inactivate superoxides and their derivatives [31] , and the OxyR response upregulates catalases and alkylhydroperoxidases that inactivate hydrogen peroxide and its radical derivatives including hydroxyl radicals [32] . We used the oxyR2 [58] and soxR104 [59] alleles ( separately ) , which cause constitutive expression of each response at induced levels , in the absence of oxidative inducers , to reduce ROS levels normally present during MBR . soxR104 reduced Lac+ MBR mutation rate , conferring a 90 ± 5 . 5% reduction ( Fig 2B and 2C ) ( p = 0 . 004 , Student’s 2-tailed t-test ) . Depression of MBR by oxyR2 ( 59 ± 11% ) was not quite significant at the 5% level ( p = 0 . 07 , Student’s 2-tailed t-test ) when tested in the wild-type strain . However , oxyR2 completely eliminated the Δdps-mediated increase in Lac+ MBR , showing a 70 ± 7 . 8% reduction in mutation frequency ( Fig 2B and 2D ) , ( p = 0 . 014 , Student’s 2-tailed t-test ) to the same level as the strain with oxyR2 alone ( p = 0 . 98 , Student’s 2-tailed t-test ) . The data indicate that ROS are required for MBR generally , and we infer that the mechanism of Dps inhibition of MBR is its protection of DNA against ROS . We removed ROS from cells by overexpressing sodB , which encodes a superoxide dismutase . We used a plasmid from the mobile plasmid library [60] that carries each gene under the control of an isopropyl β-D-1-thiogalactopyranoside ( IPTG ) -inducible promoter in cells used for the Tet MBR assay [7] . The Tet MBR assay reports on indel mutations that revert a chromosomal tetA +1bp frameshift allele during starvation in liquid minimal medium . In the Tet assay , DSBs are induced near a chromosomal tet mutation-reporter gene by the I-SceI site-specific double-strand endonuclease expressed in F-plasmid-free cells during starvation . tet reversion in this assay requires the key MBR proteins , as in Lac+ MBR , including RpoS , the SOS response regulators , Pol IV , and DSB-repair proteins [6 , 7 , 16] . The Tet assay differs from the Lac MBR assay not only in having no F-plasmids present , but also in having no selection for the reverted allele during the starvation stress; Tet-resistant ( TetR ) revertants are selected as TetR cfu after the cells are rescued from starvation in liquid . The Tet assay does not measure GCR . We found that over-expression of sodB reduced TetR MBR mutant frequencies 76% ± 12% , ( Fig 3 , 0 . 003 , Student’s 2-tailed t-test ) . These data show , by enzymatic removal of ROS , that ROS are required for starvation-stress-induced MBR . We conclude that MBR requires ROS for mutagenesis . We eliminated several possible mechanisms for the requirement for ROS in MBR . Repair of oxidized DNA bases might lead to formation of spontaneous DSBs that instigate MBR [4 , 6 , 7 , 47] . Base excision repair proceeds by excision of the damaged base by specific DNA glycosylases forming an abasic ( AP ) site , followed by nicking at that site by an AP endonuclease . If not repaired , the nick could produce a one-ended DSB when replicated , via replication fork breakage [61] . Using the Tet MBR assay , we find that even in the presence of an I-SceI-endonuclease-generated DSB , over-expression of sodB still reduced TetR MBR as described above ( Fig 3 ) . We show that induction of DSBs by I-SceI is effective when SodB is overexpressed ( S2 Fig ) . We conclude that ROS promote MBR other than or in addition to by promoting spontaneous DSBs . ROS damage proteins , lipids , RNA and DNA [30] . Oxidation of proteins , which causes carbonyl groups , can inhibit protein function ( reviewed by [33] ) . We used a mutation that reduces cellular levels of carbonylated proteins: rpsL141 , which encodes a hyper-accurate ribosomal protein that reduces intracellular carbonylated protein content [62] . We found that Lac assay MBR rates were unaffected by rpsL141 ( Fig 4A and 4B ) , implying that promotion of MBR by ROS is not mediated by oxidation of proteins . We excluded the possibility that ROS promote MBR by inducing the RpoS response by assaying MBR in an rssB mutant in the presence of TU . RssB targets RpoS , the transcriptional activator of the general stress response , for degradation by the ClpXP protease , so that when rssB is deleted , RpoS levels increase , artificially upregulating the RpoS response [64] . If ROS promoted MBR solely by inducing the general stress response , ΔrssB would be expected to counter the anti-MBR effect of TU treatment . We find that TU treatment reduced MBR dose-dependently in ΔrssB cells as in wild-type cells ( Fig 4C ) , ( p = 0 . 16; 0 . 92; 0 . 95 , comparing WT with rssB at 25 , 50 , and 100 mM TU , respectively; Student’s 2-tailed t-test ) . We conclude that ROS probably promote MBR other than or in addition to by activation of the general stress response . The SOS DNA-damage response promotes MBR via its 10-fold transcriptional upregulation of Pol IV error-prone DNA polymerase , and is not needed for MBR in cells with an operator-constitutive ( constitutively over-expressing ) dinB allele [41] . We find that the dinB ( oc ) allele did not counter TU treatment [Fig 4D , p = 0 . 001 for wild-type ( WT ) versus WT +100mM TU; p = 0 . 0003 comparing dinB ( oc ) with dinB ( oc ) +100mM TU , Student’s 2-tailed t-test] . By contrast wild-type and the dinB ( oc ) strain are not different [p = 0 . 95; and p = 0 . 10 for WT with TU versus dinB ( oc ) with TU , Student’s 2-tailed t-test] , implying that ROS promote MBR other than or in addition to by activation of the SOS response . We tested the possibility that excessive base damage from ROS might overwhelm mismatch repair , making it unable to correct base misincorporations by DinB . MutL , a required component of mismatch repair , was shown to become limiting during MBR [63] and for polymerase errors generated by a hyper-mutator Pol III [65] . We combined TU treatments with overproduction of MutL , to test whether oxidized DNA bases saturate mismatch repair capacity during MBR . Previously , increasing mismatch repair capacity by overproducing MutL reduced Lac+ MBR approximately 4-fold [63] . If ROS promoted MBR by saturating mismatch repair capacity , MutL overproduction and TU inhibition of MBR would be expected have an epistatic relationship , implying their action in the same pathway . By contrast , we found that 100mM TU combined with MutL overproduction reduced Lac+ MBR additively ( Fig 4E ) , P = 0 . 005 , 0 . 002 , 0 . 838 , and 0 . 035 for WT compared with MutL overproduction; WT compared with TU; WT + TU compared with MutL overproduction; and WT + TU compared with MutL overproduction + TU , respectively , Student’s 2-tailed t-test . The data imply that ROS promote MBR other than by causing saturation of mismatch repair . We found that removing oxidized guanine from DNA reduces MBR . E . coli has three proteins that specifically reduce the mutagenic effects of 8-oxo-dG . MutM is a DNA glycosylase that excises 8-oxo-dG from DNA , MutY is another glycosylase that excises mispaired adenine opposite to G or 8-oxo-dG , and MutT hydrolyses 8-oxo-dGTP in the nucleotide triphosphate pool , reducing incorporation of 8-oxo-dG into DNA . We ( separately ) overproduced MutM , MutT , and MutY in Tet MBR assay cells using mobile plasmid library plasmids [60] , and measured TetR mutant frequencies . We detected no decrease in cell viability during these experiments or in growth of strains containing these plasmids induced with 1mM IPTG ( S3 Fig ) . We found that overproduction of MutM and MutT reduced TetR MBR 82% ± 9 . 7% and 91% ± 5 . 3% , respectively , compared with the vector-only control ( p = 0 . 0015 and 0 . 0002 , Student’s 2-tailed t-test ) ( Fig 5 ) . ΔrpoS and dinB positive-control strains confirm that mutagenesis is via the canonical MBR pathway . In eight repeated experiments we observed a small , but not significant , decrease in Tet MBR with MutY overproduction ( 32% ± 9 . 5% decrease , p = 0 . 15 , Student’s 2-tailed t-test ) , suggesting that mispaired adenines have , at most , a minor effect on promotion of MBR . We conclude that persistent unrepaired 8-oxo-dG in DNA is required for MBR , and that ROS promote MBR via the presence of 8-oxo-dG in DNA . Furthermore , because the increased removal of 8-oxo-dG from DNA eliminates most mutagenesis , we conclude that whereas oxidized proteins or lipids , or oxidative lesions in RNA might play a role in MBR , it is at most a small one . The strong effect of overproduction of MutT , the 8-oxo-dGTP diphosphatase , implies that the MBR-promoting 8-oxo-dG in DNA results mainly from incorporation from the nucleotide pool rather than in situ oxidation of DNA . We tested the hypothesis the 8-oxo-dG in DNA might promote MBR by pausing the high-fidelity replicative DNA Pol III , active in DSB repair [20] , which might allow Pol IV and other alternative DNA polymerases to switch into the active position in the repair replisome . If this were the case , then 8-oxo-dG would not be required as a specific intermediate in MBR; any fork-stalling DNA-base damage traversed or extended from by translesion DNA polymerases would be expected to substitute for 8-oxo-dG . We tested whether other persistent DNA damage could bypass the requirement for ROS and 8-oxo-dG in MBR by using methyl methanesulfonate ( MMS ) , which methylates DNA , forming mainly N7-methyldeoxyguanosine and N3-methyldeoxyadenosine [66] , and ultraviolet-C light ( UV-C ) , which creates pyrimidine dimers and 6–4 photoproducts ( reviewed by [67] ) , all of which stall the replicative DNA Pol III and can be traversed by translesion DNA polymerases [68–70] . We pulse-treated starved cells with nonlethal doses of MMS or UV-C before plating on minimal lactose medium with 100mM TU to scavenge ROS . Whereas both MMS and UV-C increased basal mutation rates slightly ( 3-fold for each treatment , Fig 6A and 6B ) , the increase in mutagenesis seen with TU + MMS or UV treatments compared with TU alone was many-fold greater ( Fig 6C ) , 22 ± 7 . 8-fold for 10mM MMS ( p = 0 . 02 , Student’s 2-tailed t-test ) , and the increase is 16 ± 2 . 3-fold for 10 J/m2 UV , ( p = 0 . 001 , Student’s 2-tailed t-test ) . The data show a robust return to mutagenesis caused by those base-damaging agents after removal of ROS . The data imply that other base damages can substitute for 8-oxo-dG in mutagenesis . MMS or UV treatment completely countered the effect of 100 mM TU ( Fig 6B ) . Further , we show that the mutagenesis restored by MMS or UV in the absence of ROS is “on-pathway” MBR , in that mutagenesis depends completely on RecA , RpoS , Pol IV , and RuvC ( Fig 6D ) . These data show that MMS and UV did not activate an alternative mutagenesis mechanism ( s ) , but rather restored MBR to ROS-depleted TU-treated cells . Both MMS and UV-C might induce oxidative damage in addition to their more common lesions . UV-C can induce 8-oxo-dG formation in HeLa cells and in DNA , although 8-oxo-dG induction was about 2000-fold lower than pyrimidine dimer induction [71] . MMS induces ROS in yeast [72] , however , this requires the presence of glucose in the medium [73 , 74] , and glucose is absent from the medium of starved E . coli cells , making this unlikely . In Fig 1D we used a high dose ( 100mM ) of TU to scavenge ROS and found that the four-fold increase in MBR caused by Δdps is completely blocked by 100mM TU . Given the expected much greater levels of ROS in Δdps than WT cells and the ability of TU to quench the ROS in Δdps cells , these data suggest that there is still substantial ROS quenching capacity when TU is applied to wild-type cells . All of these data imply that MMS and UV-C promoted MBR not by promoting 8-oxo-dG in DNA , but rather by their abundant standard base damages . The slightly lower mutagenesis seen in the presence of TU for all mutagen treatments in Fig 6B ( p = 0 . 01 , 0 . 03 , 0 . 40 , and 0 . 20 , Student’s 2 tailed t-test for 5 J/m2 UV , 10 J/m2 UV , 5 mM MMS , and 10 mM MMS , respectively , in the presence versus the absence of TU ) might reflect a low level of MBR from oxidative damage when UV and MMS are used . However , most MBR in MMS or UV treated cells arises from non-oxidative base damage , which is not prevented by TU . These data demonstrate that persistent damaged bases in DNA are needed generally for MBR , and imply that the DNA substrate for MBR is more than a simple DSB undergoing HR repair: that , additionally , damaged DNA bases must be present in the DNA for MBR to occur .
We suggest a model in which the role of damaged bases in MBR is to allow the switch from high-fidelity DNA Pol III [20] to low-fidelity translesion DNA polymerases during homologous recombinational DSB repair , thus allowing MBR . A requirement for Pol III pausing to allow a switch to Pol IV has been shown previously [76–78] . A simple example of this idea is illustrated in Fig 7 for the formation of SNAs . The general model is that replication stalling at base damages that are not easily replicated or extended from by Pol III allows switching to alternative DNA polymerases , some of which are mutagenic , by pausing the replisome so that the DNA polymerases can switch . Damaged bases can be repaired in double-stranded DNA , but in single-stranded DNA at the replication fork they must be bypassed by translesion synthesis ( TLS ) [79] . TLS often uses alternative DNA polymerases to synthesize across damaged bases or to extend from the resultant mismatches after incorporation [80] . TLS polymerases include the Y-family polymerases Pol IV/DinB and Pol V in E . coli . DNA polymerases in the replisome and in repair complexes are attached to the β processivity clamp , which is the structural homolog of proliferating cell nuclear antigen ( PCNA ) in yeast and mammalian cells [81] . In both E . coli and eukaryotes , replicative polymerases might pause when they encounter a damaged base , either before incorporation or before extension , allowing a TLS polymerase that is already loaded on the β-clamp [77] to move onto the DNA primer end and synthesize past the damaged base or to extend beyond it . In the version of this general model shown in Fig 7 , we show Pol IV ( DinB ) replacing Pol III in the active position on the beta clamp ( Fig 7B ) . The replicative Pol III can incorporate 8-oxo-dG opposite A ( not illustrated ) or C ( Fig 7A ) [38] . Assuming that findings about eukaryotic polymerases [38] can be applied to E . coli , Pol III would then pause transiently because of poor ability to extend synthesis from the 8-oxo-dG paired with template C [38] . Alternatively Pol III could pause at an 8-oxo-dG in the template strand paired with C on the primer ( not illustrated ) [38] . In these cases , we suggest that stalling of Pol III allows it to be shifted out of the β processivity clamp site carrying the active DNA polymerase ( top of β-clamp doughnut shown Fig 7 ) , and a DNA polymerase in the clamp inactive position ( bottom of β-clamp , Fig 7 ) switched into the active position ( Pol IV shown switching places with Pol III , Fig 7B and 7C ) . With Pol IV in the β-clamp active DNA polymerase position , synthesis can resume from the 8-oxo-dG primer opposite template C ( red line , Fig 7C ) or from C across from 8-oxo-dG in the template ( not illustrated ) , and base-substitution and/or indel mutations can be made downstream after the switch , for example when Pol IV encounters a run of Gs ( Fig 7D and 7E ) , a slippery template sequence at which Pol IV makes errors with high probability [72] . Pol IV is processive for about 400 nucleotides [39] . Pol IV and its orthologous Y-family DNA polymerases in other species including human can make indel mutations because their active sites can accommodate extrahelical bases in mononucleotide repeats in the template strand , and thus insert fewer nucleotides than are on the template [82 , 83] . Their most common errors are base substitutions , however . The data presented here reveal that the availability of alternative DNA polymerases is not sufficient to allow mutagenesis , which requires polymerase switch , supporting models like this one . All three DNA damage-response ( SOS ) -induced polymerases , Pols II , IV and V , contribute to stress-induced MBR SNA formation [7 , 21–23] . Pol IV promotes 85% of indels in Lac MBR , and can increase mutation rates over a thousand-fold in the absence of exogenously induced DNA damage when overproduced in E . coli [40] . The remaining 15% of Lac MBR indels require Pol II [22] . Pol V promotes MBR base substitutions [23] and some indels [7] . GCR , seen as amplifications , does not require any of the SOS-inducible DNA polymerases , but requires DNA Pol I [45] . Thus all stress-induced MBR requires alternative DNA polymerases . Presumably the relative availability of the different DNA polymerases will influence which polymerases are switched into the MBR replisome and replace Pol III at the active position on the beta clamp . Amplification in the Lac assay is mediated by microhomology [45] and is proposed to arise by polymerase template switching during replication by a microhomology-mediated break-induced replication ( MMBIR ) mechanism [46 , 84] . Because Pol I is required for amplification [53] , we suggest that Pol I moves to the beta clamp active position when Pol III is stalled ( Fig 8B and 8C ) , and then mediates template switching creating genome rearrangements . We suggest that when Pol I binds the primer end and the replisome is dispersed ( Fig 8B , 8C and 8D ) , as will happen if the fork is stalled for a time [85] , Pol I can mediate template switching using very limited homology ( Fig 8D and 8E ) , as has been observed for the Pol I human ortholog DNA PolQ , another A-family polymerase [86–88] . This mechanism ( one specific example of which is illustrated in Fig 8 ) , parallels the model for base-substitution and indel formation ( one indel version shown in Fig 7 ) . Dissociation of the primer end with Pol I attached might be achieved if the editing function of Pol I is activated by the mismatch at ROS damaged bases , for example at an 8-oxo-dG:C base pair , causing dissociation of the 3'-end to enable it to attain the nuclease domain [89] . Such dissociation would permit polymerase template switching ( Fig 8E ) , the postulated first step in microhomology-mediated recombination . Other possible mechanisms that might have explained the role of ROS in MBR have been described , but do not fit the data presented here . First , because oxidation induces nicks and DSBs directly in DNA [90 , 91] , and both UV and MMS cause DNA breakage [92 , 93] , together with the established requirement for DSBs for MBR [6] , we considered the possibility that ROS were needed for MBR for the provision of DSBs . We answered this by showing that expression of I-SceI double-strand endonuclease did not raise MBR in a strain over-expressing the SodB superoxide dismutase to the level seen in a strain also expressing I-SceI but not over-expressing SodB ( Fig 3 ) . This shows that the loss of MBR associated with reduction in endogenous ROS is not suppressed by provision of DSBs , and therefore that ROS has a role in MBR other than or in addition to DSB formation . Similarly , the action of I-SceI did not suppress the reduction in MBR associated with over-expression of MutT ( Fig 5 ) , showing that removal of 8-oxo-dGTP from the deoxynucleotide triphosphate pool is not overcome by provision of DSBs . If ROS were required solely for DSB formation we would have seen MBR at wild-type levels in MutT over-expressing strains . Second , Pol IV overproduction causes cell death by increasing incorporation of 8-oxo-dG into DNA with subsequent increase in DSBs [54] . This is postulated to be because Pol IV is more likely than replicative polymerases to incorporate 8-oxo-dG , which is then removed by base-excision repair in 8-oxo-dG clusters leading to DSBs [54] . This model cannot account for the role of ROS/8-oxo-dG in MBR because the requirement for 8-oxo-dG was other than or in addition to for generation of DSBs ( Figs 3 and 5 ) . Also arguing against the Pol IV/8-oxo-dG-DSB model is the observation that amplification also requires ROS ( Fig 2A ) , and neither SOS-induced levels of Pol IV nor Pol IV itself are required for amplification [21] . Thus the model of Foti et al . [54] does not provide an explanation for the requirement for ROS for MBR . Third , a model for how single-stranded ( ss ) DNA damage promotes MBR in yeast also does not explain our data in E . coli . In yeast , ssDNA regions experience hypermutability both spontaneously and induced by UV light or MMS treatment [94] , with mutagenesis increased by orders of magnitude in ssDNA regions at sites of DSB repair by HR , compared with duplex DNA nearby . The ssDNA at sites of HR DSB repair and at uncapped telomeres involved ssDNA lengths longer than 6kb [95] . A similar mechanism is implicated in human cancer cells [12] . This hypermutability in ssDNA is attributed to the absence of complementary DNA sequence during repair of base lesions formed in the ssDNA , causing their replication by mutagenic translesion DNA polymerases [95] . This model appears not to apply to MBR in E . coli both because the E . coli MBR mutation clusters cover ~100 kb rather than six [8] , and because the model offers no explanation for GCR in MBR . Finally , ROS were reported previously to participate in an RpoS-independent mechanism of starvation-associated mutation in E . coli [96] . In contrast with our findings for RpoS-dependent MBR , in that assay , although mutating mutT , encoding the 8-oxo-G nucleotide pool sanitizer , increased mutagenesis , mutation of mutM , encoding the 8-oxo-dG DNA glycosylase , showed no significant effect , showing no role for 8-oxo-dG in DNA [96 , 97] , the opposite of our finding ( Fig 5 ) . Thus , a different mutagenesis mechanism was at work . Importantly , the mutagenesis-promoting effect of ROS and 8-oxo-dG in DNA reported here is in spontaneous mutation . Although increasing ROS in cells induces mutagenesis ( e . g . [25–27] ) , whether basal levels of spontaneous mutation are also driven by ROS was unknown previously . Previously , the sequence spectrum and frequencies of mutations in anaerobic and aerobic E . coli cultures suggested that spontaneous base substitutions were oxygen-related , via hydroxyl radicals [98] . The data here show directly that spontaneous SNA and GCR MBR in starving cells require ROS and 8-oxo-dG . A large body of work has established that organisms respond to stress by increasing their mutation rate under the control of stress responses [13] possibly increasing their ability to evolve and adapt [99] . The demonstration that mutation rate in stressed cells additionally requires base damage caused by ROS , together with our previous finding [57] that there is both positive ( by H-NS ) and negative ( by Dps ) regulation of ROS , suggest that cells might regulate their spontaneous mutation rates in response to challenging circumstances by regulating/responding to ROS levels . ROS were long regarded as unwanted byproducts of oxidative metabolism , to be avoided because of the damage that they do to macromolecules . They have been held to be a primary cause of aging [100 , 101] . However , it is now apparent that ROS are also an integral component of cell physiology [102] . The use of ROS for accelerated evolution in stressed cells described here might be regarded as a parallel with the involvement of ROS in signaling [103] and in the function of the immune system [104] . Conversely , although it might be advantageous for an organism to mutate specifically when under stress [99 , 104 , 105] , it would be to our advantage to stop pathogens from responding to our immune systems and antibiotics with hypermutation , evasion of the immune system and antibiotic resistance . We have suggested that novel “anti-evolvability” drugs might stop stress-induced evolution of pathogens ( and cancers ) by suppression of the stress-responses that promote mutagenesis [13] .
The origin of these strains is listed in Table 1 . Strains used for Lac+ MBR assays are isogenic derivatives of SMR4562 , an independent construction of FC40 [106] . Genotypes were confirmed either by PCR or by sequencing as necessary . Mobile plasmid library plasmids [60] containing over-expression alleles of mutants under the control of a Ptac promoter were conjugated into SMR10866 and SMR10798 , strains expressing the chromosomal I-SceI restriction endonuclease controlled by PBAD , and an I-SceI cutsite , or the enzyme-only control , respectively . Others were constructed by transduction or linear replacement [107] as listed in Table 1 . I-SceI cutting was verified by the sensitivity of strains to arabinose that induces I-SceI enzyme . Experimental procedures are as described [53] . The strains to be compared are grown at 37° for two days in M9 medium with 0 . 1% glycerol and thiamine , then mixed with a 25-fold excess of Δlac scavenger cells , and plated in top agar on M9 minimal lactose and thiamine solid medium . Plates are incubated at 37° , and Lac+ revertant colonies counted daily . About half of the colonies appearing on day 2 result from Lac+ mutant cells that arose during growth . Stress-induced mutant colonies appear linearly from day 3 onward . These experiments were continued to day 7 to obtain measurements of lac-amplification , which forms colonies later than SNA mutations [43] . The mutation rate ( mutants per cell per day ) is taken from the linear part of the curve as the mean and SEM of three or four parallel cultures of each strain . To determine cell viability during the prolonged starvation , the Lac- lawn is sampled at intervals by plating cells on complete medium containing rifampicin , which does not allow growth of the scavenger cells . 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside ( x-gal ) is included in the medium so that only Lac- cells are counted . In all experiments reported , Lac- viable cell counts did not vary significantly over the course of the experiment . Bipyridine and TU were added to the solid lactose minimal medium on which the starved cells were spread . Bipyridine was dissolved in ethanol . TU was dissolved in distilled water . Both solutions were filter-sterilized . Errors on all reported experiments are one standard error of the mean ( SEM ) of at least three independent experiments of three or four cultures per strain per experiment . To distinguish indel from amplified Lac+ cfu , samples of Lac+ colonies were replated onto rich medium containing x-gal dye on which SNAs produce solid blue colonies and amplified isolates produce sectored blue and white colonies [43] . MutL , expressed from the chromosome regulated by a PBAD promoter , was derepressed by absence of glucose in the medium during growth and on the minimal medium lactose plates . Experimental procedures are described in [7] . Four cultures of each strain are grown in M9 glycerol liquid medium with 20μg/ml carbenicillin for plasmid maintenance , 50μg/ml proline and with 0 . 1% glucose to repress PBAD , back-diluted after 12 hours twice and cultured for 84 hours . I-SceI endonuclease is induced by exhaustion of the glucose in the medium . Samples are then plated on complete medium with 0 . 1% glucose with and without tetracycline to obtain the TetR mutant frequency , and colonies are counted after one day . Activity of I-SceI was confirmed for all cultures during the experiments by their inability to grow with 0 . 0001% arabinose in M9 glycerol medium with IPTG ( S2A Fig ) . I-SceI cutting was confirmed for each strain by the loss of viability when plated on medium without glucose containing 0 . 001% arabinose ( S2B Fig ) . Error bars on all reported data are the SEM of at least 3 independent experiments . Genes carried in mobile plasmid library plasmids [60] were introduced to strains containing inducible DSB I-SceI cut-sites and enzyme and induced by adding 1mM IPTG to the liquid cultures during growth . A control strain containing an empty overexpression plasmid vector and an I-SceI cutsite and I-SceI enzyme cassette was also similarly treated with IPTG and plated to control for effects of plasmid expression on DSB induction and growth . Immediately prior to plating , aliquots of Lac-assay cells in prolonged stationary phase per the Lac MBR assay , were exposed to either UV-C irradiation or MMS . For MMS treatment , 1 mL of starved cell culture was pulse treated with MMS at 5 or 10mM , freshly diluted in water , for 20 minutes at 37°C . For UV-C irradiation , 1 mL aliquots of each culture were irradiated at either 5 J/m2 , or 10 J/m2 using a Stratalinker 2400 UV lamp emitting at 254nm . Viable cell count was determined after MMS and UV treatment to confirm that there was no loss of viability .
|
Mutagenesis mechanisms upregulated by stress responses promote de novo antibiotic resistance and cross resistance in bacteria , anti-fungal-drug resistance in yeasts , and genome instability in cancer cells under hypoxic stress . Stress-induced mutagenesis is implicated as the main source of spontaneous mutagenesis that drives bacterial evolution , and may drive much of evolution generally . A widely useful model mechanism is mutagenic DNA break repair in Escherichia coli , in which activation of two stress responses allows error-prone DNA polymerase in the break-repair replisome and introduce misincorporations , later fixed as mutations . Both stress responses upregulate the error-prone mutagenic DNA polymerase Pol IV ( DinB ) , suggesting that the regulation of mutagenesis to times of stress is accomplished by indirect gene upregulation , followed by DNA polymerase competition . This paper describes the discovery that the stress responses are not sufficient to allow mutagenesis caused by error-prone DNA polymerases in the break-repair replisome—damaged DNA bases must also be present—and that these are caused by reactive oxygen species in starving E . coli . We hypothesize that damaged bases may inhibit the progress of the highly processive and high fidelity replicative DNA polymerase , thus allowing the switch to error-prone DNA polymerases and mutagenesis . These findings suggest the possibility that the spontaneous mutation rate is regulated by the generation and removal of reactive oxygen , a very common byproduct of metabolism .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"cellular",
"stress",
"responses",
"ultraviolet",
"radiation",
"cell",
"processes",
"dna-binding",
"proteins",
"light",
"electromagnetic",
"radiation",
"dna",
"damage",
"polymerases",
"dna",
"replication",
"dna",
"proteins",
"physics",
"mutagenesis",
"biochemistry",
"dna",
"polymerase",
"cell",
"biology",
"nucleic",
"acids",
"genetics",
"biology",
"and",
"life",
"sciences",
"dna",
"repair",
"physical",
"sciences"
] |
2017
|
Persistent damaged bases in DNA allow mutagenic break repair in Escherichia coli
|
Mitogen-activated protein kinase ( MAP ) cascades are important in antiviral immunity through their regulation of interferon ( IFN ) production as well as virus replication . Although the serine-threonine MAP kinase tumor progression locus 2 ( Tpl2/MAP3K8 ) has been implicated as a key regulator of Type I ( IFNα/β ) and Type II ( IFNγ ) IFNs , remarkably little is known about how Tpl2 might contribute to host defense against viruses . Herein , we investigated the role of Tpl2 in antiviral immune responses against influenza virus . We demonstrate that Tpl2 is an integral component of multiple virus sensing pathways , differentially regulating the induction of IFNα/β and IFNλ in a cell-type specific manner . Although Tpl2 is important in the regulation of both IFNα/β and IFNλ , only IFNλ required Tpl2 for its induction during influenza virus infection both in vitro and in vivo . Further studies revealed an unanticipated function for Tpl2 in transducing Type I IFN signals and promoting expression of interferon-stimulated genes ( ISGs ) . Importantly , Tpl2 signaling in nonhematopoietic cells is necessary to limit early virus replication . In addition to early innate alterations , impaired expansion of virus-specific CD8+ T cells accompanied delayed viral clearance in Tpl2-/- mice at late time points . Consistent with its critical role in facilitating both innate and adaptive antiviral responses , Tpl2 is required for restricting morbidity and mortality associated with influenza virus infection . Collectively , these findings establish an essential role for Tpl2 in antiviral host defense mechanisms .
Mitogen-activated protein kinase ( MAP ) cascades represent major intracellular signaling pathways activated in response to a variety of external stimuli . Their activation during infection leads to transcriptional induction of immune and inflammatory mediators . Although MAP kinase signaling is important in eliciting host protective responses , many viruses are known to utilize these pathways directly for their replication [1] . Activation of MAP kinases occurs during virus recognition by pattern recognition receptors ( PRRs ) like toll-like receptors ( TLRs ) and RIG-I-like RNA helicases ( RLH ) [2] . Virus sensing by these receptors activates multiple intracellular signaling cascades including NFκB , MAP kinase and IRF pathways that coordinately regulate induction of interferons ( IFNs ) which are important mediators of antiviral resistance [3] . Among the MAP kinases , tumor progression locus 2 ( Tpl2/MAP3K8 ) , a MAP3 kinase , plays an important role in regulating IFN production by promoting the ERK-dependent induction of c-fos , a component of AP-1 heterodimeric transcription factors [4] . While Tpl2 is required for IFNα production by plasmacytoid dendritic cells ( pDCs ) and IFNγ secretion by CD4+ T cells , it is a potent negative regulator of IFNβ in macrophages and DCs [4 , 5] . Despite being identified as a major regulator of both Type I ( IFNα/β ) and Type II ( IFNγ ) IFNs , Tpl2 regulation of Type III IFNs ( IFNλs ) has not been investigated so far . Tpl2 was initially identified as an oncogene that induces T cell lymphomas in rodents [6] , but more recent studies have established its criticality in regulating both innate and adaptive immune responses via its cell type- and stimulus-specific activation of the MEK-ERK MAPK pathway . Tpl2 regulates signal transduction and cellular responses downstream of TLRs , cytokine receptors , antigen receptors and G protein-coupled receptors [4 , 7–9] . In addition to IFNs , Tpl2 also regulates the production of other prominent immune mediators like TNFα , IL-1β IL-10 , IL-12 and COX-2 [4 , 10–12] . Consequently , Tpl2 is essential for mounting effective immune responses during infections , and Tpl2-/- mice are more susceptible to Toxoplasma gondii [5] , Listeria monocytogenes [11] , Mycobacterium tuberculosis [13] and Group B Streptococcus [14] . Surprisingly , there is still limited and contradictory information about how Tpl2 contributes to host defense against viruses . Early studies reported normal cytotoxic T cell responses against lymphocytic choriomeningitis virus [10] and resistance to mouse cytomegalovirus infection [14] . However , another study delineating the signaling circuitry in virus sensing pathways implicated Tpl2 as a key regulator of both inflammatory and antiviral gene induction in response to model viral ligands [15] . A recent study also reported increased replication of vesicular stomatitis virus in Tpl2-deficient mouse embryonic fibroblasts ( MEFs ) [16] . We recently demonstrated that among the TLRs implicated in virus sensing ( TLRs 3 , 7 and 9 ) , Tpl2 plays a prominent role in TLR7 signaling [17] . In this study , we investigated Tpl2’s regulation of antiviral responses using a murine model of influenza virus infection , which relies upon TLR7 for virus sensing [18] , ERK MAP kinase for virus replication [19] and where both IFNα/β and IFNλ are host protective [20] . Our experiments demonstrate positive regulation of IFNλ and cell-type specific regulation of IFNα/β production in Tpl2-deficient cells following stimulation with model viral ligands that trigger influenza virus sensing receptors , TLR7 or RIG-I . However , during influenza virus infection , IFNλ uniquely required Tpl2 for its induction . Moreover , Tpl2 is involved in IFN signaling , regulating ERK activation and STAT1ser727 phosphorylation , and is required for proper induction of antiviral IFN-stimulated genes ( ISGs ) . Impaired ISG induction coupled with reduced antigen-specific CD8+ T cells resulted in failure to control virus replication and significant morbidity and mortality of Tpl2-/- mice to an otherwise low pathogenicity strain of influenza virus . Collectively , this study establishes Tpl2 as a host factor that integrates antiviral responses to control influenza virus infection .
To determine whether Tpl2 regulates influenza virus replication , wild type ( WT ) and Tpl2-/- mice were infected with 104 plaque forming units ( pfu ) of mouse-adapted influenza virus A/HK-X31 ( H3N2 ) ( X31 ) , and viral titers in the lungs were evaluated on days 3 , 5 and 7 post infection ( pi ) . The average lung viral titers were significantly higher in Tpl2-/- mice compared to WT mice at all time points examined ( Fig 1A ) . Notably , average viral titers were more than ten-fold higher in Tpl2-/- lungs at day 7 pi . This increase in virus replication was also observed in littermate control mice ( S1 Fig ) . In addition to viral titers , early proinflammatory cytokines , except TNFα were significantly higher in the BALF of Tpl2-/- mice compared to WT mice ( Fig 1B ) . Consistent with increased virus replication , total cellular infiltration was also significantly increased in the lungs of Tpl2-/- mice at day 7 pi ( Fig 1C ) . The increased lung viral titers in Tpl2-/- mice early after infection on day 3 suggest a critical role for Tpl2 in limiting virus replication during influenza virus infection . Airway epithelial cells are the primary targets for influenza virus infection . Early studies after the discovery of Tpl2 demonstrated high levels of Tpl2 expression in the lungs [21] . Moreover , similar to hematopoietic cells , Tpl2 regulation of signal transduction and cytokine gene induction was also demonstrated in airway epithelial cells [22] . To elucidate whether Tpl2 functions in hematopoietic or nonhematopoietic cells to limit virus replication , we assessed lung viral titers in chimeric mice in which WT or Tpl2-/- bone marrow cells were transferred into either WT or Tpl2-/- irradiated recipients . At day 3 pi , average lung viral titers were significantly higher in Tpl2-/- mice reconstituted with WT hematopoietic cells ( Fig 1D ) . In contrast , there was no statistically significant increase in viral titers of WT mice that received Tpl2-/- bone marrow ( Fig 1E ) . These data demonstrate that Tpl2 signaling within radioresistant , nonhematopoietic lung cells is necessary for limiting virus replication early after infection . Interferons are induced early during infection and are key factors initiating host protective antiviral responses [3] . To determine whether the observed increase in viral titers in Tpl2-/- mice is due to defective induction of IFNs , WT and Tpl2-/- mice were infected with 106 pfu X31 virus , and IFNα/β/λ levels in lung homogenate or BALF were measured at day 1 or day 3 pi . Induction of both IFNα and β were comparable between WT and Tpl2-/- lung homogenates and BALF ( Fig 2A ) . Notably , IFNλ secretion was significantly reduced in Tpl2-/- mice following influenza virus infection ( Fig 2B ) . Surprisingly , while IFNλ was induced to a higher level compared to Type I IFNs in WT mice , there was minimal induction in Tpl2-/- mice in response to infection at both time points . Reduced IFNλ production in Tpl2-/- mice was independent of viral titers which were similar between WT and Tpl2-/- mice at day 1 pi ( S2 Fig ) . Despite differences in IFNλ induction , total cellular infiltration and IFNγ levels in BALF were significantly elevated in Tpl2-/- mice compared to WT mice at day 3 pi ( S3 Fig ) . The observation that Tpl2 is uniquely required for IFNλ , but not IFNα or IFNβ , production in influenza-infected lungs is especially significant , because IFNλ is regarded as the principal IFN induced during influenza virus infection . Airway epithelial cells and pDCs are considered the major sources of IFNs during respiratory virus infections , including influenza [20 , 23] . Although we observed a decrease in IFNλ levels in Tpl2-/- mice at day 1 pi , a more consistent and significant reduction was observed at day 3 pi , which corresponds to the migration of pDCs to infected lungs [23] . Since Tpl2 is required for macrophage and neutrophil migration during acute inflammation [9 , 24] , we investigated whether Tpl2 similarly regulates the recruitment of pDCs to the infected lung . The reduction in IFNλ levels in influenza-infected Tpl2-/- mice was not due to impaired recruitment of pDCs ( S4 Fig ) . To investigate whether defective IFN induction by pDCs contributes to the reduced IFNλ in BALF from Tpl2-/- mice during influenza infection , bone marrow-derived pDCs ( CD11c+B220+CD11b- ) from WT and Tpl2-/- mice were infected with influenza virus A/WSN/1933 ( H1N1 ) , and the production of IFNα , β and λ was assessed . Consistent with in vivo infections , the levels of both IFNα and IFNβ were comparable between WT and Tpl2-/- cells , whereas IFNλ secretion was significantly less in Tpl2-/- pDCs infected with influenza virus ( Fig 2C ) . A similar reduction in IFNλ induction was also observed in Tpl2-deficient cells infected with X31 influenza virus strain ( S5 Fig ) . Collectively , these data demonstrate the unique requirement for Tpl2 in IFNλ production during influenza infection in vitro and in vivo . During influenza virus infection , receptors from both TLR and RLR families recognize viral PAMPs and trigger rapid induction of IFNs . Recognition of viral components by PRRs typically occurs in respiratory epithelial cells , alveolar macrophages , DCs and pDCs in a cell type-specific manner [25] . The major receptors involved in recognition of influenza virus are TLR7 , which recognizes single-stranded viral RNA , and RIG-I , which recognizes the 5’-triphosphate of single-stranded RNA genomes ( 5’ppp-RNA ) . The single-stranded RNA genome is recognized through endosomal TLR7 in pDCs [18] in contrast to epithelial cells and DCs where virus recognition is mediated primarily by the cytosolic sensor RIG-I [26] . We therefore investigated whether differential regulation of IFN production observed during infection is due to differences in Tpl2-mediated sensing by PRRs . MEFs and bone marrow-derived macrophages ( BMDMs ) from WT and Tpl2-/- mice were either transfected with the RIG-I ligand 5’ppp-RNA or stimulated with the TLR7 ligand R848 [27] , and IFNβ production was measured by ELISA . Consistent with previous studies using the TLR4 ligand LPS [4] , IFNβ production was significantly increased in Tpl2-/- cells treated with both 5’ppp-RNA and R848 ( Fig 3A–3C ) . This increase in IFNβ correlated with impaired ERK phosphorylation in Tpl2-deficient cells in response to these ligands ( S6 Fig ) . Unlike epithelial cells and DCs , virus recognition in pDCs is mediated via TLRs rather than RLHs , and Type I IFN production occurred normally in RIG-I-deficient pDCs infected with RNA viruses [18 , 26] . To determine whether Tpl2 regulates TLR7-mediated IFN production by pDCs , bone marrow-derived pDCs from WT and Tpl2-/- mice were treated with the TLR7 ligand , R848 , and IFN levels were quantitated . Consistent with previous studies using the TLR9 ligand CpG [4] , and in contrast to BMDMs , secretion of both IFNα and IFNβ were significantly decreased in culture supernatants from Tpl2-/- pDCs treated with R848 ( Fig 3D ) . Notably , IFNλ secretion was also significantly less in Tpl2-/- pDCs compared to WT cells in response to R848 ( Fig 3D ) . Unlike Ifna but similar to NFκB-regulated Il12p40 and Tnfa [28] , IFNλ3 ( Il28b ) transcription occurred early , by 2 hr of stimulation ( S7 Fig ) . Collectively , these data demonstrate that Tpl2 differentially regulates IFN production downstream of PRRs involved in influenza virus sensing in a cell type-specific manner . The importance of IFNλs in host protection against many viruses is well established , however , the mechanisms that regulate their production are largely unexplored . Common mechanisms have been postulated to regulate Type I and III IFNs during viral infections [29 , 30] . Despite their importance in mediating Type I IFN production in pDCs [4 , 31] , the significance of MAP kinase and PI3 kinase cascades in murine IFNλ production has not been directly investigated . In order to elucidate the potential mechanism by which Tpl2 regulates IFNλ production in pDCs , we evaluated the involvement of ERK and PI3K-mTOR signaling in IFNλ induction . Tpl2 regulation of both ERK and mTOR-Akt signaling in different cell types has been reported previously [8 , 32–34] . In addition to the MEK/ERK pathway [4] , we demonstrate that Tpl2 also promotes mTOR/Akt signaling in pDCs as determined by a decrease in the proportion of phospho-Akt+ pDCs in the absence of Tpl2 signaling ( Fig 4A and 4B ) . To confirm whether ERK , PI3K or mTOR signaling also contributes to IFNλ production in pDCs , cells were pre-treated with rapamycin ( mTOR inhibitor ) , LY294002 ( PI3K inhibitor ) or U0126 ( MEK inhibitor ) 30 min prior to TLR stimulation , and CpG-induced IFNλ secretion was measured by ELISA . CpG was used as the stimulant in these experiments because TLR9 ligation induced higher levels of IFNλ compared to TLR7 stimulation with R848 . Pharmacological inhibition of each of these signaling pathways significantly reduced IFNλ secretion to the levels observed in Tpl2-/- cells ( Fig 4C ) . In contrast , only a modest reduction in IFNλ induction was observed in Tpl2-deficient cells treated with rapamycin or U0126 ( S8 Fig ) . These results demonstrate the significance of Tpl2 and both MAPK and PI3 kinase signaling cascades in regulating IFNλ production in pDCs . Robust production of Type I IFNs in pDCs is dependent upon IRF7 and autocrine IFN signaling , and consequently IFNα secretion is abrogated in both Irf7-/- and Ifnar1-/- pDCs [35] . Similar to IFNα , and as reported previously [20] , IFNλ production was abolished in Ifnar1-/- pDCs infected with influenza virus ( Fig 5A ) demonstrating the absolute requirement for IFNAR signaling in IFNλ secretion by pDCs . Induction of IFNλ in response to direct IFN stimulation has been reported in hepatocyte carcinoma HepG2 cell lines [36] . Although a high dose of IFNβ could induce modest IFNλ secretion , the levels induced were lower than that induced by TLR-stimulation , demonstrating that IFN/IRF7 signaling alone is not sufficient for driving high levels of IFNλ secretion ( Fig 5B ) . Nevertheless , Tpl2 contributed to IFNAR-induced IFNλ production , since significantly less IFNλ was secreted by Tpl2-/- pDCs directly treated with IFNβ ( Fig 5B ) . In addition to demonstrating the role of Tpl2 in IFNAR-mediated IFNλ production , these data also suggest a role for Tpl2 in directly transducing Type I IFN signals . To determine whether Tpl2 regulates IFNλ production in influenza virus-infected lungs directly via virus sensing pathways or indirectly via IFNAR feedback signaling , we assessed IFNλ levels in lung homogenates from mice that are deficient in both Tpl2 and IFNAR1 . Consistent with reduced IFNλ levels in BALF from Tpl2-/- mice day 3 pi ( Fig 2B ) , IFNλ levels were similarly reduced in day 3 lung homogenates ( Fig 5C ) . IFNλ levels were significantly decreased in Ifnar1-/-Tpl2-/- compared to Ifnar1-/- mice , demonstrating that Tpl2 promotes early IFNλ induction independent of Type I IFN signaling ( Fig 5D ) . Notably , the level of IFNλ induction was similar in Tpl2-/- and Ifnar1-/-Tpl2-/- mice ( Fig 5C and 5D ) . In striking contrast to the abrogation of IFNλ production in Ifnar1-/- pDCs ( Fig 5A ) , IFNλ production occurred normally in Ifnar1-/- mice ( Fig 5D ) . Consistent with the critical role of IFNAR signaling in IFNα induction , IFNα levels were significantly diminished in both Ifnar1-/- and Ifnar1-/-Tpl2-/- mice ( Fig 5E ) . These data demonstrate that Tpl2-dependent IFNλ production during influenza virus infection is IFNAR-independent . Both IFNα/β and IFNλ are known to induce expression of ISGs that establish an antiviral state in infected tissue to prevent virus replication and spread [3 , 37] . Because of the observed increase in early virus replication in Tpl2-/- mice ( Fig 1A ) , we questioned whether Tpl2 regulates the induction of ISGs . We first addressed whether Tpl2 regulates IFN signaling . BMDMs from WT and Tpl2-/- mice were stimulated with IFNα or IFNβ , and activation of downstream cascades , especially STAT1 , which is the principle regulator of IFN responses , was evaluated by immunoblotting . BMDMs were used in these experiments due to limited availability of pDCs . The phosphorylation of STAT1 Tyr701 and Ser727 , which is necessary for maximal STAT1 transcriptional activation , were examined [38] . While phosphorylation of Tyr701 occurred normally in Tpl2-deficient cells in response to stimulation with Type I IFNs , a consistent reduction in Ser727 phosphorylation was observed in Tpl2-/- cells compared to WT cells ( Fig 6A–6C ) . In addition to the classical JAK-STAT pathway , signaling via the Type I IFN receptor also activates other downstream cascades including MAP kinases [39] . Despite the existence of multiple MAP3 kinases , Tpl2 has an essential , non-redundant role in transducing ERK activation signals during TLR , TNF- and IL-1-receptor signaling [7 , 8] . We therefore investigated whether Tpl2 is similarly required for ERK activation during Type I IFN signaling , or whether other MAP3Ks could fulfill this role . ERK phosphorylation was strongly induced by both IFNα and IFNβ . Importantly , ERK phosphorylation was absent in Tpl2-/- BMDMs stimulated with IFNα/β demonstrating an absolute requirement for Tpl2 in transducing ERK activation signals in response to Type I IFNs ( Fig 6A ) . Of note , unlike LPS- and TNFα-treated BMDMs and similar to poly I:C- , CpG- , and IL-1β-treated BMDMs [17 , 40] , no mobility shift ( indicative of phosphorylation ) or degradation of the p58 isoform of Tpl2 was detected following stimulation with Type I IFNs ( Fig 6A ) . Consistent with our previous studies [41] , both Tpl2 protein and mRNA expression were induced upon either Type I IFN stimulation or influenza virus infection ( Fig 6A and S9 Fig ) . Overall , these data demonstrate that Tpl2 contributes to Type I IFN signaling . Since Tpl2 is known to modulate the antiviral transcriptome [16] , we next investigated whether the induction of ISGs in infected lungs is impaired in the absence of Tpl2 . The induction of Ifitm3 , Isg15 and Oasl2 , ISGs known to be important in limiting influenza virus infection [25] , were measured . We observed a modest , but statistically significant decrease in Ifitm3 and Oasl2 expression in Tpl2-/- compared to WT mice infected with influenza virus ( Fig 6D ) . A trend towards reduction in Isg15 expression was also noted in Tpl2-/- mice ( Fig 6D ) . In addition to infected lungs , the induction of Oasl2 was significantly reduced in Tpl2-/- BMDMs , while the expression of Ifitm3 and Isg15 was largely unaffected by Tpl2 ablation ( S10 Fig ) . These data demonstrate that Tpl2 promotes the induction of ISGs in influenza-infected lungs to limit virus replication . Although Tpl2 is important in transducing Type I IFN signals , this would not alone account for the increase in viral titers or reduction in ISGs observed in Tpl2-/- mice , since either Type I or Type III IFN is sufficient for protection during influenza virus infection . This is because both types of IFNs drive redundant amplification loops inducing the expression of similar antiviral genes [42] . To investigate whether IFNAR signaling contributes to the observed increase in virus replication , we next assessed lung viral titers in mice deficient in both Tpl2 and IFNAR1 . Consistent with previous studies [20] , viral titers were comparable between WT and Ifnar1-/- mice ( Fig 6E ) . Although average lung viral titers were significantly higher in Ifnar1-/-Tpl2-/- mice compared to both WT and Ifnar1-/- mice ( Fig 6E ) , the titers were similar to those observed in Tpl2-/- mice ( Fig 1A ) . These data demonstrate that Tpl2 restricts early virus replication in an IFNAR-independent manner . Even though the observed reduction in ISGs helps to explain the early increase in viral titers , a more pronounced and biologically significant increase in viral titers was observed at day 7 pi which correlates with the recruitment of influenza-specific CD8+ T cells to the lungs [43] . Since many seminal studies have identified CD8+ T cells as the major mediators of influenza virus clearance from infected lungs [44 , 45] , we investigated whether virus-specific CD8+ T cell responses are impaired in Tpl2-/- mice . Consistent with defective viral clearance observed in Tpl2-/- mice , induction of protective nucleoprotein ( NP ) -specific CD8+ T cells [46] was significantly reduced in BAL cells from Tpl2-/- mice compared to WT animals ( Fig 7A and 7B ) . In addition , antigen-specific secretion of IFNγ was also decreased in BAL cells from Tpl2-/- mice ( Fig 7C ) . During the course of this experiment , we unexpectedly observed severe clinical signs in Tpl2-/- mice despite the fact that the mice were infected with the low pathogenicity A/HK-X31 ( H3N2 ) ( X31 ) influenza virus . To confirm whether Tpl2 ablation alters the susceptibility to influenza virus infection , WT and Tpl2-/- mice were infected with 104 pfu of X31 virus , and weight loss and clinical symptoms were monitored over a period of 14 days . All Tpl2-/- mice exhibited severe clinical signs and succumbed to infection by day 10 pi , whereas all WT animals survived and returned to pre-infection body weights by day 14 pi ( Fig 7D and 7E ) . Similar to infection with X31 virus , Tpl2-/- mice infected with the virulent PR8 [A/Puerto Rico/8/34 ( PR8; H1N1 ) ] strain showed increased disease severity compared to WT mice , although not to the same extent seen with the low pathogenicity virus ( S11 Fig ) . Body weights were collected to day 10 pi , at which time the Tpl2-/- mice met the humane endpoints of the study . At this time point , the body weights were just beginning to show the characteristic switch between the WT and Tpl2-/- mice , such that the Tpl2-/- mice were showing more severe clinical signs of disease . Accordingly , systemic pro-inflammatory cytokine levels were also increased in the Tpl2-/- mice at day 10 pi . Analysis of BAL cells also showed decreased antigen-specific CD8+ T cell responses in Tpl2-/- mice compared to WT mice at this late time point , consistent with the observations with X31 infections . Collectively , these data demonstrate the critical role of Tpl2 in promoting viral clearance and restricting morbidity and mortality associated with influenza virus infection .
Tpl2 is now appreciated to regulate the induction of Type I and Type II IFNs as well as other cytokines that may contribute to antiviral responses . However , there is very limited information on how Tpl2 coordinates antiviral immune responses in vivo . In this study , we demonstrate Tpl2’s obligate role in promoting antiviral responses and viral clearance during influenza virus infection . These findings are important because influenza virus is a ubiquitous seasonal virus that afflicts millions of people annually , causing significant morbidity , mortality and socio-economic burdens [47] . Therefore , understanding the role of host factors like Tpl2 in restricting morbidity and mortality associated with influenza virus infection is critical for developing disease intervention strategies . Mechanistically , Tpl2 promotes the induction of ISGs and virus-specific CD8+ T cells that facilitate viral clearance as shown in the proposed model ( Fig 8 ) . Thus , the findings reported here establish an essential role for Tpl2 in host protective innate and adaptive antiviral responses . Tpl2 deficiency led to cell-type specific alterations in the regulation of Type I IFN production . Specifically , IFNβ production was increased in response to LPS , R848 and the RIG-I ligand , 5’-triphosphate RNA in Tpl2-/- MEFs and BMDMs . In contrast , Type I IFN was significantly reduced in pDCs in response to TLR7 stimulation with R848 . This differential regulation of Type I IFN production by Tpl2 in different cell types in response to TLR ligands is consistent with a previous report by O’Garra and colleagues [4] . Importantly , we also demonstrated that Tpl2 similarly functions as a negative regulator of Type I IFN production upon activation of the RIG-I cytosolic sensor with 5’-triphosphate RNA . One striking observation was the absolute requirement for Tpl2 in the TLR-dependent induction of both Type I ( IFNα/β ) and Type III IFNs ( IFNλ ) by pDCs . The fact that pDCs uniquely require Tpl2 for production of both Type I and Type III IFNs suggests that pDCs differ fundamentally from BMDMs and MEFs in their signaling pathways . Indeed , impaired IFN production correlated with reduced activation of the PI3K/Akt signaling pathway in Tpl2-/- pDCs . This finding is also consistent with the observation that the PI3K/Akt pathway appears to be especially important in driving TLR-dependent IFN expression by pDCs [31] . In addition to cell-type specific regulation , Tpl2 also differentially regulates the production of Type I and Type III IFNs during viral infection . Importantly , influenza virus has been reported to utilize the Raf pathway to activate ERK , which explains why Type I IFN induction occurs in a Tpl2-independent manner in mice and pDCs infected with influenza virus [48] . On the contrary , IFNλ production was uniquely dependent upon Tpl2 during the course of influenza infection in vitro and in vivo . Although Type I and Type III IFNs have common regulatory elements in their promoters and are usually co-expressed in response to viruses and TLR ligands [36] , selective induction of IFNλ by transcription factors NFκB and IRF1 has been reported [49 , 50] . The distinct requirement for Tpl2 in IFNλ induction in virus-infected pDCs likely represents the unique requirement of the IFNλ promoter for an early NFκB-dependent priming event . In support of this , our own data demonstrate that IFNλ induction is rapid and parallels the regulation of NFκB-dependent genes more closely than IFNα ( S7 Fig ) . With the exception of a recent study reporting that p38 , but not ERK , is required for Ifnl1 expression in human cells [49] , the roles of MAPK or PI3K pathways in the regulation of IFNλs have not been evaluated . Although the regulation of IFNλ1 by PI3K-mTOR is still unexplored , our data demonstrate a different mechanism of IFNλ3 regulation that relies on the Tpl2-ERK pathway in contrast to the p38-dependent regulation described for IFNλ1 . Therefore , in addition to transcription factors [30] , diverse signaling cascades also specify induction of different IFNs . The complexity of the IFN response is not completely understood , since multiple signaling cascades and transcription factors activated during IFN signaling can independently or cooperatively regulate the transcriptional response to IFNs [39] . Importantly , our data demonstrate the involvement of Tpl2 in IFN signaling leading to the phosphorylation of ERK and STAT1Ser727 . Previous studies have demonstrated the significance of STAT1Ser727 phosphorylation for full transcriptional activation and induction of ISGs [38 , 51] . Conflicting reports exist regarding the identity of the serine kinase responsible for STAT1Ser727 phosphorylation; different kinases including p38 , ERK and PKC-δ have been implicated [52–54] . Importantly , an association of ERK with STAT1 and a requirement of ERK activity for expression of ISGs have been demonstrated [55] . Tpl2 regulation of STAT1Ser727 phosphorylation and induction of ISGs might be indirect via its regulation of ERK phosphorylation during IFN signaling . In addition to regulating ISG transcription , Tpl2-ERK signaling also regulates the phosphorylation and dissociation of the translation initiation factor 4E-Bp-eIF4E complex , which is involved in cap-dependent translation of many genes , including ISG15 [34 , 56] . Therefore , the Tpl2-ERK pathway regulates the biological effects of IFNs at the transcriptional level and possibly also at the posttranscriptional level . Although MAP kinase pathways are known to be activated in response to IFNs , the importance of Tpl2 in regulating IFN-inducible effectors has not yet been described . The induction of ISGs is mainly attributed to IFN-stimulated gene factor-3 ( ISGF3; consists of STAT1 , STAT2 and IRF9 ) . In addition to ISGF3 , IRF7 can also act independently to regulate transcription of antiviral genes , and Tpl2 has been shown to promote IRF7-dependent transcription [16] . However , normal induction of IFNα/β during influenza virus infection argues against a major role for IRF7 in the observed phenotype , since IRF7 is regarded as the ‘master regulator’ of Type I IFN induction [35] . To understand the mechanism by which Tpl2 exerts its antiviral effect , we examined the contribution of Tpl2 to virus replication in different cellular compartments and in the context of IFNAR deficiency . Using bone marrow chimeras , we demonstrated that Tpl2 was required within the nonhematopoietic compartment to restrict early virus replication . This likely reflects Tpl2 functions in airway epithelial cells , the primary target of influenza virus . In this regard , Tpl2 is known to be expressed and to regulate inflammation within airway epithelial cells [22] . Studies using Ifnar1-/-Il28ra-/- mice have also demonstrated that interferon responsiveness of these cells is critical for restricting early viral replication [42] . It is well known that abrogation of Type I IFN signaling does not increase influenza virus replication due to the presence of compensatory Type III IFNs [57] . Consistent with this , we observed that Tpl2 ablation promoted virus replication to the same extent on both Ifnar1+/+ and Ifnar1-/- genetic backgrounds . The 50% reduction in IFNλ levels that we observed in Tpl2-/- mice on day 3 pi is unable to explain the increase in virus replication , because compensatory Type I IFNs are induced to normal levels . Furthermore , the presence of IFNs , rather than quantity , is important in driving antiviral responses [42] . One possible explanation for the increased viral replication in Tpl2-/- mice is that Type III IFN signaling is also Tpl2-dependent , like we have demonstrated for Type I IFNs . Additional studies using Il28ra-/- mice are needed to determine the contribution of Tpl2 to Type III IFN signaling . In addition to antiviral innate responses , we also identified a critical role for Tpl2 in the induction of antigen-specific CD8+ T cell responses . This is in contrast to a recent study reporting a major role for Tpl2 in human , but not murine , CD8+ T cell responses [58] . The impaired induction of virus-specific CD8+ T cells resulting in defective viral clearance and increased mortality in Tpl2-/- mice clearly warrants detailed studies on Tpl2 regulation of effector CD8+ T cell responses . The increased mortality observed in Tpl2-/- mice infected with X31 virus was surprising because infection with this low pathogenicity virus does not typically cause severe clinical signs or mortality in mice . Even though IFNλ production was impaired in Tpl2-/- mice , this defect is not sufficient to explain their increased morbidity and mortality , because several studies have shown that either Type I or Type III IFN alone is sufficient to limit influenza virus infection [20 , 42 , 59] . In addition to impaired CD8+ T cell responses [45] , the reduction in expression of some ISGs may also contribute to the enhanced pathogenesis , since defective expression of individual antiviral factors , like IFITM3 , can alter the course of infection [60] . Early increases in virus replication in Tpl2-deficient lung stromal cells , demonstrated by bone marrow chimera experiments , coupled with defective viral clearance by CD8+ T cells likely potentiate the inflammatory response , which is considered a major factor contributing to morbidity and mortality during pathogenic influenza infection [61] . Overall , our study establishes Tpl2 as a host factor with intrinsic ability to restrict influenza virus replication and also demonstrates immune regulatory functions of Tpl2 within the lungs . The involvement of Tpl2 in major virus sensing pathways as well as antiviral signaling cascades suggests a key role for Tpl2 in integrating antiviral responses . These results are especially significant considering a very recent study demonstrating the requirement of IRF7 as well as Type I and Type III IFNs , all regulated by Tpl2 , in protecting humans from life-threatening influenza virus infection [62] . Whether Tpl2 similarly restricts the replication of other classes of viruses requires further investigation . The findings reported here also suggest that therapeutic inhibition of Tpl2 during chronic inflammatory diseases might predispose patients to viral infections .
All animal experiments were performed in accordance to the national guidelines provided by “The Guide for Care and Use of Laboratory Animals” and The University of Georgia Institutional Animal Care and Use Committee ( IACUC ) . The Institutional Animal Care and Use Committee ( IACUC ) of the University of Georgia approved all animal experiments ( Assurance Number A3437-01 ) . Wild type ( WT ) C57BL/6J ( CD45 . 2+ ) mice were purchased from The Jackson Laboratory . Tpl2-/- mice backcrossed to C57B6/J were kindly provided by Dr . Philip Tsichlis ( Tufts University ) and Thomas Jefferson University . For some experiments , littermate control WT and Tpl2-/- mice were obtained by interbreeding Tpl2+/- mice . Ifnar1-/- mice were kindly provided by Dr . Biao He ( University of Georgia ) . Mice deficient in both IFNAR1 and Tpl2 were generated by interbreeding single knockout animals . To generate chimeric mice , WT or Tpl2-/- recipient mice ( both CD45 . 2+ ) were lethally irradiated with 1100 rad and reconstituted with donor B6 . SJL-PtprcaPepcb/BoyJ ( WT CD45 . 1+ congenic ) or Tpl2-/- bone marrow cells . Chimeric mice were maintained for 8 weeks . Animals were housed in sterile microisolator cages in the Central Animal Facility of the College of Veterinary Medicine . Embryonated specific pathogen free ( SPF ) chicken eggs were purchased from Sunrise Farms , New York . Influenza viruses A/HKX31 ( H3N2 ) , A/Puerto Rico/8/34 ( PR8; H1N1 ) and A/WSN/1933 ( H1N1 ) stocks were propagated in the allantoic cavity of 9- to 11-day-old embryonated SPF chicken eggs at 37°C for 72 hr , and viral titers were enumerated by plaque assays [63] . Age-matched , 6- to 8-week-old WT , Tpl2-/- , Ifnar1-/- , Ifnar1-/-Tpl2-/- or chimeric mice were anesthetized with 250 mg/kg Avertin ( 2 , 2 , 2-tribromoethanol ) followed by intranasal infection with influenza A/HK-X31 ( H3N2 ) in 50 μl PBS . Control mice were mock-infected with a similar dilution of allantoic fluid . To determine lung viral titers , whole lungs from WT and Tpl2-/- mice infected with 104 pfu of X31 virus were harvested on days 3 , 5 and 7 pi . Lungs were placed in 1 ml PBS and dissociated with a bead mill homogenizer ( Qiagen ) , and virus titers were enumerated by plaque assays . To assess susceptibility to influenza infection , WT and Tpl2-/- mice infected with 104 pfu of X31 virus were observed over a period of 14 days . Body weights were recorded daily , and mice exhibiting severe signs of disease or more than 30% weight loss were euthanized . To measure IFN and cytokine secretion , mice infected with 106 or 104 pfu of X31 virus were euthanized 3 or 7 days pi , and bronchoalveolar lavage fluid ( BALF ) or BAL cells were obtained by washing the lungs twice with 1 mL PBS . Cells were recovered by centrifugation of the lavage fluid for 10 min at 250xg . BALF from the first wash was used for quantitation of cytokine secretion . Cellular recruitment was assessed by quantifying total leukocyte recovery from both washes . Mice infected with 104 pfu of X31 virus were euthanized on day 10 pi , and cells were obtained by washing the lungs twice with 1 mL PBS . BAL cells were stained with anti-CD4 , anti-CD8 ( eBiosciences ) , and H2DbNP366–374 tetramer ( NIH Tetramer Core Facility , Emory University , Atlanta , GA ) for 30 min at 4°C and fixed in 1% formaldehyde . Cells were acquired on a BD LSRII flow cytometer and analyzed using FlowJo software ( Tree Star , Inc . ) . For IFNγ measurement , BAL cells were stimulated with a cocktail of influenza immunodominant peptides ( NP366–374 , PA224–233 , PB1703–711 ) ( 1 μg/mL ) for 24 hr at 37°C , and IFNγ levels in culture supernatant was measured by ELISA ( eBiosciences ) . Bone marrow derived macrophages ( BMDMs ) , pDCs and mouse embryonic fibroblasts ( MEFs ) were generated from age- and sex-matched mice as described previously [17 , 64] . CD11c+CD11b-B220+ pDCs were sorted using a Beckman Coulter MoFlo XDP cell sorter . In some experiments , cells were used on day 10 without sorting ( referred as Flt3 ligand-derived DCs ) . Triggering of RIG-I was accomplished by directly delivering 5’-triphosphate RNA ( 5’ppp-RNA; 0 . 5 μg/mL ) or a control RNA to the cytosol of BMDMs or MEFs using LyoVec transfection reagent ( InvivoGen ) . 20 μL 5’ppp RNA or control RNA ( 100 μg/mL ) was incubated with 200 μL LyoVec ( 62 . 5 μg/mL ) at room temperature for 15 min to form complexes . Twenty-five microliters of the complexes were used to stimulate 2 . 5x105 BMDMs or 0 . 5x105 MEFs per well for 24 hr . BMDMs at 1x106/mL were also treated with R848 ( InvivoGen ) ( 1 μg/mL ) for 24 hr . To investigate IFN signaling , BMDMs at 1x106/mL were treated with rmIFNα ( 2000 IU/mL; R&D Systems ) , or rhIFNβ ( 10 ng/mL; Peprotech ) for 1–4 hr . Plasmacytoid DCs at a concentration of 0 . 5-1x106/mL were left untreated or stimulated with R848 ( 1 μg/mL ) , CpG ODN2395 ( 10 μg/mL ) ( InvivoGen ) , 50 ng/mL rhIFNβ ( PeproTech ) or infected with WSN virus at a MOI of 0 . 2 for 24 hr . In some experiments , cells were pretreated with LY294002 hydrochloride ( 20 μM ) , rapamycin ( 30 nM ) or U0126 ( 20 μM ) ( Sigma ) for 30 min before stimulating with CpG . Cytokine levels were measured by ELISA ( IFNα , IFNλ and IFNγ , eBioscience; IFNβ , PBL Interferon Source ) or bead-based detection assays ( Mouse IFNα Flowcytomix simplex , eBioscience; Mouse inflammation cytokine bead array , BD Biosciences ) . Cells stimulated with R848 or IFNs were lysed using TRK lysis buffer ( Omega Bio-Tek ) . For in vivo infections , RNA lysates were prepared from tissue after homogenizing whole lungs . RNA was extracted using a Total RNA Kit ( Omega Bio-Tek ) . Real-time PCR was performed after synthesizing cDNA using a High capacity cDNA Reverse Transcription kit ( Applied Biosystems ) . The expression of Irf7 ( Mm00516791_g1 ) , Il28b ( ifnl3 ) ( Mm00663660_g1 ) , Ifitm3 ( Mm00847057_s1 ) , Isg15 ( Mm01705338_s1 ) , Oasl2 ( Mm00496187-m1 ) , Il12b ( Mm00434174_m1 ) , Il6 ( Mm00446190_m1 ) , Tnfa ( Mm00443258_m1 ) , Ifna ( Mm03030145-gH ) , Ccl5 ( Mm01302427-m1 ) and Actinb ( 4352341E-1112017 ) were determined by RT-PCR ( Applied Biosystems ) . RT-PCR reactions were performed in microAmp Fast plates ( Applied Biosystems ) using SensiFAST Probe Hi-ROX kit ( Bioline ) and a StepOnePlus RT-PCR machine ( Applied Biosystems ) . Relative gene expression levels were calculated by normalizing the Ct levels of the target gene to both endogenous actin levels and an unstimulated WT control using the ΔΔCt method . Cell lysates were separated on 4–12% gradient gels ( Invitrogen ) and were transferred to PVDF membranes using the iBlot Gel Transfer system ( Invitrogen ) . Membranes were probed with various antibodies followed by horseradish peroxidase ( HRP ) -labeled secondary antibodies . Protein bands were visualized by enhanced chemiluminescent reagent ( Lumigen ) and Amersham Hyperfilm ECL ( GE Healthcare ) . The following antibodies were used for immunoblotting: Tpl2 ( Cot M-20 ) , ERK1 , ERK2 and β-actin ( Santa Cruz Biotechnology ) , p-ERK1/2 ( Thr202/Tyr204 ) , p-STAT1 ( Tyr701 ) , p-STAT1 ( Ser727 ) and STAT1 ( Cell Signaling Technology ) . Cells harvested after overnight stimulation were fixed , permeabilized with triton buffer ( PBS+0 . 5%triton+0 . 1%BSA ) and stained for p-Akt ( Ser473 ) according to manufacturers’ protocol ( Cell Signaling Technology ) . Samples were acquired on a BD LSRII flow cytometer and analyzed using FlowJo software ( Tree Star , Inc . ) . Data represent means ± SEM , except where indicated . P-values were determined by Students t-test , and significance was assigned for p-values <0 . 05 . Kaplan-Meier analysis using PRISM software was performed to estimate percentage survival of WT and Tpl2-/- groups infected with influenza virus , and p value was determined using a Mantel-Cox test .
|
Influenza viruses infect millions of people annually causing significant morbidity , mortality and socio-economic burdens . Host immune responses against influenza virus are initiated upon virus recognition by specific intracellular receptors . Signals relayed from these receptors trigger various signaling cascades , which induce an antiviral immune response to control infection . Herein , we identified the serine-threonine kinase tumor progression locus 2 ( Tpl2 ) as an essential component of virus sensing pathways , regulating induction of interferons ( IFNs ) and IFN-induced antiviral genes that restrict virus replication . We also demonstrate that Tpl2 is necessary for generation of effector CD8+ T cells , which are required for viral clearance from infected lungs . Consistent with the impaired antiviral responses , Tpl2-deficient mice are defective in controlling virus replication and succumb to influenza virus infection with a normally low pathogenicity strain . Thus , our study identifies Tpl2 as a host factor that integrates antiviral innate and adaptive responses to restrict morbidity and mortality during influenza virus infection .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Tumor Progression Locus 2 Promotes Induction of IFNλ, Interferon Stimulated Genes and Antigen-Specific CD8+ T Cell Responses and Protects against Influenza Virus
|
Antibodies are thought to play an essential role in naturally acquired immunity to malaria . Prospective cohort studies have frequently shown how continuous exposure to the malaria parasite Plasmodium falciparum cause an accumulation of specific responses against various antigens that correlate with a decreased risk of clinical malaria episodes . However , small effect sizes and the often polymorphic nature of immunogenic parasite proteins make the robust identification of the true targets of protective immunity ambiguous . Furthermore , the degree of individual-level protection conferred by elevated responses to these antigens has not yet been explored . Here we applied a machine learning approach to identify immune signatures predictive of individual-level protection against clinical disease . We find that commonly assumed immune correlates are poor predictors of clinical protection in children . On the other hand , antibody profiles predictive of an individual’s malaria protective status can be found in data comprising responses to a large set of diverse parasite proteins . We show that this pattern emerges only after years of continuous exposure to the malaria parasite , whereas susceptibility to clinical episodes in young hosts ( < 10 years ) cannot be ascertained by measured antibody responses alone .
Naturally acquired immunity to malaria is a complex and poorly understood process , by which individuals living in P . falciparum endemic areas develop protection against clinical and symptomatic infections over years of repeated exposure . Since the first experimental evidence demonstrating how passively transferred immunoglobulins from immune adults can dramatically reduce parasitaemia in infected recipients [1 , 2] there has been a growing body of evidence that antibody ( Ab ) responses play an important role for parasite control and protective immunity . However , the unambiguous identification of the target antigens involved has been difficult , and even after decades of research there is still no strong consensus about which candidates could be considered as potential components of an anti-asexual stage vaccine . Prospective cohort studies , in which individuals’ immune responses against panels of P . falciparum-specific antigens at time zero are related to their subsequent risk of developing clinical malaria , have frequently shown how responses to various antigens correlate with increased protection against clinical malaria in an age- and/or exposure dependent manner [3–14] . Proteins expressed by the merozoite life-stage of P . falciparum , such as the merozoite surface protein ( MSP ) or apical membrane protein ( AMA ) , are often the focus of such studies , partially due to their higher sequence conservation compared to other immunogenic but highly polymorphic variant surface proteins ( e . g . PfEMP1 ) that are expressed during the intra-erythrocytic life-stages of the parasite . The protective potentials of anti-merozoite antibodies have been confirmed in in vitro and animal studies , which led to those antigens now being considered as potential vaccine targets ( see e . g . [15] for a review ) . However , their contribution to clinical immunity in a field-setting is yet to be quantified . Small effect sizes and the difficulty in reliably quantifying previous exposure [16] makes the distinction between markers of exposure and markers of protective immunity problematic and has resulted in inconsistent and contradictory findings in the past [17] . More importantly , though , routine analytical approaches based on comparisons between population-level mean responses often fail to convey information about the robustness of the derived associations and how sensitive they are to even small changes in the observed data . The shortcomings of traditional statistical methods are highlighted when trying to predict individual-level protection from population-wide associations . In particular , when dealing with high dimensional data , where a vast number of combinations and interactions must be tested . Here , practitioners typically rely on univariate tests , whilst adjusting for common markers of exposure , thus ignoring potential interplay between different antigens . Conversely , predictive modelling frameworks based on machine learning offer a systematic way to consider all possible combinations of immune responses against various antigens . These hypothesis-free approaches do not assume a priori functional relationships between the measured variables ( e . g . Ab-levels ) and the response ( e . g . the risk of a clinical episode ) , and test whether these associations could be due to chance ( i . e . the ubiquitous P-value ) . Instead , the outcome of interest is the predictive accuracy , i . e . the degree by which the model can predict the response at the level of the individual . They further provide a better understanding of the contribution of individual predictors towards model performance . Thus , machine learning techniques have become popular choices for the analysis of high dimensional datasets in biology and ecology ( see e . g . [18–23] ) . Here we used a random forests machine learning approach to analyse antibody profiles against panels of P . falciparum-specific antigens with the aim to identify signatures that are predictive of an individual’s protective status against clinical malaria episodes . Our results show that immune signatures that clearly distinguish clinically immune individuals can be found only when considering a broad set of antigens from individuals spanning a sufficiently wide age-range , whereas the responses taken from young cohorts are less likely to be informative of the individual’s susceptibility to malaria .
We analysed previously published data from three prospective cohort studies conducted in Kenya , Kenya/Tanzania and Mali , which we simply refer to here as KEN , KTZ and MAL , respectively . The datasets can be found as supporting information ( S1 Data , S2 Data and S3 Data ) . The underlying studies are described in detail elsewhere [8 , 12 , 24] and summarised in Table 1 , so here we just provide a brief overview . The KEN dataset contains immune profiles for 286 individuals . However , in line with the original study [12] , our analysis was performed on the subset of children who were parasite-positive at screening ( N = 121 , age = 1-10 years ) . Immune profiles are ELISA-based antibody titres against 36 P . falciparum-specific antigens , taken at the start of the transmission season , with host age and schizont extract reactivity used as exposure proxies . The response variable was incidence of a clinical malaria episode , defined as an axillary temperature of > 37 . 5°C , plus any parasitaemia for children less than 1 year , and an axillary temperature of > 37 . 5°C , plus parasitaemia > 2500/μl for individuals older than 1 year , during a 6-months follow-up . The KTZ dataset is based on luminex-derived IgG levels against 46 individual PfEMP1 domains of 447 children ( 5-18 months old , mean = 11 . 4 months ) living in Kilifi ( Kenya ) and Korogwe ( Tanzania ) , taken from the placebo arm of the RTS , S malaria vaccine trial [25] . Individuals were followed for an average of 8 months with multiple samples taken over the time course , resulting in a total of 1269 immune profiles . The outcome of interest was the incidence of at least one clinical malaria episode , defined as an axillary temperature of > 37 . 5°C plus parasitaemia > 2500/μl . Age and bednet use were used as proxy variables for exposure risk . The MAL dataset comprises protein microarray-based antibody reactivity of 186 individuals aged 2-25 years against a panel of 2320 P . falciparum-specific epitopes of the 3D7 line , representing 1204 unique proteins ( ∼ 23% of the P . falciparum proteome ) , taken before the start of the transmission season . The response variable was incidence of clinical malaria , defined as axillary temperature of > 37 . 5°C plus parasitaemia > 5000/μl , over an 8-months period of follow-up . Age and infection status ( parasite positive or negative ) at first screening were used as exposure proxy variables . To identify predictive immune signatures underlying clinical protection we employed a random forests [26] machine learning approach , using the randomForest package [27] in R [28] . Each dataset ( KEN , KTZ , MAL ) was analysed separately . As input variables for our models we used the measured immune profiles and , where applicable , their respective exposure proxies . The response variable , i . e . the outcome to be predicted , was the incidence of clinical infections during follow-up as defined by the respective studies . For this work we classified individuals as susceptible if they had a recorded episode of clinical malaria within the specified time window , and protected otherwise . Unavoidably , protected individuals also included those who may have been uninfected . However , this scenario is probably minimised by the fact that these studies were conducted in villages of moderate to high transmission . Random forests are able to deal with data sets where the number of predictors is larger than the sample size . However , when the number of features greatly outweighs the number of samples , such as in the MAL dataset ( sample size N = 186 , number of features M = 2320 ) , feature selection is advised to remove uninformative variables and focus on the ones that exhibit sufficient predictive power [29 , 30] . We start by first ignoring strongly linearly correlated responses ( above a Pearson correlation coefficient of ρ = 0 . 8; which corresponds to around 36% of the original feature set ) , to avoid biasing the variable importance measures computed by the random forests [31] . Recall that only around half of the measured responses represent unique proteins ( see Table 1 ) . Note that these correlated variables are reintroduced in the interpretation stage if any of the features they are associated with have been selected for the final model . The remaining covariates undergo a rigorous supervised feature selection process , based on the mProbes [32] and xRF [33] algorithms , as follows: A detailed layout of the random forests model fitting procedure for the MAL dataset can be found in S1 Text; S1 Script provides the R code to run our feature selection procedure . We used two measures to assess and report on the models’ predictive accuracies: ( i ) the receiver operating characteristic ( ROC ) curve , which is generated by plotting the true positive rate against the false positive rate ( i . e . the observed incidence against the false predicted incidence ) at various threshold settings . The area under the curve ( AUC ) is a measure of predictive accuracy , with an AUC = 1 equating to zero error and an AUC = 0 . 5 equating to random guessing; and ( ii ) by means of a confusion matrix , which contrasts the instances of the predicted classes ( protected or susceptible ) against the observed classes . The misclassification-rates are based on the so-called out-of-bag ( OOB ) errors [27] . These are derived by iteratively testing the model’s performance against subsets of data left out during the fitting process ( recall that each decision tree in a random forests is built on a bootstrapped sample of the original data ) . OOB errors represent an estimate of the generalisation error , that is , how well the model would fare against previously unseen data . For the MAL dataset , the OOB error computed on a model fitted only on the selected features would be over-optimistic due to selection bias [34] . Instead , we use our feature selection algorithm inside a five-fold cross-validation loop . Within each fold , the model fitted using the selected features ( for each fold the number of selected features may vary ) is tested against the left out fold . We report the average AUC across all folds .
We first analysed two datasets comprising ELISA and Luminex-derived antibody profiles against P . falciparum—specific antigens obtained from prospective cohort studies in Kenya ( KEN ) and Kenya / Tanzania ( KTZ ) ( see Material and methods ) . For each dataset , we used a random forests ( RF ) machine learning approach to predict individual-level protection against clinical immunity over a specified period of time based on ( i ) measured antibody levels ( Ab ) , ( ii ) proxies for exposure ( such as age or bednet use ) ( Exp ) , and ( iii ) all measured variables ( Ab and Exp ) . Fig 1 illustrates the outcome of this analysis by two measures of predictive accuracy: the receiver operator characteristic ( ROC ) curves ( Fig 1A and 1C ) and illustrated confusion matrices ( Fig 1B and 1D ) . To our surprise , Ab-levels considered in these studies , including those against previously proposed vaccine targets , such as MSP1 or AMA1 , are poor predictors of individual-level protection , with misclassification rates of up to 56% . A weak signal for protective immunity could be found in the KEN dataset using both antibodies and exposure variables . However , a null-model based solely on exposure proxies ( age and schizont extract , blue line in Fig 1A ) was equally predictive of an individual’s risk of malaria as the more complex model that also included their immune profiles ( green line , Fig 1A ) . High misclassification was also found in the KTZ cohort data ( Fig 1C and 1D ) , which we believe was mainly due to the very young host ages in this study ( between 4-18 months at recruitment ) , where individuals were still experiencing their first malaria infections . Therefore , exposure did not contribute to the model’s predictive performance . What these results demonstrate is that even in the case where some responses might show univariate statistically significant associations with protection at the population-level , their effect sizes are too small to be able to predict whether someone with elevated titres will be protected during the next transmission season or not . It is equally possible that the data was simply too limited with respect to the age-range and/or the specificity and number of ( allelic ) antigens considered in the respective assays to identify immune signatures that clearly distinguish clinically immune individuals . The protein-microarray-based MAL dataset comprises a much broader set of antigens , representing over 1000 unique proteins , and contained individuals of two different age classes: children between the age of 2-10 years ( mean = 5 . 8 , N = 149 ) , which were mostly classed as susceptible ( 117/149 ) , and young adults between 18-25 years ( mean = 20 . 8 , N = 37 ) , who had predominantly acquired a state of clinical protection ( 33/37 individuals remained symptom-free ) . Fig 2A shows the mean antibody measures against all antigens stratified by either age or infection outcome . In both cases we find a consistent , qualitative shift in the immune profiles of protected and older individuals where sets of high-titre antibodies are further elevated and sets of low-titre responses further reduced . Importantly , and as shown in the beanplots [35] in Fig 2B , there is no notable difference in the mean reactivity between the various strata ( in all three cases P > 0 . 2 , Welch two sample t-test ) , implying that clinical protection within this dataset is not characterised by an increase in overall reactivity . To elucidate the most predictive antigens at the individual-level in this dataset , where the number of immune responses ( M = 2320 ) was much larger than the number of samples ( N = 186 ) , we performed feature selection inside a five-fold cross-validation loop ( see Material & methods and S1 Text ) . We first fitted a model to all individuals ( 2-25 years ) using all available predictors ( 2320 antigenic responses , age and parasite status at screening ) . The average AUC across all testing folds was 0 . 83 ( Fig 3A ) , exhibiting very good discrimination between the protected and susceptible individuals . However , this predictive performance dropped considerably when the model was identified using only the children ( 2-10 years ) , average AUC across four of the five folds was 0 . 56 ( Fig 3B ) . For one of the cross-validation folds , no predictors were retained by the feature selection algorithm , so no model could be built . Fig 4 shows a heatmap of immunoreactivity using the set of antigens selected in at least one of the cross-validation folds , and a word cloud for the protein product description based on the selected features and any other proteins they are highly correlated with ( ρ > 0 . 8 ) . The set of predictive antigens consists of those with either an increasing or a decreasing response as individuals grow older and gain clinical protection . Interestingly , none of the 47 proteins which showed an increasing response with age were correlated with others . Whereas , 232 proteins were correlated with the 32 responses that exhibited a decreasing response with age . This temporal pattern , as well as the strong relationship between age and immune status , suggests that the change in the immune responses that allows us to distinguish susceptible from clinically immune individuals ( at least as measured by the protein microarray ) is taking shape through continuous exposure to the malaria parasite during childhood , but does not fully develop until early adulthood . In part , this explains why we see such an extensive degradation to model performance when only the children are considered , and why 79 antigens were selected for the model with all individuals , but only 9 when considering only the children . Moreover , whilst age was an important predictor for all individuals ( coming up in all five-folds ) , it was never selected for the model based on only the children . The word cloud suggests that the responses that increase with age are related to surface variable antigens , whilst the ones that decrease consist of a number of conserved proteins of ( currently ) unknown function . S4 Data contains the complete list of selected antigens and their annotations .
Despite decades of intensive research , we still do not know which antigens are central in the induction of immune responses that protect against clinical malaria . Small effect sizes of individual responses and the often polymorphic nature of many immune targets , coupled with considerable inter-individual heterogeneity , are partially to blame for the lack of a clear relationship between a measured response and the level of protection against malaria it offers . Osier and colleagues [3 , 12] have previously proposed that protection could in fact be due to the accumulation of responses against sets and or specific combinations of antigens in a threshold-dependent manner , which would certainly help to overcome the issue of individual responses having small effect sizes . However , the number of combinations and possible interactions between antigens that need to be tested and compared across studies to draw robust conclusions can easily become infeasible using standard statistical approaches . Mainly due to identifiability issues , where insufficient samples are available to estimate each model parameter . The predictive modelling framework based on machine learning described here offers a systematic approach to consider all possible combinations of measured antibodies and to extract the most distinguishing features from these high-dimensional datasets in a hypothesis-free way . Furthermore , as the outcome of this approach is the predictive accuracy , i . e . the degree by which the model can predict the response at the level of the individual , results are easily comparable across studies , in contrast to P-values , which are strongly dependent both on sample size and the chosen statistical test and/or model . Our analysis of different sets of cohort data based on immune profiles against relatively limited sets of P . falciparum-specific antigens demonstrated that commonly assumed immune correlates and potential vaccine candidates ( e . g . MSP-1 , MSP-2 , or AMA-1 ) are poor predictors of clinical protection in children . Apart from small effect sizes and the fact that in most studies only a limited set of target alleles are investigated , the age-range of individuals considered in these studies may also play a role in explaining our findings . That is , previous studies have shown that protection against life-threatening disease might be acquired early in life after only a few infections ( see e . g . [36] ) . On the other hand , clinical protection is a more gradual process whereby the probability of a clinical episode declines slowly under repeated exposure . This may make the identification of immune signatures that are highly predictive at the individual-level problematic , unless the considered age ranges , and therefore the levels of cumulative exposure , are sufficiently broad . What our results also point towards is that a diverse set of antigens must be considered to robustly identify predictive immune signatures . The distinct pattern that we found in the data based on protein-microarrays was characterised by an exposure-driven change in the responses to several surface-expressed and internal , conserved and polymorphic parasite proteins . Importantly , a small subset of antigens was sufficient to predict an individual’s risk of presenting with a symptomatic malaria infection during the following transmission season with a high degree of accuracy , at least within a set of individuals covering a suitably wide age-range . In contrast to previous studies , the difference in the immune profiles between the two phenotypes consisted of both increased and decreased responses to certain antigens , which intensified with age . Whereas those responses which showed increasing intensities with exposure contained various known immune targets , such as PfEMP1 , the responses that decreased with age were mostly against proteins of unknown function . It is possible that these decreasing responses are an artefact of the microarray data as they mostly consisted of low-titre responses ( see S1 Text ) , i . e . those with small signal-to-noise ratios . In their original analysis , Crompton et al . [8] only considered responses immunogenic if they were 2 SDs above the controls , whereas we considered all responses and are therefore more likely to pick up immunologically irrelevant features . Further investigations are therefore required to verify if and to what degree these responses relate to ( protective ) immunological pathways . The fact that the predictive responses showed such a strong association with age also begs the question whether they are true targets of the protective responses or whether they simply mirror infection histories; the latter is probably more plausible for the internal rather than surface-expressed antigens . Finding reliable markers of previous/cumulative exposure is arguably one of the most fundamental problems in correctly identifying antibody-based correlates of protection . Not only are they important in assessing how many undetected infections get past passive and/or active surveillance during follow-up periods , but the ability to discern between cases where someone is truly protected , i . e . infected but without showing clinical symptoms , or simply has a lower exposure risk , would allow us to define reliable phenotypes . A more direct approach would be the inclusion of only those individuals with documented infections [37] . However , this relies on a much higher sampling frequency or reliable markers of recent infections . Age , homestead location and previous malaria incidence are common markers , but none of these directly quantify how often the immune system was challenged by a P . falciparum infection . Using a predictive framework , Helb and colleagues [21] recently established an alternative way for estimating recent exposure by identifying key antigens that were most predictive of days since the last P . falciparum infection and incidence of symptomatic malaria during the previous year . The immune signature identified in our analysis is not so much indicative of recent infections but more of repeated challenges over years of continuous exposure . In order to make our findings and methodological approach relevant , not only for understanding the process of natural acquired immunity to malaria , but also with regards to future intervention measures , including vaccines , our results need to be validated against independent datasets . In the first instance these should involve replicate studies in similar transmission settings , including the follow up of individuals over successive transmission seasons to test for the robustness of the identified immune signatures . More importantly , though , is the issue of differentiating between individuals who are protected and those who did not get challenged . In the studies considered here this was not a major concern as they were all based in moderate to high transmission settings with most children experiencing a clinical episode over the study periods . However , in areas of lower transmission intensities this is a pressing concern . One obvious way around this would be the use of longitudinal cohort studies with active surveillance . Another , and much cheaper option would be to exploit data analytical approaches as employed by Helb and colleagues [21] . A good correlate of protection should be a universal biomarker reflecting an immune response that prevents the parasite from causing clinical and life-threatening disease . Our results need to be validated against independent datasets to investigate whether the predictive signatures we identified perform equally well in different malaria endemic settings and to test to what degree they truly capture functional immune mechanisms . In that respect we believe that predictive models offer clear advantages over univariate association analyses . Not only is predictive accuracy directly comparable between studies , but these frameworks also provide a systematic way to consider all putative correlates of protection whilst reducing the chances of false discoveries . With the advent of more detailed and complex big data sets in the field of malaria immuno-epidemiology these models should therefore be considered more prominently alongside standard statistical approaches in an attempt to unravel the complex interplay between exposure and infection outcome in P . falciparum malaria .
|
Understanding naturally acquired immunity against P . falciparum malaria is of fundamental importance for malaria control and elimination efforts . The identification of parasite antigens that could potentially be considered as vaccine targets often relies on prospective cohort studies where observed infection rates are related to measured immune responses . However , what is unknown , is how these population-level associations between antibody titres and protection from severe malaria can predict the risk of an infection for an individual . We therefore analysed three sets of cohort-based immune profiles using a machine learning approach in order to identify distinct immune signatures that are predictive of protection at the individual-level . Our results show that even statistically significantly associated responses fail to provide robust information about an individual’s risk of malaria and that machine learning approaches should be considered more prominently alongside traditional methods for analysing these complex and high dimensional datasets .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"immunology",
"tropical",
"diseases",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"protozoans",
"forecasting",
"mathematics",
"statistics",
"(mathematics)",
"artificial",
"intelligence",
"antibodies",
"research",
"and",
"analysis",
"methods",
"immune",
"system",
"proteins",
"computer",
"and",
"information",
"sciences",
"malarial",
"parasites",
"proteins",
"mathematical",
"and",
"statistical",
"techniques",
"immune",
"response",
"biochemistry",
"machine",
"learning",
"eukaryota",
"immunity",
"physiology",
"biology",
"and",
"life",
"sciences",
"malaria",
"physical",
"sciences",
"statistical",
"methods",
"organisms"
] |
2017
|
Identification of immune signatures predictive of clinical protection from malaria
|
Ribosomes are the highly complex macromolecular assemblies dedicated to the synthesis of all cellular proteins from mRNA templates . The main principles underlying the making of ribosomes are conserved across eukaryotic organisms and this process has been studied in most detail in the yeast Saccharomyces cerevisiae . Yeast ribosomes are composed of four ribosomal RNAs ( rRNAs ) and 79 ribosomal proteins ( r-proteins ) . Most r-proteins need to be transported from the cytoplasm to the nucleus where they get incorporated into the evolving pre-ribosomal particles . Due to the high abundance and difficult physicochemical properties of r-proteins , their correct folding and fail-safe targeting to the assembly site depends largely on general , as well as highly specialized , chaperone and transport systems . Many r-proteins contain universally conserved or eukaryote-specific internal loops and/or terminal extensions , which were shown to mediate their nuclear targeting and association with dedicated chaperones in a growing number of cases . The 60S r-protein Rpl4 is particularly interesting since it harbours a conserved long internal loop and a prominent C-terminal eukaryote-specific extension . Here we show that both the long internal loop and the C-terminal eukaryote-specific extension are strictly required for the functionality of Rpl4 . While Rpl4 contains at least five distinct nuclear localization signals ( NLS ) , the C-terminal part of the long internal loop associates with a specific binding partner , termed Acl4 . Absence of Acl4 confers a severe slow-growth phenotype and a deficiency in the production of 60S subunits . Genetic and biochemical evidence indicates that Acl4 can be considered as a dedicated chaperone of Rpl4 . Notably , Acl4 localizes to both the cytoplasm and nucleus and it has the capacity to capture nascent Rpl4 in a co-translational manner . Taken together , our findings indicate that the dedicated chaperone Acl4 accompanies Rpl4 from the cytoplasm to its pre-60S assembly site in the nucleus .
The biogenesis of ribosomes is a fundamental cellular process whose main principles are conserved from the lower eukaryote Saccharomyces cerevisiae to mammalian organisms . S . cerevisiae 80S ribosomes are composed of two unequal subunits , a small 40S ( SSU ) and a large 60S ( LSU ) ribosomal subunit ( r-subunit ) , which contain 33 ribosomal protein ( r-proteins ) and the 18S ribosomal RNA ( rRNA ) or 46 r-proteins and the 25S , 5 . 8S , and 5S rRNA , respectively [1] . The synthesis of ribosomes basically consists in the ordered assembly of the r-proteins with the rRNAs; however , the efficient and accurate assembly of r-subunits vitally depends on a multitude ( >200 ) of transiently acting biogenesis factors [2 , 3 , 4 , 5] . Ribosome assembly takes successively place in the nucleolus , nucleoplasm , and cytoplasm . Within the nucleolus , it is initiated by the transcription of the ribosomal DNA ( rDNA ) into a long , polycistronic 35S precursor rRNA ( pre-rRNA ) , which contains the 18S , 5 . 8S , and 25S rRNAs , and a pre-5S rRNA [6] . Concomitant to transcription , the 35S pre-rRNA already associates with several SSU r-proteins and early-associating biogenesis factors to give birth to the first detectable pre-ribosomal particle , referred to as 90S or SSU-processome [2 , 3 , 4 , 5] . In a predominantly co-transcriptional process , the pre-rRNA within this initial pre-ribosomal particle undergoes cleavage at processing site A2 ( for a pre-rRNA processing scheme , see S1 Fig ) [7 , 8] , thus leading to the formation of nuclear 43S pre-ribosomal particles containing the 20S pre-rRNA . These pre-40S ribosomes are rapidly exported to the cytoplasm , where they are converted in a series of concerted events , including processing of 20S pre-rRNA to 18S rRNA , into mature 40S subunits [2 , 3 , 4 , 5 , 9 , 10] . The first pre-60S ribosome , termed 66S particle , also assembles on nascent pre-rRNA and contains , upon termination of transcription , the 27SA2 pre-rRNA [2 , 8] . This 66S pre-ribosomal particle is associated with selected r-proteins , mainly binding to domains I and II of 25S rRNA , that promote the initial compaction of the emerging LSU and a characteristic set of early-acting biogenesis factors [2 , 3 , 11] . Maturation of and pre-rRNA processing within nuclear pre-60S particles proceeds in a hierarchical manner and involves the sequential recruitment of r-proteins , which shape and stabilize the pre-60S subunit with the aid of specific biogenesis factors . Affinity purifications have revealed the protein and pre-rRNA/rRNA composition of pre-60S particles and , accordingly , the existence of several distinct nuclear intermediates could be established [2 , 3 , 4 , 5 , 12] . Pre-60S subunits acquire export competence by associating with several export-promoting factors; at this stage , the particles contain most r-proteins and only relatively few biogenesis factors [2 , 3 , 4 , 13] . Upon appearance in the cytoplasm , a cascade of sequential events leads to the dissociation of the export and biogenesis factors and the incorporation of the final eight 60S r-proteins; thus enabling subunit joining and engagement of 80S ribosomes in translation [2 , 3 , 4 , 5 , 10] . An exponentially growing yeast cell must initially synthesize at least 2’000 molecules of each r-protein per minute to meet the demand for assembly-competent r-proteins [14] . Many r-proteins exhibit rather unique folds and they often contain , besides featuring in many cases a globular domain , disordered loops and extensions that stabilize the tertiary structure of rRNA [1 , 15 , 16] . Notably , eukaryotic r-subunits have acquired , when compared to their bacterial counterparts , additional rRNA elements , referred to as expansion segments ( ES ) , and 46 eukaryote-specific r-proteins [1 , 15] . Moreover , many of the evolutionarily conserved r-proteins contain eukaryote-specific extensions , which , together with the ES and the eukaryote-specific r-proteins , form most of the solvent-side surface of eukaryotic ribosomes [15 , 17] . Conversely , the evolutionarily conserved loops and extensions mostly penetrate into the interior of the rRNA cores of the r-subunits [15 , 16] . In agreement with their prevalent role in mediating interactions with the negatively charged rRNA phosphate backbones , most r-proteins are very basic with their loops and extensions being especially rich in lysine and arginine residues [16] . Since ribosome assembly mainly occurs in the nucle ( ol ) us , most r-proteins have to undertake the cumbersome journey from their cytoplasmic site of synthesis to the nucleus . However , due to their structural characteristics and highly basic nature , they need , prior to their ribosome incorporation , to be protected from engaging in illicit interactions with non-cognate RNAs or polyanions that may promote their aggregation [18] . Despite their small size , nuclear import of r-proteins largely depends on active transport across the nuclear pore complex ( NPC ) [19 , 20 , 21] . Notably , importins may not only act as transporter receptors for r-proteins , but as they were shown to prevent their aggregation , a chaperone role for importins has been put forward [18] . As already implied above , the equimolar synthesis of assembly-competent r-proteins represents a major challenge for the cell . Each non-assembled r-protein likely exhibits a distinct intrinsic stability and propensity for aggregation , as suggested by the occurrence of different folds and the unequal partitioning into globular domains and disordered extensions [15] . Moreover , proper folding of a large number of , but apparently not all , r-proteins especially depends on two functionally collaborating ribosome-associated chaperone systems [22] , consisting of the chaperone triad SSB/ribosome-associated complex ( RAC ) and the nascent polypeptide-associated complex ( NAC ) [23] . Since r-proteins associate at different spatiotemporal entry points with the evolving pre-ribosomal subunits , their non-assembled forms , which may in many cases be unstable [24 , 25] , are exposed for different durations to the hostile intracellular environment before being finally stabilized by encountering the cognate rRNA binding context at the pre-ribosome . Therefore , it is not surprising that additional mechanisms , besides the general chaperone and transport systems , evolved to ensure the stable expression of r-proteins and the subsequent delivery to their assembly site . While one such strategy , utilized by two r-proteins , comprises the initial synthesis as a precursor protein carrying an N-terminal ubiquitin moiety [26 , 27 , 28 , 29] , an emerging and prevalent theme involves the association of r-proteins with specific binding partners , also referred to as dedicated chaperones . Recent evidence revealed that such binding partners prevent r-proteins from aggregation , promote their nuclear import and/or coordinate their assembly into pre-ribosomal particles . The ankyrin-repeat protein Yar1 interacts specifically with Rps3 ( uS3 according to the recently proposed r-protein nomenclature [30] ) and acts as an anti-aggregation chaperone that may accompany Rps3 to its nuclear assembly site [31 , 32] . The nuclear Tsr2 promotes the safe transfer of importin-bound Rps26 ( eS26 ) to the 90S pre-ribosome [33] . The transport adaptor Syo1 binds simultaneously to Rpl5 ( uL18 ) and Rpl11 ( uL5 ) and mediates , upon nuclear import of the trimeric complex via the transport receptor Kap104 , their synchronized delivery to the 5S rRNA by serving as a 5S RNP assembly platform [19 , 34 , 35] . Additionally , the eight-bladed WD-repeat β-propeller protein Sqt1 and the predicted WD-repeat β-propeller protein Rrb1 are dedicated chaperones of Rpl10 ( uL16 ) and Rpl3 ( uL3 ) , respectively [36 , 37 , 38 , 39 , 40] . While Syo1 , Sqt1 , and Rrb1 recognize the N-terminal extensions of their binding partners [35 , 38] , Yar1 mainly binds to the solvent-exposed side of the first α-helix within the N-terminal globular domain of Rps3 [31] . Interestingly , and in line with a protective function , these four chaperones were recently shown to have the capacity to capture their r-protein clients at the earliest possible moment in a co-translational manner [38] . We are interested in unravelling the assembly paths , from stable cytoplasmic synthesis , along nuclear import to ribosome incorporation , of r-proteins and in understanding if and how dedicated chaperones contribute to these events . The essential 60S r-protein Rpl4 ( uL4 ) is a particularly interesting candidate for studying its assembly path , since it associates very early with pre-60S particles and displays remarkable structural features [2 , 15 , 41 , 42 , 43] . Rpl4 is mostly located on the solvent-side surface of the mature 60S r-subunit and is composed of a universally conserved globular domain and a prominent eukaryote-specific C-terminal extension ( see Fig 1 ) [15 , 41] . Notably , the globular domain contains an insertion of a long internal loop , which penetrates deep into the interior of the 60S core and whose tip region forms part of the constriction point within the polypeptide exit tunnel [15 , 41] . Additionally , a small internal loop also emanates from the surface-exposed globular domain into the 60S subunit . While the globular domain almost exclusively interacts with conserved , interconnected rRNA segments of domains I and II of the 25S rRNA , the eukaryote-specific extension spans across more than half the width of the solvent-side 60S surface and thereby engages in an intricate network of interactions , primarily with eukaryote-specific rRNA and r-protein moieties ( Fig 1A and 1B ) [11 , 15 , 17] . The first part of the eukaryote-specific extension is accommodated by Rpl18 ( eL18 ) and ES15L ( H45 ) , and the second part is sandwiched between Rpl7 ( uL30 ) , mostly by its long , eukaryote-specific N-terminal α-helix , and helices ES7Lc/ES7Lb of ES7L ( Fig 1B ) . Moreover , the eukaryote-specific r-proteins Rpl20 ( eL20 ) and Rpl21 ( eL21 ) contact the C-terminal residues of Rpl4 . In this study , we show that both the long internal loop and the C-terminal eukaryote-specific extension are essential features of Rpl4 . We further reveal that Rpl4 contains at least five distinct nuclear localization signals ( NLS ) , which map to different regions of Rpl4 , including the long internal loop and the C-terminal extension . Notably , we have identified a previously uncharacterized protein , termed Acl4 , as a specific binding partner of Rpl4 . Acl4 interacts with the C-terminal part of the long internal loop of Rpl4 and both genetic and biochemical evidence suggests that Acl4 can be considered as a dedicated chaperone of Rpl4 . Furthermore , we show that Acl4 , which localizes to the cytoplasm and the nucleus , has the capacity to capture nascent Rpl4 in a co-translational manner . Taken together , our data indicate that the dedicated chaperone Acl4 accompanies Rpl4 from the cytoplasm to its pre-60S assembly site in the nucleus .
In S . cerevisiae , the paralogous RPL4A and RPL4B genes encode the essential r-protein Rpl4 of 362 amino acids length whose two versions , Rpl4a and Rpl4b , only differ at amino acid position 356 ( threonine versus alanine ) . Notably , Rpl4 is composed of a globular domain , which contains a small ( amino acids 184–205 ) and a long internal loop ( amino acids 44–113 ) , and a eukaryote-specific C-terminal extension ( amino acids 264–362 ) ( see Introduction and Fig 1 ) . To determine the contribution of the long internal loop and the C-terminal extension to Rpl4 function , we selected Rpl4a since Δrpl4a null mutant cells , but not Δrpl4b null mutant cells , showed a moderate growth defect and reduced steady-state levels of 60S subunits , as evidenced by a shortage of free 60S subunits and the accumulation of half-mer polysomes ( S2 Fig ) . For phenotypic analysis , plasmids encoding the Rpl4a deletion variants were transformed into a Δrpl4a/Δrpl4b strain harbouring an instable URA3/ADE3 plasmids containing RPL4A ( RPL4 shuffle strain ) . Importantly , expression of wild-type Rpl4a under the control of its cognate promoter from a monocopy plasmid in the RPL4 shuffle strain conferred , upon plasmid shuffling on 5-FOA containing plates , almost wild-type growth and resulted only in a very minor 60S deficiency ( S2 Fig ) . Complete deletion of the C-terminal extension ( N264 construct; i . e . : Rpl4a deletion variant consisting of amino acids 1–264 ) did not support growth ( Fig 1C ) . To map more precisely the important regions within the C-terminal extension , we generated a series of progressive C-terminal deletion variants . This analysis revealed that the last 30 amino acids ( N332 construct ) of Rpl4 are completely dispensable; conversely , removal of the C-terminal 71 amino acids ( N291 construct ) resulted in lethality ( Fig 1C ) . The first rpl4 truncation mutant showing a slow-growth phenotype lacked the last 37 amino acids ( N325 construct ) and further deletion led to a severe slow-growth phenotype ( N312 and N301 constructs ) ( Fig 1C and S3 Fig ) . Moreover , the severity of the observed growth defects correlated with the extent of the deficiency in 60S subunits , as indicated by the more dramatic reduction in polysome content in Rpl4 . N312 and Rpl4 . N301 expressing cells ( S3 Fig ) . All the viable C-terminal deletion variants localized , as also observed for N-terminally yEGFP-tagged Rpl4a , mainly to the cytoplasm ( S3 Fig ) . While removal of the small internal loop ( deletion of amino acids 185–200 ) did not affect growth , deletion of the long internal loop ( deletion of amino acids 46–110 ) resulted in a non-functional Rpl4 protein ( Fig 1C ) . We conclude that , in agreement with its central location within 60S subunits , the presence of the long internal loop is strictly required for the synthesis of functional 60S subunits . Concerning the role of the eukaryote-specific C-terminal extension , we conclude that its interaction with Rpl20 and Rpl21 is dispensable for full functionality of Rpl4 . More importantly , the interaction network formed by the second part of the C-terminal extension ( from amino acids 308 onwards ) with Rpl7 and ES7Lb/c contributes majorly , as evidenced by the severe slow-growth and the temperature sensitivity of the Rpl4a . N301 and Rpl4a . N312 constructs ( Fig 1C and S3 Fig ) , to the efficient recruitment of Rpl4 and/or the assembly of functional 60S subunits . Further deletion of Rpl4 sites ( amino acids 292–301 ) that mediate some of the interactions with Rpl18 and ES15L conferred lethality; thus , indicating an additional important role of these contacts for incorporation of Rpl4 , pre-60S assembly and/or the functional integrity of 60S subunits ( see also Discussion ) . At this point however , it cannot be ruled out that the lethality of Rpl4a . N291 and Rpl4a . N264 variants may not be simply due to their inefficient nuclear import . To determine whether the lethal C-terminal deletion variants of Rpl4 could still enter the nucleus , we expressed them from plasmid , under the transcriptional control of the cognate promoter , as fusion proteins with a C-terminal yEGFP in a wild-type strain containing the nucleolar marker protein Nop58-yEmCherry . While Rpl4 . N291 localized almost exclusively to the nucleus , Rpl4 . N264 showed clear nuclear enrichment but also some cytoplasmic signal ( Fig 2A ) . Further C-terminal deletion revealed that constructs Rpl4 . N210 and Rpl4 . N173 exhibited a complete or partial nuclear accumulation . On the other hand , and as observed above ( S3 Fig ) , the viable C-terminal deletion constructs displayed a mainly cytoplasmic localization ( Fig 2A ) , indicating that they are assembled into mature 60S subunits . Conversely , an Rpl4 construct lacking the long internal loop displayed a striking nuclear enrichment , suggesting that the presence of the long internal loop is required for the incorporation of Rpl4 into and/or the nuclear maturation of pre-60S subunits ( see below and Discussion ) . Furthermore , we conclude that the Rpl4 variants lacking completely or more than two-thirds of the C-terminal extension ( N264 and N291 constructs ) enter the nucleus but are presumably not efficiently assembled into pre-60S subunits . To obtain a complete overview of the different Rpl4 regions that may confer nuclear localization , we next wished to precisely map the individual NLSs . To reduce passive diffusion across the NPC , we fused a C-terminal triple yEGFP ( 3xyEGFP ) preceded by a short ( GA ) 5 linker , consisting of five glycine-alanine repeats , to the different Rpl4 fragments . Since an Rpl4 fragment consisting of the complete C-terminal extension ( 263C construct; amino acids 263–362 ) localized to the nucleus ( Fig 2B ) , we first determined the NLS region ( s ) within the C-terminal extension . This analysis revealed that the C-terminal extension contains two distinct , but partially overlapping regions , consisting of amino acids 263–325 and 312–341 , which conferred nuclear targeting , albeit slightly less efficiently than amino acids 263–362 ( Fig 2B ) . These two regions can be considered as minimal NLSs since their further N- or C-terminal shortening resulted in a mostly cytoplasmic signal ( S4 Fig ) . In agreement with the exclusive nuclear localization of the N210 construct ( Fig 2A ) , we could map a strong NLS to amino acids 183–210 , which basically encompasses the small internal loop region of Rpl4 ( Fig 2B ) . Despite the finding that the N173 construct only displayed a partial nuclear enrichment , we could identify therein two quite efficient , but again partially overlapping , NLS regions consisting of amino acids 43–114 and 101–173 ( Fig 2B ) . Since their further N- and/or C-terminal shortening strongly decreased the intensity of the nuclear signal ( S4 Fig ) , these two regions likely correspond to minimal NLSs . Notably , the NLS region defined by amino acids 43–114 corresponds to the long internal loop of Rpl4 . Taken together , our comprehensive analysis revealed that Rpl4 contains at least five distinct NLSs , with four of these being part of two larger , overlapping NLS regions ( Fig 2A and S4 Fig ) . However , due to the enormous combinatorial complexity—there are ten importin-β proteins in yeast , which , moreover , often display significant overlap in substrate specificity [44]–we have not attempted to assign the importin ( s ) responsible for the nuclear import of the five individual NLSs within Rpl4 . To explore a possible role of the C-terminal extension for assembly of Rpl4 into pre-60S particles , we first expressed the partially functional ( N325 and N301 constructs ) and lethal ( N291 and N264 constructs ) C-terminal deletion variants from plasmid , under the transcriptional control of the cognate promoter , in a wild-type strain and examined their effects on growth . While additional expression of Rpl4a did not affect the growth of wild-type cells , all of the above C-terminal deletion variants conferred a similar slow-growth phenotype and decrease in 60S and polysome content ( Fig 3A and 3B ) . Moreover , related growth defects were observed when these C-terminal deletion variants were overexpressed from a galactose-inducible promoter ( S5 Fig ) . In agreement with their cytoplasmic localization , the partially functional Rpl4 . N325 and Rpl4 . N301 proteins , expressed as fusion proteins with an N-terminal TAP tag in wild-type cells , got incorporated into 60S , 80S , and translating ribosomes ( Fig 3B ) . Conversely , the almost exclusively nuclear Rpl4 . N291 only showed a very minor 60S association and was mostly detectable in the soluble fractions ( Fig 3B ) , while Rpl4 . N264 , as already suggested by its dual cytoplasmic and nuclear localization , was both present in the soluble and ribosome-associated fractions ( Fig 3B ) . We conclude that the C-terminal extension contributes to the efficient assembly of Rpl4 into nuclear pre-60S particles . Interestingly , expression of Rpl4 . N291 , which is not stably incorporated into 60S subunits , confers a severe slow-growth phenotype to wild-type cells . This observation raised the possibility that free Rpl4 . N291 may titrate a protein that acts positively on wild-type Rpl4 . Finally , the Rpl4 protein lacking the long internal loop mainly migrated in the fractions surrounding the 60S peak and exhibited a clearly reduced 80S and polysome distribution when compared to Rpl3 ( Fig 3B ) . Considering that the Rpl4 ( Δ46–110 ) protein localized predominantly to the nucleus ( Fig 2A ) , we conclude that this Rpl4 variant gets efficiently incorporated into pre-60S subunits and , therefore , that the long internal loop is not a critical determinant for the assembly of Rpl4 . Moreover , expression of Rpl4 ( Δ46–110 ) in wild-type cells not only conferred a slow-growth phenotype but also some decrease in 60S subunits and a drastic reduction in polysome content ( Fig 3A and 3B ) ; thus , pointing to a possible role of the long internal loop in maturation events that are necessary for the productive assembly of export-competent pre-60S subunits ( see Discussion ) . In order to corroborate the sucrose gradient fractionation data and , if possible , to identify the anticipated binding partner of free Rpl4 , we performed tandem-affinity purifications ( TAP ) of N-terminally TAP-tagged Rpl4 . N264 and Rpl4 . N291 . In agreement with its dual gradient localization , purification of NTAP-Rpl4 . N264 not only yielded the bait protein but also sub-stoichiometric amounts of r-proteins ( Fig 4A ) . Most strikingly , a prominent band , corresponding to a protein migrating at around 43 kDa , was observed in the final EGTA eluates of both purifications ( Fig 4A ) . Mass spectrometric analysis revealed that this band contained the previously uncharacterized protein Ydr161w , which was recently assigned the name Acl4 ( Assembly Chaperone of RpL4 ) [45] . Moreover , both purifications contained low levels of early 60S biogenesis factors ( Fig 4A ) , indicating that both Rpl4 . N264 and Rpl4 . N291 have the capacity to be incorporated into nucleolar pre-60S particles . In further validation of an association with early pre-60S particles , N-terminally GFP-tagged Rpl4a . N264 precipitated the 27SA2 and 27SB pre-rRNAs ( S6 Fig ) . To address whether Acl4 was also associated with full-length Rpl4 under normal conditions , we expressed N-terminally TAP-tagged full-length Rpl4a from plasmid , under the transcriptional control of the cognate promoter , in Δrpl4a/Δrpl4b cells . In agreement with the good functionality of this construct ( S7 Fig ) , the purification revealed that NTAP-Rpl4a was efficiently incorporated into mature 60S subunits ( Fig 4B ) . Notably , significant amounts of Acl4 could be co-purified with the NTAP-Rpl4a bait . Next , we purified in a reciprocal experiment C-terminally TAP-tagged Acl4 , which was expressed as a fully functional protein from its genomic locus ( S7 Fig ) , thereby revealing that Acl4 specifically co-purified free Rpl4 ( Fig 4C ) . Moreover , the only other prominent band , migrating slightly above 70 kDa , contained the Hsp70 chaperones Ssa1 , Ssa2 , Ssa3 , and Ssa4 . However , we have not further explored the functional significance of their co-enrichment with the Acl4-TAP bait in this study , since Hsp70 chaperones , belonging to the Ssa and Ssb subfamilies , are rather commonly found in TAP purifications . Altogether , these in vivo purifications provide strong evidence that Acl4 is a specific binding partner of non-ribosome bound Rpl4 . Acl4 is an acidic protein ( pI 4 . 15 ) of 387 amino acids with a calculated molecular mass of 42 . 94 kDa . The eukaryote-specific Acl4 is predicted to contain 16 α-helices that , based on our bioinformatics analysis , may form up to eight tetratrico peptide repeats ( TPR ) or TPR-like repeats . Notably , TPR domains build the scaffolds that mediate protein-protein interactions and the assembly of multiprotein complexes in a versatile manner [46 , 47] . Acl4 is not only conserved in fungi , such as filamentous ascomycetes ( e . g . : Chaetomium thermophilum , Accession: XP_006691536 ) and fission yeasts ( e . g . : Schizosaccharomyces pombe , Accession: NP_596496 ) , as orthologues can also be found in protists ( e . g . : Trypanosoma brucei , Accession: XP_011776831 ) , arachnids ( e . g . : Stegodyphus mimosarum , Accession: KFM57536 and Metaseiulus occidentalis , Accession: XP_003742882 ) , ray-finned and coelcanth fish ( e . g . : Danio rerio , Accession: XP_009290679 and Latimeria chalumnae , Accession: XP_006002768 ) , and amphibians ( e . g . : Xenopus tropicalis , Accession: XP_002934631 ) . However , there are no orthologues in the evolutionary more advanced classes of reptiles , birds , and mammals . Moreover , Acl4 is not present in plants ( e . g . : Arabidopsis thaliana ) , insects ( e . g . : Drosophila melanogaster ) , and nematodes ( e . g . : Caenorhabditis elegans ) , indicating that Acl4 can also be specifically lost in earlier evolutionary branches . To address the functional significance of the identification of Acl4 as an Rpl4 binding protein , we first determined whether Acl4 contributed to the biogenesis of 60S subunits . Haploid cells with a chromosomally disrupted ACL4 gene ( Δacl4 ) were viable , but displayed a severe slow-growth phenotype at all tested temperatures ( Fig 4D ) . In agreement with an involvement in 60S biogenesis , polysome profile analysis revealed that Δacl4 cells contained reduced levels of 60S subunits , as evidenced by a shortage of free 60S subunits and the accumulation of half-mer polysomes , resulting in a substantial decrease in overall polysome content ( Fig 4E ) . Compared to cells that were genetically depleted for Rpl4 , which showed as previously reported a striking decrease in 27SB and 7S pre-rRNAs ( S8 Fig ) [43 , 48] , the reduction of these pre-rRNA species was clearly observable but less pronounced in Δacl4 cells ( S8 Fig ) . In line with the in vivo purification , sucrose gradient fractionation revealed that Acl4-TAP was exclusively present in the soluble fractions ( Fig 4F ) ; thus , providing further evidence that Acl4 is not stably associated with pre-60S or mature 60S subunits . Finally , Acl4-GFP , expressed from its genomic locus as a functional protein at 30°C ( S7 Fig ) , localized to both the nucleus and cytoplasm ( Fig 4G ) . Since Acl4 lacks a predicted NLS , it is highly likely that Acl4 may already bind to Rpl4 in the cytoplasm and may be imported in complex with Rpl4 , which contains several experimentally established NLS regions ( Fig 2B ) . To precisely map the binding site of Acl4 on Rpl4 , we first conducted yeast two-hybrid ( Y2H ) interaction assays ( Fig 5A ) . As already indicated by the co-purification of Acl4 with NTAP-Rpl4 . N264 ( Fig 4A ) , the C-terminal extension was neither required for nor sufficient to mediate the interaction with Acl4 ( Fig 5A ) . Further C-terminal deletion analysis revealed that the first 114 amino acids ( N114 construct ) were sufficient for a robust interaction , while the Rpl4 . N104 construct did not show any interaction with Acl4 ( Fig 5A ) . We therefore tested next whether Acl4 recognized the long internal loop and , indeed , a strong Y2H interaction was observed between Acl4 and amino acids 43–114 of Rpl4 . By shortening the long internal loop from the N-terminal side , amino acids 88–114 were identified as the minimal , albeit less efficient , interaction fragment . However , amino acids 88–264 resulted in an interaction that was almost as strong as the one between Acl4 and full-length Rpl4 , thus indicating that the region around amino acid 88 likely corresponds to the N-terminal border of the minimal interaction fragment . In further support of this notion , amino acids 96–114 or 96–264 of Rpl4 were insufficient to promote an interaction with Acl4 . Finally , deletion of the long internal loop from full-length Rpl4 ( deletion of amino acids 46–110 ) completely abolished the Y2H interaction . To corroborate the Y2H data , we turned to in vitro binding assays ( Fig 5B ) . To this end , we co-expressed C-terminally ( His ) 6-tagged Rpl4 or fragments thereof with full-length Acl4-Flag in Escherichia coli and subsequently performed Ni-affinity purification of the different Rpl4 baits . These binding assays confirmed that the long internal loop contained the Acl4 binding site and defined amino acids 72–114 of Rpl4 as the minimal region conferring a robust interaction ( Fig 5B ) . In contrast to the Y2H data , however , amino acids 88–114 of Rpl4 were insufficient to mediate the interaction in vitro . Finally , the Rpl4 bait lacking the long internal loop did not yield co-purification of Acl4 ( Fig 5B ) . To corroborate the in vitro binding data obtained with the S . cerevisiae proteins and to obtain , if possible , structural insight into the Acl4-Rpl4 interaction , we turned to the orthologous proteins from the thermophilic , filamentous ascomycete C . thermophilum ( ct ) . Proteins from this organism often display excellent biochemical properties and are well suited for structural studies [49] . In agreement with being a functional orthologue , ctAcl4 complemented the slow-growth phenotype of Δacl4 mutant cells to the wild-type extent ( S9 Fig ) . Ni-affinity purification of ctRpl4- ( His ) 6 yielded stoichiometric amounts of co-expressed ctAcl4-Flag ( S9 Fig ) . Moreover , these in vitro binding assays also confirmed that the long internal loop is sufficient to mediate the interaction . Most notably and in contrast to the in vitro binding data of the S . cerevisiae proteins , amino acids 89–115 of ctRpl4 resulted in an efficient co-purification of ctAcl4-Flag ( S9 Fig ) , suggesting that these residues indeed constitute the minimal , evolutionary conserved binding site for Acl4 . Unfortunately however , we could so far not gain structural insights into the interaction mode , since our attempts to obtain crystals of full-length ctAcl4 or co-crystals of the ctRpl4 ( 89–115 ) /ctAcl4 complex were not yet successful . To confirm that the C-terminal part of the long internal loop ( amino acids 88–114 ) is required for the interaction in S . cerevisiae and to delineate important residues , we next generated four different Rpl4 variants harbouring non-overlapping , consecutive alanine substitutions ( block-I: F90A/N92A/M93A/C94A/R95A , block-II: R98A/M99A/F100A , block-III: P102A/T103A/K104A/T105A , and block-IV: W106A/R107A/K108A/W109A ) ( see Fig 6A ) . Contrary to the complete deletion of the long internal loop ( Fig 1C ) , all four alanine-block substitution mutants were viable , albeit displaying different degrees of growth deficiencies ( S10 Fig ) . Moreover , these mutant Rpl4 proteins were similarly expressed as wild-type Rpl4 and their expression did not confer a slow-growth phenotype to wild-type cells ( S10 Fig ) . Subsequent Y2H analyses and in vitro binding assays revealed that none of these four Rpl4 variants retained the capacity to interact with Acl4 ( Fig 6B and 6C ) . Taken together , we conclude that Acl4 recognizes Rpl4 by directly interacting with the C-terminal part of the long internal loop ( amino acids 88–114 ) and that , notably , residues dispersed along the entire , linear interaction segment are critical binding determinants . Having established Acl4 as a physical interaction partner of Rpl4 , we next wished to assess the functional relevance of their association . Genetic analyses revealed that the slow-growth phenotype of Δacl4 cells could be efficiently suppressed by overexpression of Rpl4a from a monocopy plasmid ( Fig 7A ) . Moreover , cells simultaneously lacking Acl4 and Rpl4a exhibited a pronounced synthetic enhancement phenotype , as evidenced by their severely reduced growth rate compared to the one of Δacl4 and Δrpl4a single mutant cells ( Fig 7B ) . These findings point to an important function of Acl4 in ensuring that cells are provided with sufficient amounts of assembly-competent Rpl4 . Along these lines , we observed that overexpression of Acl4 suppressed the growth defects associated with the expression of Rpl4 . N264 and Rpl4 . N291 ( Fig 7C ) , which efficiently associate with Acl4 ( Fig 4A ) and therefore likely compete with endogenous Rpl4 for Acl4 binding . Finally , newly synthesized Rpl4-2xHA , expressed for 20 min from a copper-inducible promoter , was only found to be soluble in the presence , but not in the absence , of Acl4 ( Fig 7D ) . Taken together , the genetic and biochemical evidence indicates that Acl4 can be considered as a specific chaperone of Rpl4 . Acl4 , which lacks a predicted NLS , localizes both to the cytoplasm and nucleus ( Fig 4G ) . Moreover , we observed that overexpression of Acl4 from a galactose-inducible promoter led to the nuclear accumulation of Rpl4 ( Fig 7E ) . Therefore , Acl4 may already bind to Rpl4 in the cytoplasm and travel in complex with Rpl4 to the nucleus . To obtain evidence for a cytoplasmic interaction , which would most efficiently already be established during translation of Rpl4 , we assessed whether Acl4 was recruited to nascent Rpl4 . To this end , we employed , as recently described [38] , a method coupling chaperone purification to the detection of associated mRNAs by real-time quantitative reverse transcription PCR ( real-time qRT-PCR ) . Briefly , we purified Acl4-TAP and , as a control , Syo1-FTpA by IgG-Sepharose pull-down from extracts of cells that were , prior to harvesting , treated with cycloheximide in order to preserve the translating ribosomes on the mRNAs ( see Methods ) . The purified chaperones were then released from the IgG-Sepharose beads by TEV cleavage and the associated RNA was extracted and transcribed into cDNA , which was used as the template for the assessment of the levels of the different r-protein mRNAs ( RPL3 , RPL4 , RPL5 and RPL11 ) by real-time qRT-PCR . As recently reported [38] , purification of Syo1-FTpA specifically enriched the mRNA encoding Rpl5 , while the amounts of the RPL4 mRNA were similar to the one of the negative control mRNA RPL3 ( Fig 8 ) . Interestingly , the mRNA encoding Rpl11 , which forms together with Syo1 and Rpl5 a co-imported complex [34 , 35] , was not enriched , indicating that only Rpl5 , but not Rpl11 , is captured by Syo1 in a co-translational manner . On the other hand , purification of Acl4-TAP yielded a specific and robust enrichment of the RPL4 mRNA ( Fig 8 ) . Notably , the fold enrichment ( around 150 fold ) of the RPL5 or RPL4 mRNAs , compared to the non-specific r-protein mRNAs , was similar in the case of the Syo1-FTpA and Acl4-TAP purification , respectively . We conclude that Acl4 has the capacity to recognize Rpl4 in a co-translational manner .
Recent evidence has highlighted that r-proteins rely in several cases on specific binding partners , also referred to as dedicated chaperones , which favour the soluble expression of r-proteins , promote their nuclear import and/or coordinate their assembly into pre-ribosomal subunits ( see Introduction ) . In the present study , we have identified such a specific binding partner , termed Acl4 , which exclusively interacts with the LSU r-protein Rpl4 . The genetic and biochemical data reported here indicate that Acl4 can be considered as a dedicated chaperone of Rpl4 . Notably , Acl4 has the capacity to recognize , by directly interacting with the C-terminal part of the long internal loop , nascent Rpl4 in a co-translational manner . Further , the identification of several NLS regions within Rpl4 suggests that Acl4 , which lacks a predicted NLS , gets imported into the nucleus in complex with Rpl4 and accompanies Rpl4 to its nucleolar pre-60S incorporation site . Finally , we show that both the eukaryote-specific C-terminal extension and the long internal loop are essential features of Rpl4 , which are required for the assembly of Rpl4 into and the nuclear maturation of pre-60S subunits . While we compiled our manuscript , the Hurt laboratory , in collaboration with the Hoelz group , independently reported the identification and characterization of Acl4 as a binding partner and assembly chaperone of Rpl4 [45] . Notably , their study also described the partial crystal structure of C . thermophilum Acl4 , revealing that , in agreement with our bioinformatics prediction , the central core of Acl4 is made up of 6 . 5 TPR repeats [45] . Moreover , the Woolford laboratory investigated in a recent study the contribution of the long internal loop and the eukaryote-specific C-terminal extension of Rpl4 to the assembly of 60S subunits [48] . By assimilating the data from these three different studies , we outline an integrated model , whose main steps are discussed below , describing how Acl4 may ensure the synthesis of assembly-competent Rpl4 and how the incorporation of Rpl4 into early pre-60S particles may proceed ( Fig 9 ) . Several lines of evidence indicate that Acl4 can be considered as a dedicated ‘holding’ chaperone of Rpl4 . First , Acl4 specifically binds to free Rpl4 and is therefore only , if at all , very transiently associated with pre-60S particles during the initial docking phase of Rpl4 ( Fig 4C and 4F ) . Second , genetic experiments reveal an important function of Acl4 in ensuring that cells are provided with sufficient amounts of assembly-competent Rpl4 ( Fig 7A and 7B ) [45] . Suppression of the growth defects , entailed by the absence of a dedicated chaperone , by overexpression of the respective r-protein partner has been previously observed for other r-protein/chaperone pairs [32 , 35 , 38] . Moreover , we observed that cells simultaneously lacking Acl4 and Rpl4a are substantially sicker than the individual single mutants ( Fig 7B ) . Third , by binding to the long internal loop of Rpl4 , which penetrates deeply into the rRNA core of the 60S subunit ( see Fig 1 ) , Acl4 may shield this highly basic region from engaging in illicit interactions with polyanions and , hence , aggregation prior to its final insertion into the cognate rRNA environment within nascent pre-60S subunits . Fourth , and in line with a protective function , Acl4 is required for the soluble expression of newly synthesized Rpl4 in yeast cells ( Fig 7D ) . Elegant experiments from the Görlich laboratory have demonstrated that r-proteins are prone to aggregation in the presence of polyanions , such as tRNA [18] . Notably , precipitation of selected r-proteins could be reversed by incubation with specific subsets of importins , thus establishing , besides their classical role as nuclear transport receptors , an anti-aggregation function for importins . It has therefore been postulated that , in order to be most efficiently protected , newly synthesized r-proteins should be immediately , possibly even co-translationally , shielded [18] . While it has not yet been revealed whether importins might already be recruited to nascent r-proteins , we have recently shown that four dedicated chaperones capture their specific r-protein partner in a co-translational manner [38] . In this study , we provide evidence that , likewise , Acl4 has the capacity to recognize Rpl4 as it is synthesized by the ribosome ( Fig 8 ) . Contrary to the other four cases , where the N-terminal regions of the r-proteins constitute the chaperone binding sites , Acl4 binds to an internal region , the C-terminal part of the long internal loop , of Rpl4 . Further support for an early , cytoplasmic interaction is provided by the finding that Rpl4 is already bound by Acl4 within 5 min of its induced expression in an experiment combining the pulse-chase epitope labelling of Rpl4 with its affinity purification [45 , 50] . The necessity to protect Rpl4 as early as possible is also highlighted by the observation that it is susceptible to aggregation , like many other r-proteins , in the absence of the ribosome-associated chaperone systems ( SSB/RAC and NAC ) [22] . Collectively , this raises interesting questions about the coordination of the different co-translational processes , e . g . : how is a nascent r-protein , such as Rpl4 , transferred from the general co-translational chaperones to its dedicated chaperone in order to ensure its productive synthesis as a soluble protein ? In agreement with cytoplasmic formation and nuclear co-import of the Acl4-Rpl4 complex , in vitro reconstitution experiments revealed that the importin Kap104 is capable of forming a stoichiometric trimeric complex with Acl4 and Rpl4 [45] . Notably , the association of the Acl4-Rpl4 heterodimer with Kap104 is dependent on the presence of Rpl4’s C-terminal extension , which harbours a complex NLS region consisting of two partially overlapping NLSs ( Fig 2B ) [45] . However , Rpl4 lacking the C-terminal extension is still targeted to the nucleus ( Fig 2A ) , indicating that the other three identified NLSs are sufficient to mediate nuclear import . Since binding of Acl4 covers a region of Rpl4 , encompassing amino acids 101–114 , that is a critical nuclear-targeting determinant for two of these NLSs ( Fig 2B and S4 Fig ) , it can be inferred that this NLS region should only be responsible for the import of free Rpl4 and not the Acl4-Rpl4 complex . Taken together and considering the quasi-essential nature of Acl4 , we propose that the co-import of the Acl4-Rpl4 complex , mediated by binding of Kap104 to the C-terminal extension of Rpl4 , represents the major import and pre-60S assembly pathway ( Fig 9 ) . However , it is very likely that other , yet to be determined importins may also recognize the complex NLS region within Rpl4’s C-terminal extension since its nuclear localization is not abolished in kap104-16 mutant cells ( S11 Fig ) . Given that yeast cells still grow , albeit at a very slow rate , in the absence of Acl4 ( Fig 4D ) [45] , we suggest the existence of alternative , Acl4-independent import and pre-60S assembly routes for Rpl4 ( Fig 9 ) . In these minor import pathways , nuclear transport of Rpl4 is likely promoted by importin binding to one of the two internal or the C-terminally located NLS regions . Irrespective of the import route , Ran-GTP binding to the importin will release free or Acl4-bound Rpl4 , which can then be incorporated into early pre-60S particles . Additionally , it is also possible that Rpl4 , imported via an Acl4-independent route , will encounter and be transferred to an Acl4 molecule in the nucleoplasm . In this scenario , Acl4 would act as an escortin , as recently proposed to be the role of the Rps26 chaperone Tsr2 [33] , connecting the nuclear import of Rpl4 to its fail-safe deposition on pre-60S subunits . Finally , it has to be assumed that most Acl4 molecules , in order to sustain the enormous demand for newly synthesized and assembly-competent Rpl4 , should travel back , after pre-60S delivery of Rpl4 , from the nucleus to the cytoplasm . However , it remains to be determined whether translocation of Acl4 across the NPC occurs by facilitated diffusion or relies on an active export mechanism . Contrary to the result obtained by the Woolford laboratory [48] , our study , as well as the one from the Hurt laboratory [45] , clearly revealed that the eukaryote-specific extension of Rpl4 harbours an essential function ( Fig 1C ) . Due to the importance of the eukaryote-specific C-terminal extension in coupling Acl4 release with the incorporation of Rpl4 into nascent pre-60S subunits , it was proposed that this region delivers the Acl4-Rpl4 complex to the pre-ribosomal particle by contacting co-evolved , eukaryote-specific sites [45] . Given that the efficient assembly of Rpl4 relies on critical hydrophobic contacts between residues of the C-terminal extension ( e . g . : Ile289 , Ile290 , and Ile295 ) with Rpl18 , a hierarchical assembly model has been put forward [45] . Accordingly , assembly of Rpl18 would precede and mediate , by contributing to the recruitment of the eukaryote-specific C-terminal extension , the initial docking of Rpl4 , which would then be followed by the insertion of the long internal loop and the concomitant release of Acl4 [45] . However , several lines of evidence indicate that a complete understanding of the order of the assembly events will require further clarification . Contrary to the above-described contribution of certain hydrophobic residues of Rpl4 to its pre-60S incorporation and release from Acl4 , Rpl4 gets , while only showing a moderate increase in Acl4 association , efficiently assembled into mature 60S subunits in cells expressing an Rpl18 variant ( L32E , V129D ) with mutations in the Rpl4 interaction surface [45] . Moreover , our C-terminal deletion analysis revealed that interactions of Rpl4 , besides the contacts with Rpl18 , with Rpl7 and ES7L , involving the middle part of the C-terminal extension ( amino acids 302–332 ) , are required for optimal cell growth and production of 60S subunits ( Fig 1C and S3 Fig ) . Nevertheless , these mutant Rpl4 proteins ( N325 and N301 constructs ) get , in the presence of wild-type Rpl4 , incorporated into mature 60S subunits and translating ribosomes ( Fig 3B ) . More strikingly even , we observed in the same experimental setting that Rpl4 lacking the complete C-terminal extension ( N264 construct ) can be assembled , albeit less efficiently , into mature 60S subunits ( Fig 3B ) ; thus , indicating the possibility of an independent pre-60S assembly of the universally conserved part of Rpl4 . However , the assembly of the globular domain and C-terminal extension seems to be tightly interconnected , as suggested by our observation that their separate expression confers growth to yeast cells and leads to the stable incorporation of both protein fragments into mature 60S subunits ( S12 Fig ) . Further , given that the globular domain and the N-terminal part of the long internal loop of Rpl4 engage in a significant number of interactions with LSU rRNA domain I , while Rpl18 almost exclusively forms polar contacts with rRNA domain II , the initial association of Rpl4 with nascent pre-rRNA may occur prior to Rpl18 recruitment . Taken together , an alternative model for Rpl4 assembly , involving a series of interconnected steps , may be envisaged: ( i ) initial docking of Rpl4 is established by contacts of the globular domain and parts of the long internal loop with rRNA domain I; ( ii ) the C-terminal extension of Rpl4 facilitates Rpl18 and Rpl7 recruitment , thereby enabling formation of the eukaryote-specific interaction network and , thus , fortifying Rpl4 association and promoting correct pre-rRNA folding within this pre-60S region; ( iii ) either concomitantly or subsequently , the long internal loop of Rpl4 gets completely inserted into the rRNA core of pre-60S subunits , thereby leading to the dissociation of Acl4 from its Rpl4 binding site . These highly complex assembly events , coupling stable Rpl4 incorporation with Acl4 release , are expected to occur very fast since Acl4 is not detectably associated with pre-60S particles ( Fig 4C and 4F ) . Moreover , the cellular Acl4 levels seem to influence the efficiency of Rpl4 association with pre-60S subunits , by affecting the equilibrium between Acl4-bound and pre-60S-associated Rpl4 , since overexpression of Acl4 confers a strong slow-growth phenotype to wild-type cells ( S13 Fig ) ; this observation also suggests that Acl4 does not actively promote Rpl4 assembly . The combined data from the three studies also provide compelling evidence for a role of the essential long internal loop of Rpl4 during late pre-60S maturation events that are necessary for the productive assembly of export-competent pre-60S subunits . In contrast to depletion of Rpl4 , which entails early pre-60S assembly defects ( S8 Fig ) [43 , 48] , it was shown that Rpl4 depleted cells expressing an Rpl4 protein lacking its long internal loop display a strong accumulation of the 7S pre-rRNA [48] . In line with such a requirement during late pre-rRNA processing steps leading to the formation of mature 5 . 8S rRNA , purification , upon pulse-chase epitope labelling , of Rpl4 lacking the tip region ( amino acids 63–87 ) of the long internal loop yielded late pre-60S ribosomes , as inferred by the co-enrichment of the export adaptor Nmd3 and the GTPases Nog1 and Lsg1 [45] . Consistently , Rpl4 lacking the long internal loop , when expressed in wild-type cells , confers a slow-growth phenotype and , as revealed by sucrose gradient fractionation , gets efficiently incorporated into pre-60S subunits ( Fig 3A and 3B ) . While the other two studies did not investigate in which compartment these aberrant pre-60S subunits are stalled , our report notably reveals , as suggested by the predominant nuclear localization of Rpl4 lacking the long internal loop ( Fig 2A and 2B ) , that maturation of pre-60S particles is blocked at a nuclear step . Future experiments will be required to better define the specific step at which assembly of these pre-60S subunits is halted and how the long internal loop of Rpl4 promotes progression of late pre-60S maturation . An interesting , yet puzzling observation is that the eukaryote-specific Acl4 , despite its almost essential function in yeast , is not conserved in all evolutionary more advanced classes , including mammals , and also conspicuously absent from certain early evolutionary branches; thus , indicating that other proteins may relatively easily replace Acl4 . Comparison of the Acl4 binding site within the long internal loop of Rpl4 between archaea and yeast reveals that this region , which has only limited sequence similarity , notably displays eye-catching differences with respect to the electrostatic surface properties ( S14 Fig ) . This observation suggests that Acl4 might have arisen due to the necessity to shield the more positively charged , eukaryotic surface during transport of Rpl4 to its nuclear pre-60S assembly site . Given that certain importins ( the importin-β/importin-7 heterodimer and importin-9 ) were shown to counteract the aggregation of mammalian rpL4 in the presence of polyanions [18] , it is reasonable to speculate that these importins may fulfil a dual role as dedicated chaperone and transport receptor of rpL4 in mammalian cells . In conclusion , our study has identified the previously uncharacterized Acl4 as a dedicated chaperone of the 60S subunit r-protein Rpl4 and has revealed that recognition may already occur during the cytoplasmic synthesis of Rpl4 . Our findings underscore the necessity to protect r-proteins on their path to their ribosomal incorporation site and further validate co-translational capturing of r-proteins by dedicated chaperones as an advantageous and prevalently used concept to efficiently fulfil this task . Clearly , future work will be required to decipher the molecular details of the Acl4-Rpl4 interaction and to unveil the precise mechanisms that promote the stable assembly of Rpl4 into pre-60S subunits .
The S . cerevisiae strains used in this study are listed in Supporting S1 Table; all strains , unless otherwise specified , are derivatives of W303 [51] . For yeast two-hybrid analyses the reporter strain PJ69-4A was used [52] . Deletion disruption and C-terminal tagging at the genomic locus were performed as described [53 , 54] . Preparation of media , yeast transformation , and genetic manipulations were done according to established procedures . For the experiments involving induction of expression by addition of copper sulfate , media were prepared with copper-free yeast nitrogen base ( FORMEDIUM ) . All recombinant DNA techniques were according to established procedures using Escherichia coli DH5α for cloning and plasmid propagation . Codon-optimized ( for E . coli expression ) C . thermophilum ctACL4 and ctRPL4 genes were generated by custom DNA synthesis ( Eurofins ) . All cloned DNA fragments generated by PCR amplification were verified by sequencing . More information on the plasmids , which are listed in Supporting S2 Table , is available upon request . For Y2H-interaction assays , plasmids expressing bait proteins , fused to the Gal4 DNA-binding domain ( G4BD ) , and prey proteins , fused to the Gal4 activation domain ( G4AD ) , were co-transformed into reporter strain PJ69-4A . Y2H interactions were documented by spotting representative transformants in 10-fold serial dilution steps onto SC-Trp-Leu , SC-Trp-Leu-His ( HIS3 reporter ) , and SC-Trp-Leu-Ade ( ADE2 reporter ) plates , which were incubated for 3 d at 30°C . Growth on SC-Trp-Leu-His plates is indicative of a weak/moderate interaction , whereas only relatively strong interactions permit growth on SC-Trp-Leu-Ade plates . Live yeast cells were imaged by fluorescence microscopy using an Olympus BX54 microscope . Nop58-yEmCherry , expressed from the genomic locus under the control of the cognate promoter , was used as a nucleolar marker . The Image J software was used to process the images . Cells expressing Acl4-TAP and NTAP-Rpl4 were grown at 23°C in 4 l YPD medium to an optical density ( OD600 ) of 2 . Wild-type cells expressing NTAP-Rpl4a . N264 and NTAP-Rpl4a . N291 from plasmid were grown at 30°C in 4 l SC-Leu medium to an OD600 of 1 . 5 . Cell extracts were obtained by glass bead lysis with a Pulverisette ( Fritsch ) . Tandem-affinity purifications were performed in a buffer containing 50 mM Tris-HCl pH 7 . 5 , 100 mM NaCl , 1 . 5 mM MgCl2 , 5% glycerol , and 0 . 1% NP-40 as described [55] . The EGTA eluates were precipitated by the addition of TCA to a final concentration of 10% and , after an acetone wash , dissolved in 80 μl of 3x SDS sample buffer . Protein samples were separated on NuPAGE 4–12% Bis-Tris 12-well gels ( Novex ) , run in 1x MES SDS running buffer , and subsequently stained with Brilliant Blue G Colloidal Coomassie ( Sigma ) . The identity of the proteins contained in Coomassie-stained bands was determined by mass spectrometric analysis of peptides obtained by digestion with trypsin . For in vitro binding assays between Rpl4- ( His ) 6 and Acl4-Flag or between ctRpl4- ( His ) 6 and ctAcl4 , proteins were co-expressed from pETDuet-1 ( Novagen ) in Rosetta ( DE3 ) ( Novagen ) or BL21 ( DE3 ) ( Novagen ) E . coli cells , respectively . Cells were grown in 200 ml of lysogeny broth ( LB ) medium at 37°C and protein expression was induced at an OD600 of around 0 . 6 to 0 . 8 by the addition of IPTG to a final concentration of 0 . 5 mM . After 3 h of growth at 30°C , cells were harvested and stored at -80°C . Cells were resuspended in 25 ml lysis buffer ( 50 mM Tris-HCl pH 7 . 5 , 200 mM NaCl , 1 . 5 mM MgCl2 , 5% glycerol ) and lysed with a M-110L Microfluidizer ( Microfluidics ) . The lysate ( 30 ml volume ) was adjusted by the addition of 300 μl 10% NP-40 to 0 . 1% NP-40 ( note that from here onwards all buffers contained 0 . 1% NP-40 ) . An aliquot of 100 μl of total extract ( sample T ) was taken and mixed with 100 μl of 6x loading buffer . The total extract was then centrifuged at 4°C for 20 min at 14’000 rpm . The soluble extract was transferred to a 50 ml Falcon tube and , as above , an aliquot of 100 μl of soluble extract ( sample S ) was taken and mixed with 100 μl of 6x loading buffer . The insoluble pellet fraction ( sample P ) was resuspended in 3 ml of lysis buffer and 10 μl thereof were mixed with 90 μl of lysis buffer and 100 μl of 6x loading buffer . The soluble extract ( 30 ml ) was adjusted to 15 mM imidazole by adding 180 μl 2 . 5 M imidazole pH 8 . Upon addition of 250 μl of Ni-NTA Agarose slurry ( Qiagen ) , samples were incubated for 2 h on a turning wheel at 4°C . Then , the Ni-NTA Agarose beads were pelleted by centrifugation at 4°C for 2 min at 1’800 rpm , resuspended in 2 ml of lysis buffer , and transferred to a 2 ml Eppendorf tube . The Ni-NTA Agarose beads were first washed five times with 1 ml of lysis buffer containing 15 mM imidazole and then two times for 5 min , by rotation on a turning wheel at 4°C , with 1 ml lysis buffer containing 50 mM imidazole . Elution of bound proteins was carried out by incubation of the Ni-NTA Agarose beads with 1 ml of lysis buffer containing 500 mM imidazole for 5 min on a turning wheel at 4°C . The eluate ( sample E ) was transferred to a 1 . 5 ml Eppendorf tube and 100 μl thereof were mixed with 100 μl of 6x loading buffer . Protein samples ( 5 μl of samples T , P , S , and E ) were separated on NuPAGE 4–12% Bis-Tris 15-well gels ( Novex ) , run in 1x MES SDS running buffer , and subsequently stained with Brilliant Blue G Colloidal Coomassie ( Sigma ) . For Western analysis , appropriate dilutions of the above samples were separated on Bolt 4–12% Bis-Tris Plus 15-well gels ( Novex ) , run in 1x MES SDS running buffer , and proteins were subsequently blotted onto nitrocellulose membranes ( GE Healthcare ) . Cell extracts for polysome profile analyses were prepared as previously described [56] and eight A260 units were layered onto 10–50% sucrose gradients that were centrifuged at 38’000 rpm in a Sorvall TH-641 rotor at 4°C for 2 h 45 min . Sucrose gradients were analysed using an ISCO UA-6 system with continuous monitoring at A254 . For the fractionation experiments , five A260 units were subjected to sucrose gradient centrifugation for 2h 45 min and 20 fractions of around 500 μl were collected and processed as described [57] . Precipitated proteins were resuspended in 50 μl 3x sample buffer and 5 μl each fraction was separated on NuPAGE 4–12% Bis-Tris 26-well gels ( Novex ) , run in 1x MES or 1x MOPS SDS running buffer , and subsequently analyzed by Western blotting . As an input control , 0 . 05 A260 units of total cell extract was run alongside the fractions . Total yeast protein extracts were prepared as previously described [58] . Cultures were grown to an OD600 of around 0 . 8 and protein extracts were prepared from an equivalent of one OD600 of cells . Western blot analysis was carried out according to standard protocols . The following primary antibodies were used in this study: mouse monoclonal anti-FLAG ( 1:2’000–1:10’000; Sigma ) , anti-GFP ( 1:2’000; Roche ) , anti-HA ( 1:3’000; BAbCO ) , anti-His6 ( 1:500; Roche ) , and anti-Rpl3 ( 1:5’000; J . Warner , Albert Einstein College of Medicine , New York ) ; rabbit polyclonal anti-Adh1 ( 1:50’000; obtained from the laboratory of C . De Virgilio , University of Fribourg ) , anti-CBP ( 1:15’000; Open Biosystems ) , anti-Rpl5 ( 1:5’000; S . R . Valentini , São Paulo State University , Araraquara ) , and anti-Rps3 ( 1:20’000; M . Seedorf , ZMBH , University of Heidelberg , Heidelberg ) . Secondary goat anti-mouse or anti-rabbit horseradish peroxidase-conjugated antibodies ( Bio-Rad ) were used at a dilution of 1:10’000 . For detection of TAP-tagged proteins , the Peroxidase-Anti-Peroxidase soluble complex was used at a dilution of 1:20’000 ( Sigma ) . Immobilized protein-antibody complexes were visualized by using enhanced chemiluminescence detection kits ( Amersham ECL , GE Healthcare; PicoDetect , Applichem; WesternBright Sirius , Advansta ) . Total RNA was extracted from exponentially grown cells ( 10 OD600 units ) by the acid-phenol method and equal amounts of total RNA ( 5 μg ) were separated on 1 . 2% agarose gels containing 6% formaldehyde or on 7% polyacrylamide gels containing 8 M urea . Northern hybridization was performed as previously described [59] , utilizing the following oligonucleotides as probes: Probe b ( 18S ) 5’-CATGGCTTAATCTTTGAGAC-3’ Probe c ( D/A2 ) 5-GACTCTCCATCTCTTGTCTTCTTG-3’ Probe d ( A2/A3 ) 5’-TGTTACCTCTGGGCCC-3’ Probe e ( 5 . 8S ) 5’-TTTCGCTGCGTTCTTCATC-3’ Probe f ( E/C2 ) 5’-GGCCAGCAATTTCAAGTTA-3’ Probe g ( C1/C2 ) 5’-GAACATTGTTCGCCTAGA-3’ Probe h ( 25S ) 5’-CTCCGCTTATTGATATGC-3’ Probe 5S 5’-GGTCACCCACTACACTACTCGG-3’ The radioactive signals on the hybridized membranes were revealed using the Typhoon FLA 9400 imaging system and the supplied software ( GE Healthcare ) . For the determination of the rRNA composition of ribosomal particles , GFP-tagged Rpl4a was precipitated by a one-step GFP-Trap_A procedure that was slightly modified from the one suggested in the manufacturer’s instructions ( ChromoTek ) . Briefly , wild-type cells expressing untagged Rpl4a ( negative control ) or N-terminally yEGFP-tagged Rpl4a or Rpl4a . N264 from plasmid were grown in 200 ml SC-Leu medium to an OD600 of 0 . 8 . Cells were then washed twice with ice-cold water and finally resuspended in 500 μl of ice-cold lysis buffer ( 20 mM Tris-HCl pH 8 . 0 , 5 mM Magnesium acetate , 200 mM KCl , 0 . 2% Triton X-100 ) supplemented with 1 mM DTT and containing a protease inhibitor cocktail ( Complete , Roche ) . Cells were disrupted with glass beads by vigorous vortexing at 4°C for 12 min . Lysates were clarified by centrifugation in a microfuge at the maximum speed ( approximately 16’100x g ) for 15 min at 4°C . To each of the resulting total cell extracts , 30 μl of GFP-Trap_A beads , equilibrated with the same buffer , were added and the mixture was incubated for 1 h 30 min at 4°C with end-over-end tube rotation . After incubation , the beads were extensively washed seven times with 1 ml of the same buffer at 4°C and finally collected . RNA was extracted from the beads and the total cell extracts as previously described [60] , and the extracted RNA was analysed by Northern blotting as above . The Δacl4 mutant cells , either containing empty vector or a centromeric plasmid expressing Acl4 from the ADH1 promoter , were grown in a volume of 100 ml to an OD600 of around 0 . 7 and expression of C-terminally 2xHA-tagged Rpl4a was induced for 20 min from the CUP1 promoter with 500 μM copper sulfate . After harvesting , cells were lysed with glass beads in a buffer containing 50 mM Tris-HCl pH 7 . 5 , 100 mM NaCl , 1 . 5 mM MgCl2 , 5% glycerol , and 0 . 1% NP-40 and cell extracts were centrifuged for 3 min at 3’000 rpm . Then , total cell extracts , 10 A260 units in a final volume of 500 μl , were subjected to centrifugation at 200’000 g for 1 h . Pellets were resuspended in 100 μl lysis buffer and equal amounts of the total extracts ( T ) , soluble extracts ( S ) , and pellet fractions ( P ) were analyzed by SDS-PAGE and Western blotting using an anti-HA antibody . Co-translational association of Syo1-FTpA and Acl4-TAP with nascent r-proteins was assessed by IgG-Sepharose pull-down and real-time quantitative reverse transcription PCR ( real-time qRT-PCR ) as previously described [38] . Oligonucleotide pairs for the specific amplification of DNA fragments , corresponding to the RPL3 and RPL5 mRNA , from the input cDNAs , obtained from total RNA or chaperone-associated RNA , have been previously described [38] . The following oligonucleotide pairs were used for the specific amplification of DNA fragments corresponding to the RPL4 and RPL11 mRNAs: RPL4-I-forward 5’-ACCTCCGCTGAATCCTGGGGT-3’ RPL4-I-reverse 5’-ACCGGTACCACCACCACCAA-3’ ( amplicon size 72 bp ) RPL11-I-forward 5’-ACACTGTCAGAACTTTCGGT-3’ RPL11-I-reverse 5’-TTTCTTCAGCCTTTGGACCT-3’ ( amplicon size 81 bp ) Multiple sequence alignments of orthologous proteins were generated in the ClustalW output format with T-Coffee using the default settings of the EBI website interface [61] . Secondary structure prediction was performed with the PSIPRED v3 . 3 prediction method available at the PSIPRED website interface [62] . Potential tetratrico peptide repeats within Acl4 were identified by using the TPRpred website interface [63] , in combination with secondary structure prediction , multiple sequence alignments , and visual inspection of the occurrence of TPR consensus residues [46 , 47] . To identify orthologues of S . cerevisiae Acl4 , the sequence of the protein YD161_YEAST was searched for orthologues against the OMA database for orthology prediction ( http://omabrowser . org/cgi-bin/gateway . pl ? f=DisplayEntry&p1=YD161_YEAST ) [64] . OMA identified 1:1 orthologues in 23 species in the first group , with orthologues in fungi , parasites , and fishes . A total of 97 orthologous sequences from eight groups ( OMA groups: 351324 , 181749 , 130365 , 204390 , 227596 , 336561 , 539094 , and 370534 ) , as well as their corresponding NCBI taxid , were extracted from the orthoXML file . One group ( 273573 with six sequences ) was excluded since it did not contain the characteristic TPR-repeat domain of this family . The sequences were aligned with MAFFT [65] and viewed with Jalview [66] . The taxids were pasted into the phyloT web server ( http://phylot . biobyte . de ) to generate the tree and forwarded to the iTOL web site to visualize the tree [67] . As shown by the multiple sequence alignment and the tree , the gene encoding Acl4 orthologues is found mainly in fungi , but some dispersed branches kept it ( e . g . : some fishes , invertebrates , and parasites ) . Analysis and image preparation of three-dimensional structures , downloaded from the PDB archive , was carried out with the PyMOL ( PyMOL Molecular Graphics System; http://pymol . org/ ) or Chimera ( http://www . cgl . ucsf . edu/chimera ) software . The coordinates of the following ribosome structures were used: S . cerevisiae 60S subunit ( PDB 3U5H and 3U5I; [15] ) , S . cerevisiae 80S ribosome ( PDB 4V88; [15] ) , and Haloarcula marismortui 50S subunit ( PDB 4V9F; [68 , 69] . The representation of the electrostatic surface potential of the universally conserved part of yeast and archaeal Rpl4 was generated with Chimera by coulombic surface colouring .
|
Ribosomes are the molecular machines that generate proteins from mRNA templates . The biogenesis of eukaryotic ribosomes is an outstandingly complex process , in which around 80 ribosomal proteins and four ribosomal RNAs are accurately pieced together . Actively growing yeast cells must produce more than 160’000 ribosomal proteins per minute in order to meet the cellular demand for new ribosomes . Many ribosomal proteins are prone to aggregation and need therefore to be protected on their path from the cytoplasm to their mostly nuclear incorporation sites within ribosome precursors . Recent evidence has highlighted that specific binding partners , referred to as dedicated chaperones , may ensure the soluble expression , nuclear import and/or correct assembly of ribosomal proteins . Here , we have identified such a dedicated chaperone , termed Acl4 , which exclusively interacts with and accompanies the ribosomal protein Rpl4 to its nuclear assembly site . Notably , Acl4 has the capacity to recognize Rpl4 as it is synthesized by the ribosome . Our findings emphasize that co-translational capturing of ribosomal proteins by dedicated chaperones is an advantageous strategy to provide sufficient amounts of assembly-competent ribosomal proteins . A detailed knowledge of eukaryotic ribosome assembly is instrumental to eventually understand and treat ribosomopathies , diseases frequently caused by altered functionalities of ribosomal proteins .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[] |
2015
|
The Dedicated Chaperone Acl4 Escorts Ribosomal Protein Rpl4 to Its Nuclear Pre-60S Assembly Site
|
Kaposi's sarcoma ( KS ) is a mesenchymal tumour , which is caused by Kaposi's sarcoma herpesvirus ( KSHV ) and develops under inflammatory conditions . KSHV-infected endothelial spindle cells , the neoplastic cells in KS , show increased invasiveness , attributed to the elevated expression of metalloproteinases ( MMPs ) and cyclooxygenase-2 ( COX-2 ) . The majority of these spindle cells harbour latent KSHV genomes , while a minority undergoes lytic reactivation with subsequent production of new virions and viral or cellular chemo- and cytokines , which may promote tumour invasion and dissemination . In order to better understand KSHV pathogenesis , we investigated cellular mechanisms underlying the lytic reactivation of KSHV . Using a combination of small molecule library screening and siRNA silencing we found a STE20 kinase family member , MAP4K4 , to be involved in KSHV reactivation from latency and to contribute to the invasive phenotype of KSHV-infected endothelial cells by regulating COX-2 , MMP-7 , and MMP-13 expression . This kinase is also highly expressed in KS spindle cells in vivo . These findings suggest that MAP4K4 , a known mediator of inflammation , is involved in KS aetiology by regulating KSHV lytic reactivation , expression of MMPs and COX-2 , and , thereby modulating invasiveness of KSHV-infected endothelial cells .
Kaposi's sarcoma ( KS ) is a mesenchymal tumour caused by Kaposi's sarcoma herpesvirus ( KSHV ) [1] , which originates from blood and lymphatic vessels and develops under the influence of inflammatory cytokines [2]–[4] . Local or systemic inflammation and immunosuppression are important additional risk factors [5] , [6] . In addition to KS , KSHV is involved in the pathogenesis of primary effusion lymphoma ( PEL ) [7] , and the plasma cell variant of multicentric Castleman's disease ( MCD ) [8] . KS is characterised by multiple patch , plaque or nodular lesions on the skin of the extremities or involving the mucosa and visceral organs [9] . KSHV-infected spindle cells , which were shown to be of vascular or lymphatic endothelial origin , represent the main proliferative element in KS and are the distinctive histological signature of advanced KS tumours [10] , [11] . The lesions also contain slit-like neovascular spaces , which represent aberrant new vessels [5] , [12] . KS spindle cells were shown to have increased invasiveness [13] , which has been attributed to the enhanced expression of several matrix metalloproteinases ( MMPs ) [14] , including MMP-1 , MMP-2 , MMP-3 , MMP-7 , MMP-9 , and MMP-13 [13] , [15] , [16] . MMPs are zinc-dependent endopeptidases involved in extracellular matrix remodelling during tumour progression , invasion and metastasis [17] , [18] . In addition to MMPs , the key enzyme for inducible prostaglandin synthesis – cyclooxygenase 2 ( COX-2 ) [19] – has also been implicated in KS progression and invasion [20] . Increased COX-2 expression in inflammation-driven tumours contributes to neoangiogenesis and activates MMPs , which promote invasiveness [21] , [22] . COX-2 is highly expressed in KS tumour tissue and is involved in KS pathogenesis [20] , [23] , [24] . Several KSHV proteins were shown to enhance COX-2 expression , including K15 [25] , and vGPCR [26] . This could explain how KSHV may increase COX-2 gene expression . In KS tumours , the majority of KSHV-infected cells harbour latent viral genomes , which are characterised by a restricted viral gene expression pattern that involves the major latent nuclear antigen LANA , homologues of a cellular D-type cyclin and a FLICE inhibitory protein , v-Cyclin and v-FLIP , respectively , and 12 viral miRNAs [6] , [27] . However , a minority of infected cells show evidence of productive ( ‘lytic’ ) replication and produce not only new virions [28] , but also secrete viral or cellular cyto- or chemokines [6] , [10] , [27] , [29] , [30] . These are thought to promote the pathological angiogenesis typical for KS lesions , increased invasion , and tumour dissemination [31] . Epidemiological findings also indicate that the prophylactic use of ganciclovir , which inhibits KSHV lytic replication , may reduce the incidence of KS in AIDS patients [32] . In addition , it is thought that the long-term persistence of KSHV in vivo may require periodic reactivation from latency and reinfection of new cells [33] . Experimentally , reactivation of KSHV from latency can be initiated by various chemical agents: these include phorbol esters and histone deacetylase inhibitors , which lead to chromatin remodelling and activation of the viral replication and transcription activator ( RTA ) [34]–[37] . So far , several signalling pathways were reported to be involved in the reactivation of KSHV from latency: PKCδ [38] , b-Raf/MEK/ERK [39] , PKA [40] , Notch and RBP-Jκ [41] , [42] , p38 and JNK [43] , Pim-1 and Pim-3 [44] , PI3K and Akt [45] , TLR7/8 signalling [46] and others . Given the importance of the KSHV lytic cycle in KS pathogenesis and the angiogenic and invasive phenotype of KSHV infected cells , we aimed at identifying ‘druggable’ cellular kinases required for KSHV reactivation from latency . To this end , we screened a library of kinase inhibitors and found the STE20 kinase family member MAP4K4 to be a novel mediator of KSHV lytic reactivation . MAP4K4 is known to play an important role in inflammation , insulin resistance , and invasiveness of several malignancies [3] , [47]–[55] . We found that MAP4K4 regulates the expression of COX-2 , MMP-7 and -13 , and thereby modulates the invasiveness of KSHV infected primary and immortalized endothelial cells . Moreover , we found MAP4K4 to be strongly expressed in KSHV-infected endothelial spindle cells in KS tissue , consistent with a role of MAP4K4 in KS pathogenesis .
Productive replication of KSHV in infected individuals is thought to contribute to viral persistence and the pathogenesis of this virus [56] , [57] . Activation of several cellular kinases , involved in different signalling pathways , promotes viral reactivation [58] , [59] . In order to identify novel “druggable” cellular kinases required for KSHV reactivation we screened a library of 486 small molecule kinase inhibitors ( figure 1A ) in a KSHV reactivation assay based on Vero cells infected with the recombinant KSHV strain rKSHV . 219 ( VK . 219 ) [60] . The activation of productive replication cycle was achieved by treatment with Na-butyrate and infection with a baculovirus expressing KSHV immediate-early protein RTA . Toxicity of the compounds was determined by crystal violet staining of VK . 219 and HEK293 cells after treatment . As a result , 105 compounds showed moderate to strong effects on virus production and infectivity without being toxic . Among them , 92 compounds were able to directly inhibit KSHV lytic protein expression in VK . 219 cells . The results were validated in BCBL1 [61] , and KSHV-infected EA . hy 926 [62] cells . As a result , we identified 18 compounds able to inhibit KSHV lytic protein expression in all three cell lines ( figure S1A ) . Interestingly , among them were 11 compounds identical to , or derived from , known p38 MAP kinase inhibitors , in line with earlier reports on the role of this kinase in KSHV reactivation [43] , [58] . When comparing the effects of commercially available p38 inhibitors with compounds in the VICHEM library , we noted that p38 inhibitors SB202190 , SB203580 , VX745 , SKF86002 , SB220025 , and a derivative of SB220025 ( VI18802 ) differed in their ability to block KSHV reactivation , as shown by their effect on the expression of KSHV envelope glycoprotein K8 . 1 ( figure 1B ) , although their ability to inhibit the phosphorylation of MK2 , a p38 target , seemed comparable ( figure 1B ) . Of these compounds , SB220025 was the most potent with regard to inhibiting K8 . 1 expression ( figure 1B ) or virus production ( not shown ) . To validate the effect of SB220025 on KSHV reactivation , we titrated this compound in KSHV-infected endothelial cells ( KSHV-infected EA . hy 926 ) and found it to inhibit KSHV reactivation at submicromolar concentrations ( figure 1C ) . We then determined which other cellular kinases are inhibited by SB220025 and its derivative VI18802 . Compound SKF86002 , although a strong inhibitor of lytic reactivation , was not included in this comparison , as it reduced the levels of total MK2 ( figure 1B ) . We used a commercial screening assay that measured the ability of these compounds to compete with an immobilized ligand for binding to a panel of 442 recombinant kinases in an in vitro assay ( see www . discoverx . com ) . The list of cellular kinases inhibited by SB220025 and VI18802 is shown in a table presented in figure S1B , which also includes previously published data on compounds SB203580 , SB202190 and VX745 [63] . To explore if , apart from p38 , any of the other kinases inhibited by SB220025 could account for the strong inhibition of KSHV reactivation observed with this compound , we used small molecule inhibitors or siRNAs against CSNK1D , CSNK1E , CSNK1A1L , MINK , CDC2L1/2 , JNK1 , MAP4K4 , STK36 and TNIK ( data not shown ) . As a result of these experiments we identified the upstream MAP kinase MAP4K4 ( data not shown ) , a member of the STE20 kinase family , which has previously been shown to be involved in inflammation , response to LPS , inflammation-dependent insulin resistance of peripheral tissues , and also invasiveness of several types of cancer cells [47] , [48] , [53] , [64] , [65] . In order to explore if MAP4K4 affected KSHV reactivation in a cell type that is known to be infected by KSHV in vivo , we used a pool of siRNAs to silence MAP4K4 expression in the immortalized HUVEC derived cell line EA . hy 926 [62] , which we had stably infected with rKSHV . 219 . As shown in figure 2 , silencing of MAP4K4 in these cells significantly reduced production of infectious viral progeny by more than 60% ( figure 2A ) , as well as the expression of immediate-early ( RTA ) , early ( KbZIP , ORF45 ) and late ( K8 . 1 ) lytic proteins ( figure 2B ) . The effect on K8 . 1 expression was confirmed using four individual siRNAs targeting MAP4K4 , all of which were able to reduce MAP4K4 and K8 . 1 levels ( figure S2A , B ) . In contrast to other lytic KSHV proteins , the expression of the viral homologue of IL-6 , vIL-6 , was slightly increased by MAP4K4 knockdown ( figure 2B ) . vIL-6 expression is known to be regulated independently of the productive replication cycle [66] and may therefore not be affected by MAP4K4 silencing . Consistently with the observed decrease in virus production and lytic protein expression , MAP4K4 depletion also reduced KSHV genome replication ( figure 2C ) , similarly to foscarnet , an inhibitor of KSHV DNA polymerase [67] . However , while foscarnet only inhibited the expression of a late viral gene ( K8 . 1 ) , MAP4K4 silencing also affected early KSHV gene ( KbZIP ) expression ( figure 2D ) , suggesting that this kinase exerts its effect early in the replication cycle . To control whether MAP4K4 knockdown affects transduction or expression of baculovirus RTA , we evaluated the levels of RTA mRNA transcripts before and after MAP4K4 depletion in cells not infected with KSHV that had been treated with baculovirus RTA alone or in combination with Na-butyrate . In these cells , RTA expression was not dependent on MAP4K4 presence ( figure S2C–D ) . Taken together , the observed decrease in KSHV titre , lytic protein expression , and replication in the absence of MAP4K4 suggests that this kinase contributes to the successful completion of the KSHV lytic programme . MAP4K4 is also known to promote tumour cell migration , invasion , and loss of adhesion [49] , [68] . KS tumour derived cells have been reported to show an invasive phenotype [69] . This phenomenon can be studied in vitro in a matrigel-based invasion assay , in which uninfected HuAR2T , a conditionally immortalized HUVEC cell line [70] , fails to invade into matrigel , whereas HuAR2T cells infected with rKSHV . 219 show increased invasiveness after the treatment with Na-butyrate to induce the KSHV lytic replication cycle ( figure 3A–B ) . Thus , lytic reactivation of the virus promotes invasiveness of these immortalized endothelial cells infected with KSHV . As we observed that MAP4K4 supports the KSHV lytic cycle ( figure 2 ) and since it had been reported to be a promigratory kinase [49] , we investigated if its silencing might affect the ability of KSHV-infected endothelial cells to invade matrigel . Indeed , after silencing of MAP4K4 expression with siRNA , KSHV-infected HuAR2T endothelial cells failed to invade matrigel beyond the levels seen in uninfected control cells ( figure 3C–D ) . MAP4K4 and KSHV lytic protein expression was controlled by Western blot analysis as presented in figure 3E . Together , these data suggest a role for MAP4K4 signalling in the KSHV-dependent invasiveness of infected endothelial cells . In an attempt to understand how MAP4K4 promotes lytic reactivation and leads to the increased invasiveness of KSHV-infected endothelial cells we compared the transcriptome of reactivated KSHV-infected HuAR2T cells , in which the expression of MAP4K4 had been silenced with siRNA , with KSHV-infected , reactivated HuAR2T cells treated with control siRNA . We were able to identify 54 cellular genes that showed at least a 1 . 5-fold decrease in their expression levels after MAP4K4 knockdown in HuAR2T rKSHV . 219 undergoing viral reactivation as compared to control siRNA treated , reactivated HuAR2T rKSHV . 219 cells in at least two out of three independent experiments ( figure 4A ) . Successful knockdown of MAP4K4 , and the subsequent inhibition of lytic gene expression , was controlled by Western blot analysis ( figure S2E ) . Among the cellular genes regulated by MAP4K4 silencing in KSHV-infected endothelial cells were three that have previously been reported to contribute to the invasive phenotype of tumour cells: PTGS2 , encoding cyclooxygenase 2 ( COX-2 ) , and the genes coding for matrix metalloproteinases 7 and 13 ( MMP-7 and MMP-13 ) ( figure 4A ) . In order to validate the results of the transcriptome analysis , the expression levels of COX-2 were evaluated by qPCR and Western blot analysis before and after the induction of the lytic cycle . As shown in figure 4B–C , COX-2 mRNA and protein expression is upregulated following induction of the viral lytic cycle and can be reduced by silencing MAP4K4 . Likewise , we could show that the expression of both MMP-7 and MMP-13 mRNAs increased after the induction of the lytic cycle and was significantly reduced after MAP4K4 depletion ( figure 4B ) . These data support the notion that MAP4K4 may mediate the increased invasiveness of KSHV-infected endothelial cells due to its ability to modulate not only COX-2 , but also MMP-7 and MMP-13 expression . KS cells are known to express high levels of MMP-1 , -2 , -3 , -7 , -9 , -13 , -19 , and previous reports suggest that some of these metalloproteinases may contribute to the invasive phenotype of the tumour [14] , [16] , [71] , [72] . Overexpression of MMP-7 has been reported in several other malignancies [73]–[75] , and its depletion with siRNA resulted in a significant decrease in the invasive potential of different cancer cell types [76]–[78] . Similarly , MMP-13 has been reported to confer the ability to penetrate basement membranes and ECM upon malignant cells [79] . Given these proinvasive properties of MMP-7 and MMP-13 , and taking into account the ability of MAP4K4 to regulate their expression ( figure 4 ) , we addressed the involvement of these metalloproteinases in the invasiveness of KSHV-infected cells in a matrigel-based invasion assay . We found that depletion of both MMP-7 and MMP-13 , similarly to MAP4K4 knockdown , led to a significant reduction of the number of invasive KSHV-infected endothelial HuAR2T cells following activation of the viral lytic replication cycle ( figure 5A ) . The efficiency of silencing the expression of MAP4K4 , MMP-7 and MMP-13 with siRNA was controlled by Western blot analysis for MAP4K4 ( figure 5B ) and qPCR for MMP-7 and MMP-13 ( figure 5C ) . We noted that silencing of MAP4K4 led to a reduced expression of the early KSHV protein KbZIP ( figure 5B ) and its mRNA transcript , as well as K8 . 1 mRNA expression ( figure 5D ) , whereas silencing of MMP-7 and MMP-13 had no effect on the protein and mRNA levels of KbZIP ( figure 5B , 5D ) or mRNA levels of K8 . 1 ( figure 5D ) . These results suggest that MAP4K4 is involved in the activation of the lytic replication cycle , which , in turn , promotes the expression of MMP-7 and MMP-13 . As shown in figure 4 , silencing of MAP4K4 reduces the expression of PTGS2 , encoding cyclooxygenase 2 ( COX-2 ) . COX-2 has previously been shown to be overexpressed in KSHV-infected endothelial cells and to play a role in inflammation , angiogenesis and invasion [20] . The KSHV K15 and vGPCR proteins induce the expression of COX-2 [25] , [26] . COX-2 catalyses the production of prostaglandin E2 ( PGE2 ) after stimulation with inflammatory cytokines [80] . Depletion of COX-2 reduced invasiveness of KSHV-infected endothelial cells , similar to MAP4K4 knockdown ( figure 6A ) . Interestingly , both MAP4K4 and COX-2 silencing inhibited KSHV lytic reactivation ( figure 6B ) . To corroborate the effect of COX-2 depletion on KSHV lytic reactivation , we used a specific inhibitor , which does not affect constitutively active COX-1 [81] . Application of this inhibitor , NS-398 , led to a dramatic decrease , comparable to the effect of MAP4K4 silencing , in the invasiveness of KSHV-infected endothelial cells undergoing lytic reactivation ( figure 6C ) . NS-398 treatment not only led to a reduction of invasiveness , but also effectively blocked KSHV lytic protein expression ( figure 6D–E ) , as well as the production of viral progeny ( figure 6F ) . This suggests that , in response to MAP4K4 signalling , COX-2 mediated production of PGE2 contributes to the successful completion of KSHV lytic cycle and KSHV driven invasiveness of infected endothelial cells . To extend our observations , which were obtained with the immortalized endothelial cell line HuAR2T , to primary endothelial cells , we investigated the role of MAP4K4 in the invasiveness of human umbilical vein endothelial cells ( HUVECs ) following their infection with rKSHV . 219 ( figure 7 ) . On day 5 after infection , KSHV-infected HUVECs showed a markedly increased invasiveness compared to uninfected cells , and this increased invasiveness depended on the expression of MAP4K4 , since silencing of MAP4K4 with siRNA reduced their invasiveness to background levels ( figure 7A–B ) . Similar to KSHV-infected HuAR2T cells , expression of COX-2 increased after infection of HUVECs with KSHV and silencing of MAP4K4 by siRNA reduced COX-2 levels in KSHV-infected primary endothelial cells ( figure 7C ) . We also observed that after infection with KSHV , MAP4K4 protein levels were moderately increased ( figure 7C–D ) . Moreover , KSHV lytic protein expression was inhibited after MAP4K4 depletion in primary cells , similarly to what we had found in immortalized endothelial cells ( figure 7D ) . To explore if MAP4K4 is expressed in KS tissue and could , therefore , play a role in KSHV-infected cells in vivo and contribute to the pathogenesis of KS , we stained KS biopsies with an antibody to MAP4K4 . We observed a strong expression of MAP4K4 in the KS endothelial spindle cells , which are characterised by the expression of CD34 and KSHV LANA ( figure 8A ) . Double staining for LANA and MAP4K4 confirmed the strong cytoplasmic expression of MAP4K4 in LANA-expressing cells ( figure 8A ) . Individual staining for MAP4K4 and LANA of adjacent serial sections of a KS biopsy also indicated the increased expression of MAP4K4 in LANA-expressing KS spindle cells , although a lower level of MAP4K4 expression could also be seen in other cells in the tumour ( figure 8B–C ) , and a basal expression of MAP4K4 was observed in the surrounding connective tissue ( figure 8C ) , in line with another report showing low levels of MAP4K4 cytoplasmic staining in non-neoplastic lung tissues , compared to strong expression in lung adenocarcinomas [82] . We found a moderate to strong expression of MAP4K4 in spindle cells in a total of 13 biopsies , derived from 11 patients ( figure 8D ) , confirming the consistent expression of this kinase in KS tissue . This observation is consistent with a role for MAP4K4 and MAP4K4-dependent signalling pathways in the pathogenesis of KS .
In KS tumours , a small percentage ( 1–5% ) of KSHV-infected cells show evidence of viral lytic replication [31] , [83] . Taken together with epidemiological findings indicating a beneficial effect of inhibiting viral lytic replication on the incidence of KS in AIDS patients [32] this suggests that lytic gene products may contribute to the pathogenesis of this disease . On the one hand , lytic replication can be a source of new virions and consequently newly infected cells . This is important , as KSHV does not completely immortalize spindle cells and needs to infect new cells to persist in an infected host [33] . On the other hand , lytic reactivation may lead to the production of autocrine and paracrine signalling molecules , which then promote inflammation , angiogenesis , and invasiveness . KSHV-infected endothelial spindle cells have been shown to have invasive properties [16] , [20] , [69] , [84]–[86] . In order to better understand how KSHV lytic replication cycle contributes to the increased invasiveness of infected endothelial spindle cells we investigated cellular mechanisms underlying the lytic switch of the virus . In contrast to earlier studies that had employed siRNA screens of the human kinome to identify cellular kinases involved in KSHV reactivation and had identified Pim kinases as activators of lytic replication [44] , or Tousled-like kinases as negative modulators of KSHV reactivation [87] , we screened a library of small molecule kinase inhibitors ( figure 1A ) to identify positive regulators of KSHV lytic cycle . We found several compounds , known to target p38 MAPK , to inhibit KSHV reactivation after baculovirus RTA and Na-butyrate treatment ( figure S1A ) , in line with previous reports on a role of p38 during de novo infection [88] , after induction of productive reactivation [43] , and during progression of KSHV through the lytic cycle , when , for instance , vGPCR activates p38 [89] . However , a close comparison of well-characterized p38 inhibitors [90]–[94] , showed that these compounds varied with regard to their ability to inhibit KSHV reactivation in endothelial cells , while showing comparable efficacy in inhibiting the phosphorylation of the p38 MAPK target MK2 ( figure 1B ) . This observation suggested that some of these compounds might also target other cellular kinases , which could contribute to KSHV reactivation . Off-target effects of other kinase inhibitors are well known and sometimes improve the biological activity and clinical usefulness of individual compounds [63] , [95] . Since compound SB220025 , which is known to have anti-inflammatory properties , proved to be the most efficient in reducing KSHV lytic reactivation ( figure 1B–C ) , we profiled this substance together with VI18802 , a derivative of SB220025 , against 442 kinases using the KINOMEscan platform ( DiscoverX ) . Extending previous reports on the ability of even “specific” p38 inhibitors to bind to other kinases [63] , we found a range of other kinases to be inhibited by SB220025 ( figure S1B ) . By blocking , among others , the p38 cascade , SB220025 inhibits the production of IL-1β and TNF-α [91] , [96] , and belongs to the CSAID class of cytokine biosynthesis inhibitors [97] , [98] . However , p38 is not the only regulator of inflammatory cytokine production . JNKs also regulate the expression and activation of inflammatory mediators , including TNF-α , IL-2 , and MMPs [99] , [100] . Interestingly , we identified several JNK isoforms and their putative upstream activators MAP4K4 ( NCK interacting kinase ( NIK ) or haematopoietic/germinal centre kinase ( HGK ) ) , MINK ( Misshapen/NIK related kinase ) , and TNIK ( TRAF2 and NCK interacting kinase ) as targets of SB220025 ( figure S1B ) . KSHV is known to activate the JNK pathway during primary infection [101] , and JNK is essential for KSHV infection [58] , and production of inflammatory cytokines [31] , [101] , [102] . Considering the important role of inflammation in KS development and progression , and the dependence of KSHV on the JNK pathway , we investigated if upstream regulators of JNK signalling targeted by SB220025 ( MAP4K4/NIK , TNIK , MINK ) are also critical for KSHV lytic cycle . While siRNA-mediated knockdown of TNIK and MINK did not affect KSHV reactivation ( data not shown ) , MAP4K4 silencing reduced KSHV virus production ( figure 2A ) , lytic protein expression ( figure 2B ) , and KSHV replication ( figure 2C ) in immortalized , as well as primary endothelial cells ( figure 7D ) . Interestingly , vIL-6 expression levels were increased after MAP4K4 silencing ( figure 2B ) . Although vIL-6 is a lytic gene induced by RTA [103] , it is known to be also regulated independently of the lytic switch , for instance by interferon-α [66] and microRNAs , such as miR-1293 [104] . Whether MAP4K4 also regulates the latter factors needs to be further investigated , and perhaps would explain the observed increase in vIL-6 expression in the absence of MAP4K4 . As MAP4K4 was previously shown to be overexpressed in multiple tumour cell lines and cancers [49]–[51] , [105] , [106] , and also implicated in tumour cell invasiveness [49] , [106] , we investigated its role in previously reported invasiveness of KSHV-infected endothelial cells . We could observe that KSHV-infected immortalized endothelial cells possess a much more invasive phenotype after the induction of the lytic cycle ( figure 3A–B ) . This increased invasiveness could be reduced by MAP4K4 silencing using siRNA ( figure 3C–E ) , demonstrating a role of MAP4K4 in invasive KSHV-infected endothelial cells . Similarly , silencing of MAP4K4 reduced the increased invasiveness of KSHV-infected primary umbilical vein endothelial cells ( figure 7A–C ) . The role of MAP4K4 in different cellular functions is only incompletely understood . In order to identify genes , regulated by MAP4K4 in the context of KSHV lytic reactivation , we performed a microarray-based analysis after silencing MAP4K4 and inducing the lytic cycle . Among cellular genes known to affect migration/invasion , we found PTGS2 , encoding COX-2 , to be downregulated after MAP4K4 knockdown ( figure 4A ) . Its mRNA and protein levels were highly upregulated in induced KSHV-infected cells compared to uninfected cells ( figure 4A ) . Increased levels of PGE2 in Kaposi's sarcoma tissue compared to surrounding tissues were reported already in 1992 [107] . In keeping with this observation , KSHV infected immortalized dermal microvascular endothelial cells display a strong increase in COX-2 expression and PGE2 production early during de novo infection [20] , [23] , [24] , when lytic replication may still take place [108] . Our finding suggests that COX-2 activation is , at least in part , mediated by MAP4K4 and is critical for KSHV lytic cycle progression , as treatment with a specific COX-2 inhibitor NS-398 ( figure 6E ) or COX-2 depletion ( figure 6B ) led to a dramatic decrease in expression of KSHV lytic proteins . COX-2 inhibitors are known to also block human cytomegalovirus replication [109] , [110] , as PGE2 enhances , for instance , CMV promoter activation [111] . COX-2 activation might also play a role in HHV-6 [112] , MHV-68 [113] , and HSV-1 [114] replication . Of note , MAP4K4 expression levels after COX-2 depletion and its chemical inhibition were slightly reduced ( figure 6B , 6D–E ) . MAP4K4 expression is regulated by TNF-α through TNF receptor α [53] , the expression levels of which in turn depend on PGE2 activation [115] . Hence it is conceivable that , when PGE2 production is downregulated by chemical inhibition of COX-2 , MAP4K4 levels can also decrease as expression of TNF receptors is reduced . COX-2 is a known mediator of angiogenesis and tumour cell invasiveness , as it leads to production of inflammatory cytokines , growth factors , angiogenic factors , and MMPs in various tumours , as well as in KSHV infected cells [20] , [77] , [116]–[122] . We could also show that , similarly to MAP4K4 knockdown , COX-2 silencing or chemical inhibition significantly reduces the invasiveness of KSHV-infected endothelial cells ( figure 6A , 6C ) . We also found that MAP4K4 mediates the expression of MMP-7 and MMP-13 ( figure 4A–B ) , which both contribute to the invasiveness of KSHV-infected cells ( figure 5A ) . Although matrix metalloproteinases are known to be modulated post-transcriptionally [123]–[125] , most of them , including MMP-7 and MMP-13 , can be activated also at the transcriptional level , as their promoters harbour several cis-elements , allowing activation by trans-activators , e . g . NF-κB and AP-1 [126] , [127] . These MMPs can also be induced at the mRNA level by TNF-α , IL-1 and other cytokines [61] , [128]–[131] . Given that MAP4K4 regulates inflammatory cytokine production , such as TNF-α and IL-1β [47] , it may therefore also modulate MMP-7 and MMP-13 mRNA expression . Our observation that MAP4K4 regulates MMP-7 and MMP-13 expression illustrates its multifactorial role in the increased invasiveness of KSHV-infected endothelial cells . Given the reported role of MAP4K4 as an upstream activator of JNK [64] , and the role of JNK in KSHV reactivation [43] , we also explored if silencing of MAP4K4 in KSHV-infected endothelial cells would alter the levels of JNK 1/2/3 phosphorylation , using phospho-specific antibodies in Western blot analysis . However , we could not detect any prominent effect of MAP4K4 silencing on the levels of JNK phosphorylation ( figure S3A ) , consistent with an earlier report [47] . Searching for other cellular targets that would be phosphorylated in response to MAP4K4 , we employed a commercial phosphokinase array and noted a moderate decrease of c-Jun phosphorylation following MAP4K4 silencing ( figure S3B–C ) . This was confirmed in Western Blot analysis using an antibody to c-Jun phosphorylated on S63 ( figure S3D–E ) . Phosphorylation of c-Jun may therefore provide another explanation of how the upstream kinase MAP4K4 exerts its effect on MMP-7 and MMP-13 expression . It might also lead to COX-2 overexpression . However , other possibilities remain to be investigated , as well as the mechanism of how MAP4K4 is activated in KSHV-infected endothelial cells . Having shown that MAP4K4-dependent signalling pathways are involved in the increased invasiveness of KSHV-infected primary and immortalized endothelial cells , we could demonstrate that MAP4K4 is highly expressed in the pathognomonic KSHV-infected endothelial spindle cells in KS lesions ( figure 8A–D ) , suggesting that it may indeed play a role in vivo in aspects of KSHV-induced pathogenesis . Our findings also provide an explanation for the increased expression of COX-2 in KSHV-infected endothelial cells .
The use of the human biopsies and human umbilical cords for this study was approved by the Hannover Medical School Ethics Committee and conducted in accordance with the Declaration of Helsinki . Written informed consent was obtained from all patients . HEK293 and EA . hy 926 cells were maintained in Dulbecco's modified Eagle's medium ( DMEM ) , and Vero cells in minimal essential medium ( MEM ) ( Cytogen ) supplemented with 10% foetal bovine serum ( HyClone ) , 50 U/ml penicillin , and 50 µg/ml streptomycin ( Cytogen ) at 37°C in a 5% CO2 incubator . Human umbilical vein endothelial cells ( HUVEC ) were isolated from freshly obtained human umbilical cords by collagenase digestion of the interior of the umbilical vein as described previously [132] and were cultured in EGM-2MV medium ( Lonza ) at 37°C in a 5% CO2 incubator . An endothelial cell line HuAR2T , conditionally immortalized with doxycycline dependent human telomerase reverse transcriptase ( hTERT ) and simian virus 40 ( SV40 ) large T antigen transgene expression [133] , were maintained in EGM-2MV medium in the presence of 200 ng/ml doxycycline . Transfection with small interfering RNA ( siRNA ) was performed using the Neon transfection system according to the manufacturer's instructions ( Invitrogen ) . All siRNAs were microporated at the concentration of 100 pmol into 105 cells . The following siRNAs ( siGENOME SMARTpool ) were obtained from Dharmacon , Thermo Scientific: Control ( Non-targeting siRNA Pool #2 , D-001206-14-20 ) , MAP4K4 ( M-003971-02-0005 ) , MMP-7 ( M-003782-01-0005 ) , MMP-13 ( M-005955-01-0005 ) . Sf9 cells were maintained in Grace's medium ( Gibco ) supplemented with 10% foetal bovine serum , 100 U/ml penicillin , and 50 µg/ml streptomycin ( Cytogen ) at 28°C . The generation of the recombinant baculovirus expressing KSHV ORF50/RTA was described previously [60] . To produce virus stocks , Vero cells containing recombinant KSHV ( rKSHV . 219 ) [60] were plated at 30–40% confluency in T175 flasks and induced twenty-four hours later with 1 mM Na-butyrate ( Sigma-Aldrich ) and 10% baculovirus coding for KSHV ORF50/RTA . The supernatant was harvested 72 hours later and 0 . 45 µm filtered to remove cell debris . The cleared supernatant was collected in centrifuge bottles ( 230 ml/bottle ) and centrifuged at 15000×g at 4°C for 6 hours using a Type19 rotor ( Beckman Coulter ) . The supernatant was then discarded and the pellet resuspended in 250 µl EBM2 basal medium ( Lonza ) overnight at 4°C . The resuspended virus was kept at 4°C for not longer than three weeks . For detection and quantification of KSHV titres , 2 . 3×104 HEK293 cells were plated in a 96-well plate and infected with serially diluted KSHV stocks . GFP-positive cells were counted two days after infection . To determine KSHV titres from EA . hy rKSHV . 219 or HuAR2T rKSHV . 219 cells after induction of the lytic cycle , the supernatants were cleared from the debris by 0 . 45 µm filtration and applied to HEK293 cells without dilution . Protein lysates of cells were prepared in 1× SDS sample buffer ( 62 . 5 mM Tris-HCl pH 6 . 8 , 2% w/v SDS , 10% glycerol , 50 mM DTT , 0 . 01% w/v bromophenol blue ) supplemented with cOmplete Ultra protease inhibitor cocktail and PhosSTOP phosphatase inhibitor cocktail ( Roche ) . Proteins were resolved by SDS-PAGE , transferred onto nitrocellulose membranes ( GE Healthcare ) , and detected using the following primary antibodies: rabbit polyclonal MAP4K4 ( HGK ) antibody ( #3485 , Cell Signaling Technology ) , rabbit polyclonal KSHV ORF50/RTA [37] , mouse monoclonal KSHV ORF45 antibody ( sc-53883 , Santa Cruz ) , mouse monoclonal HHV-8 KbZIP antibody F33P1 ( sc-69797 , Santa Cruz ) , rabbit polyclonal KSHV vIL-6 antibody ( 13-214-050 , Advanced Biotechnologies ) , mouse monoclonal KSHV ORFK8 . 1A/B antibody ( 13-212-100 , Advanced Biotechnologies ) , mouse monoclonal β-actin antibody ( A5441 , Sigma-Aldrich ) , rabbit monoclonal GAPDH antibody ( #2118 , Cell Signaling Technology ) , rabbit polyclonal COX-2 antibody ( #4842 , Cell Signaling Technology ) , rat monoclonal KSHV ORF73 ( LNA-1 ) antibody ( 13-210-100 , Advanced Biotechnologies ) , mouse monoclonal phospho-JNK 1/2/3 antibody 9H8 ( sc-81502 , Santa Cruz ) , rabbit polyclonal JNK 1/3 antibody C17 ( sc-474 , Santa Cruz ) , mouse monoclonal phospho-p44/42 antibody ( #9106 , Cell Signaling Technology ) , mouse monoclonal p44/p42 antibody 3A7 ( #9107 , Cell Signaling Technology ) , rabbit polyclonal phospho-p38 antibody ( #9211 , Cell Signaling Technology ) , rabbit polyclonal p38 antibody ( #9212 , Cell Signaling Technology ) , rabbit monoclonal phospho-MK2 antibody 27B7 ( #3007 , Cell Signaling Technology ) , rabbit polyclonal MK2 antibody ( #3042 , Cell Signaling Technology ) . All stainings were performed at 4°C overnight with subsequent washing in TBS-T or PBS-T and incubation with a corresponding secondary HRP-labelled antibody ( DaKo ) at RT for one hour . Following further washing steps , the proteins were detected with SuperSignal West Femto Chemiluminescent Substrate ( Pierce , Thermo Scientific ) . The “Whole Human Genome Oligo Microarray V2” ( G4845A , ID 026652 , Agilent Technologies ) used in this study contains 44495 oligonucleotide probes covering roughly 27390 human transcripts . Synthesis of Cy3-labeled cRNA was performed with the “Quick Amp Labelling kit , one colour” ( #5190-0442 , Agilent Technologies ) according to the manufacturer's recommendations . cRNA fragmentation , hybridization , and washing steps were carried out exactly as recommended in the “One-Color Microarray-Based Gene Expression Analysis Protocol V5 . 7” ( Agilent ) . Slides were scanned on the Agilent Micro Array Scanner G2565CA ( pixel resolution 5 µm , bit depth 20 ) . Data extraction and processing of raw fluorescence intensity values were performed with the “Feature Extraction Software V10 . 7 . 3 . 1” by using the recommended default extraction protocol file: GE1_107_Sep09 . xml . Processed intensity values of the green channel ( “gProcessedSignal” or “gPS” ) were globally normalized by a linear scaling approach: All gPS values of one sample were multiplied by an array-specific scaling factor . This scaling factor was calculated by dividing a “reference 75th Percentile value” ( set as 1500 for the whole series ) by the 75th Percentile value of the particular Microarray ( “Array i” in the formula shown below ) . Accordingly , normalized gPS values for all samples ( microarray data sets ) were calculated by the following formula: normalized gPSArray i = gPSArray i X ( 1500/75th PercentileArray i ) . A lower intensity threshold was defined as 1% of the reference 75th Percentile value ( = 15 ) . All normalized gPS values below this intensity threshold were substituted by the surrogate value of 15 . Data were filtered according to the following criteria: 1 ) More than 1 . 5 fold downregulation in lytically induced HuAR2T rKSHV . 219 cells after MAP4K4 knockdown compared to control siRNA treated induced cells ( each of three experiments ) . 2 ) Arithmetic mean intensity of nPS values calculated from both channels that define ratio values >25 ( each of three experiments ) . 3 ) QC flag entries “gIsNonUnifOL” ( determined by the Feature Extraction Software ) must have been “0” ( indicating reliable performance ) ( each of six relevant channels of the three experiments ) . 4 ) In cases , in which more than one probe directed against the same transcript is present on the microarray , only those transcripts passed the criteria , for which the majority of probes indicate the respective regulation . 5 ) The respective transcript has to be classified as being functionally characterized and reasonably annotated ( for details visit: www . mh-hannover . de/Transcriptomics . html and consult our manual: “Crude probe characterization_RCUT_date . pdf” ) . Just one representative probe is selected for visualization in figure 4A if many probes directed against the same transcript match the applied criteria . Total RNA was extracted from the cells with an RNeasy kit ( QIAgen ) according to the manufacturer's recommendations , followed by DNase treatment and inactivation ( Ambion ) . cDNA was synthesized using BioScript RNase H Low reverse transcriptase ( BIO-27036 , Bioline ) or Expand reverse transcriptase ( Roche ) in 20 µl reactions . 1 µl of generated cDNA samples ( 50 ng total RNA equivalents ) were used per reaction for real-time PCR with the ABI7500 system ( Applied Biosystems ) . Specific amplification was assured by utilizing TaqMan probes and gene specific primers . Amplification was performed in 10 µl reactions with TaqMan Fast Advanced Master Mix under recommended conditions ( Applied Biosystems; #4444557 ) . The following TaqMan gene expression assays ( Applied Biosystems: #4331182 ) were used: Hs00153133_m1 ( PTGS2/COX-2 ) , Hs99999908_m1 ( GUSB ) , Hs01042796_m1 ( MMP-7 ) , Hs00233992_m1 ( MMP-13 ) , Hs02758991_g1 ( GAPDH ) ; primer-probe sets for RTA [134] , KbZIP [70] , K8 . 1 [135] . The average Ct for each individual amplification reaction was calculated from duplicate measurements by means of the instrument's software in “auto Ct” mode ( 7500 System Software v . 1 . 3 . 0 ) . Average Ct values obtained for the analysed transcripts of PTGS2/COX-2 , MMP-7 or MMP-13 were normalized by subtraction from the Ct values obtained for GUSB or GAPDH ( housekeeping reference ) . Relative mRNA expression changes were calculated according to the ΔΔCt method . For quantification of KSHV genome copies , DNA was extracted using the QIAamp DNA Blood Mini Kit ( QIAgen ) according to the manufacturer's instructions . KSHV genome copy numbers were determined in a TaqMan based qPCR directed against KSHV ORF K6 with normalization to the cellular C-reactive protein ( CRP ) as described previously [136] . Briefly , qPCR for KSHV was carried out in a total volume of 50 µl containing a ready-to-use master mix ( QuantiTect multiplex PCR kit; Qiagen ) , 0 . 5 µM concentrations of each primer , 10 µl of DNA from the sample of interest , and 0 . 4 µM FAM-labeled KSHV K6 probe . Amplification was performed in the Applied Biosystems 7500 thermal cycler and visualized with ABI 7500 software . qPCR of CRP was carried out in a total volume of 20 µl containing a ready-to-use master mix ( LightCycler FastStart DNA Master HybProbe; Roche ) , 0 . 3 mM MgCl2 , 0 . 5 µM concentrations of each primer , 0 . 2 µM FAM-labeled CRP probe , and 5 µl of DNA from the sample of interest . Amplification was performed in the LightCycler 2 . 0 Instrument and analyzed with the LightCycler software . The primers ( Sigma ) and probes ( Eurogentec ) used for the quantification of KSHV and CRP had the following sequences: KSHV K6 forward ( CGCCTA ATAGCTGCTGCTACGG ) , HHV8 K6 reverse ( TGCATCAGCTGCCTAACCCAG ) , CRP forward ( CTTGACCAGCCTCTCTCATGC ) , CRP reverse ( TGCAGTCTTAGACCCCACCC ) , K6 probe [5′- ( 6 FAM ) -CAGCCACCGCCCGTCCAAATTC-TAMRA] , and CRP probe [5′- ( 6 FAM ) -TTTGGCCAGACAGGTAAGGGCCACC-TAMRA] . Cell invasiveness was measured using Matrigel coated invasion inserts ( Growth Factor Reduced Matrigel Invasion Chamber , 8 . 0 µm; 354483 , BD Biosiences ) . HuAR2T rKSHV . 219 cells were microporated with siRNAs twenty hours prior to the induction of the lytic cycle . Twenty-four hours after the induction , the cells were starved in EBM2 supplemented with 2% FCS for twelve hours . Next day , 5×104 cells were plated in the inner chambers in 500 µl of EBM2 basal medium with 2% FCS and 750 µl EBM2 was added to the outer chamber and incubated for twenty-four hours . Before the assay , Matrigel inserts were rehydrated with 500 µl EBM2 for two hours . Cells that were able to degrade the Matrigel layer , migrated to the lower surface of the filter , and were fixed with 4% paraformaldehyde , permeabilised with 0 . 2% Triton X-100 , and nuclei were stained with DAPI ( Sigma-Aldrich ) and counted under a fluorescent microscope . Four different Matrigel chambers were used for each sample . Four random fields were counted for each chamber , and the average cell number per field in a chamber was calculated using CellProfiler2 . 0 . To quantify the number of cells in the immunofluorescence images we used the CellProfiler software [137] . All pixel intensities were rescaled to 0–1 . Using the Otsu Global thresholding method [138] in the DAPI channel , the nuclear area was defined . Clumped nuclei were distinguished based on the intensity . The threshold correction factor was set to 1 . 3 . Cell invasiveness of freshly isolated HUVECs ( <p . 2 ) was measured after infection with rKSHV . 219 at MOI 30 for four days . Three days after infection the cells were microporated with siRNAs and starved in EBM2 medium with 2% FCS . After twenty-four hours the cells were seeded onto the Matrigel , and their invasiveness was quantified as described above . 3 µm thin tissue sections were cut from formalin-fixed KS samples and stained with haematoxylin-eosin ( HE ) . KS tumour cells and non-neoplastic endothelial cells were marked immunohistochemically with anti-CD34 antibody ( Menarini Corp . ) using a 1∶50 dilution . The KSHV latent nuclear antigen ( LANA ) was marked immunohistochemically with NCL-HHV8-LNA antibody , clone 13B10 , purchased from Novocastra , using a 1∶50 dilution . MAP4K4 was stained immunohistochemically with MAP4K4 monoclonal antibody M07 , clone 4A5 , produced by Abnova Corp . and purchased from Biozol Diagnostica GmbH , applied at a 1∶300 dilution . When performing MAP4K4/LANA double-staining , 1∶20 ( LANA ) and 1∶100 ( MAP4K4 ) dilutions were applied using the BenchMark Ultra staining machine . The library of kinase inhibitors was received from Vichem Chemie Research Ltd . ( Budapest , Hungary ) as lyophilized powders , and stored at room temperature . DMSO ( cell culture grade , Applichem ) was used to dissolve the inhibitors at a stock concentration of 10 mM . After reconstitution , the inhibitors were stored at room temperature protected from light . VI18802 is a phenoxypyrimidine [93] targeting p38α . As a part of the Vichem Core Validation Library it was handled as described above . SB203580 ( #13067 , Cayman Chemical ) , SB202190 ( #EI-294-0001 , Biomol ) , SB220025 ( #559396 , Calbiochem ) , SKF86002 ( #2008 , Tocris Bioscience ) , and VX-745 ( #3915 , Tocris Bioscience ) were reconstituted and stored in working aliquots at −20°C protected from light . To target COX-2 , NS-398 ( #349254 , Calbiochem ) was prepared according to the manufacturer's recommendations and applied to the cells six hours before the induction of the lytic cycle . Application of all compounds to the cells was controlled by DMSO treatment . For evaluation of KSHV reactivation inhibition , 5×103 Vero rKSHV . 219 cells per well were plated in 96-well plates twenty-four hours before the treatment with kinase inhibitors . The inhibitors were applied one hour before the induction of the lytic cycle . Forty-eight hours later the supernatants were transferred to HEK293 cells , which were then centrifuged 30 min at 30°C and 500×g and incubated at 37°C for six hours . Medium was exchanged , and the cells were incubated at 37°C for forty-eight hours . The number of infectious particles was evaluated by the mean fluorescence intensity of GFP-positive HEK293 cells . Alternatively , Vero rKSHV . 219 or EA . hy rKSHV . 219 cells were treated with kinase inhibitors , and subsequently with induction mix to assess the expression levels of KSHV lytic proteins RTA and K8 . 1 . To assay the viability of cells after treatment with kinase inhibitors , each well of a 96-well plate received 20 µl glutaraldehyde ( 25% ) and was incubated at RT for at least 20 min . After washing with water , the plates were stained with 0 . 4% crystal violet solution in methanol for 30 min . Absorbance at 590 nm was measured spectrophotometrically with a reference to 405 nm reading . The KINOMEscan of SB220025 and VI18802 was carried out by DiscoverX as described ( www . discoverx . com ) . For evaluation of Ser/Thr kinase phosphorylation in endothelial cells , a human phospho-kinase antibody array ( ARY003B , R&D Systems ) was used according to the manufacturer's recommendations . Briefly , HuAR2T rKSHV . 219 cells were transfected with control siRNA or an siRNA pool targeting MAP4K4 twenty-four hours before the induction of the lytic cycle . Cells were lysed twenty-four after lytic cycle induction and diluted lysates were applied to , and incubated overnight with , the nitrocellulose membranes with spotted capture antibodies . The array was washed to remove unbound proteins , followed by incubation with a cocktail of biotinylated detection antibodies , and streptavidin-HRP . Chemiluminescent detection reagents ( SuperSignal West Femto Chemiluminescent Substrate , 34096 , Thermo Scientific ) were applied as recommended . The signal produced at each capture spot corresponded to the amount of phosphorylated protein bound . Statistical analysis was performed using GraphPad prism software . For the comparison of more than two groups a one-way-ANOVA with Tukey's post-test was applied after using D'Agostino-Pearson's normality test where applicable . P-values <0 . 05 were considered as significant ( * ) , <0 . 01 ( ** ) , <0 . 001 ( *** ) , and <0 . 0001 ( **** ) . P-values >0 . 05 were considered non-significant ( ns ) . Error bars were calculated from means ±SD . The qPCR data are shown as means ±SEM , where one replicate is shown as a representative . Akt ( P31749 ) , b-Raf ( P15056 ) , CD34 ( P28906 ) , CDC2L1 ( A4VCI5 ) , c-Jun ( P05412 ) , COX-1 ( P23219 ) , COX-2 ( P35354 ) , CSNK1A1L ( Q8N752 ) , CSNK1D ( P48730 ) , CSNK1E ( P49674 ) , ERK1 ( P27361 ) , IFN-α ( P01562 ) , IL-1β ( P01584 ) , IL-2 ( P60568 ) , IL-6 ( P05231 ) , JNK1 ( P45983 ) , JNK3 ( P53779 ) , KSHV K15 ( Q9QR69 ) , KSHV K8 . 1 ( D2XQF0 ) , KSHV KbZIP ( E5LBX3 ) , KSHV LANA ( J9QT20 ) , KSHV ORF45 ( F5HDE4 ) , KSHV RTA ( F5HCV3 ) , KSHV v-Cyclin ( Q77Q36 ) , KSHV v-FLIP ( Q76RF1 ) , KSHV vGPCR ( Q98146 ) , KSHV vIL-6 ( Q98823 ) , MAP4K4 ( O95819 ) , MEK1 ( Q02750 ) , MINK ( Q8N4C8 ) , MK2 ( P49137 ) , MMP-1 ( P03956 ) , MMP-13 ( P45452 ) , MMP-19 ( Q99542 ) , MMP-2 ( P08253 ) , MMP-3 ( P08254 ) , MMP-7 ( P09237 ) , MMP-9 ( P14780 ) , NF-κB ( Q04206 ) , Notch ( P46531 ) , p38α ( Q16539 ) , PI3K ( P42336 ) , Pim-1 ( P11309 ) , Pim-3 ( Q86V86 ) , PKA ( P17612 ) , PKCδ ( Q05655 ) , RBP-Jκ ( Q06330 ) , STK36 ( Q9NRP7 ) , TLR7/8 ( D1CS68 ) , TNF-α ( P01375 ) , TNFR α ( P19438 ) , TNIK ( Q9UKE5 ) .
|
Kaposi's sarcoma ( KS ) is a tumour caused by Kaposi's sarcoma herpesvirus ( KSHV ) and dysregulated inflammation . Both factors contribute to the high angiogenicity and invasiveness of KS . Various cellular kinases have been reported to regulate the KSHV latent-lytic switch and thereby virus pathogenicity . In this study , we have identified a STE20 kinase family member – MAP4K4 – as a modulator of KSHV lytic cycle and invasive phenotype of KSHV-infected endothelial cells . Moreover , we were able to link MAP4K4 to a known mediator of inflammation and invasiveness , cyclooxygenase-2 , which also contributes to KSHV lytic replication . Finally , we could show that MAP4K4 is highly expressed in KS lesions , suggesting an important role for this kinase in tumour development and invasion .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2013
|
The Inflammatory Kinase MAP4K4 Promotes Reactivation of Kaposi's Sarcoma Herpesvirus and Enhances the Invasiveness of Infected Endothelial Cells
|
Recent results indicate that genome-wide association studies ( GWAS ) have the potential to explain much of the heritability of common complex phenotypes , but methods are lacking to reliably identify the remaining associated single nucleotide polymorphisms ( SNPs ) . We applied stratified False Discovery Rate ( sFDR ) methods to leverage genic enrichment in GWAS summary statistics data to uncover new loci likely to replicate in independent samples . Specifically , we use linkage disequilibrium-weighted annotations for each SNP in combination with nominal p-values to estimate the True Discovery Rate ( TDR = 1−FDR ) for strata determined by different genic categories . We show a consistent pattern of enrichment of polygenic effects in specific annotation categories across diverse phenotypes , with the greatest enrichment for SNPs tagging regulatory and coding genic elements , little enrichment in introns , and negative enrichment for intergenic SNPs . Stratified enrichment directly leads to increased TDR for a given p-value , mirrored by increased replication rates in independent samples . We show this in independent Crohn's disease GWAS , where we find a hundredfold variation in replication rate across genic categories . Applying a well-established sFDR methodology we demonstrate the utility of stratification for improving power of GWAS in complex phenotypes , with increased rejection rates from 20% in height to 300% in schizophrenia with traditional FDR and sFDR both fixed at 0 . 05 . Our analyses demonstrate an inherent stratification among GWAS SNPs with important conceptual implications that can be leveraged by statistical methods to improve the discovery of loci .
Complex traits are generally influenced by many genes with small individual effects [1] . This ‘polygenic’ architecture has been difficult to characterize . While Genome-wide association studies ( GWAS ) [2] have successfully identified thousands of trait-associated single nucleotide polymorphisms ( SNPs ) [3] , even when considered in aggregate , these SNPs explain small portions of the trait heritability [4] . Recent results indicate that GWAS have the potential to explain much of the heritability of common complex phenotypes [5] , [6] , and more SNPs are likely to be identified in larger samples [7] . However , there are few methods available for identifying more of the SNPs likely to be associated with phenotypes without increasing the sample size , as recognized by recent European and US calls for new statistical genetics methods . The crucial issue is that for most complex traits a large number of SNPs have too small an effect to pass standard GWAS significance thresholds given current sample sizes . We present results suggesting new analytical approaches for GWAS will uncover more of the polygenic effects in complex disorders and traits . We hypothesize that all SNPs in a GWAS are not exchangeable , but come from pre-determinable categories with different distributions of effects . This implies that some categories of SNPs are enriched , i . e . are more likely to be associated with a phenotype than others . This information can be used to calculate the category-specific True Discovery Rate ( TDR ) , or the expected proportion of correctly rejected null hypotheses [8] . SNPs from enriched SNP categories will have an increased TDR for a given effect size , or equivalently , for a given nominal p-value . Stratified False Discovery Rate ( sFDR ) methods [9] provide an established framework for demonstrating the utility of using enriched genic categories to increase power to discover SNPs likely to replicate in independent samples . Previous work has applied sFDR and related methods to GWAS data stratified by candidate regions determined through prior linkage analysis and/or candidate gene studies [10]–[12] and specific biological pathways related to disease etiology [13] . Others have considered stratification by genome annotations in linkage analysis [14] and Bayesian association analyses [15] , demonstrating the utility of this approach for improving power and FDR based discovery where reliable , pre-determinable strata exist . It has been suggested that variation in and around genes harbors more polygenic effects [6] , [16] . However , the particular gene elements ( i . e . , intron , exon , UTRs ) containing these variants and the distribution of effect sizes in GWAS have been left to extrapolation and speculation . Further , SNPs in and around genes have been shown to explain more variation [6] and replicate at higher rates [16] than intergenic SNPs . These studies , however , did not parse genic regions down to specific genic elements . We here hypothesize that SNPs in regulatory and coding elements of protein coding genes will show an enrichment of polygenic effects relative to intronic and intergenic SNPs which will be reflected in an increased estimated TDR and empirically confirmed through improved replication rate across independent samples . The association signal of a SNP tested in GWAS is a surrogate for , or ‘tags , ’ the potential effects of many other variants . Thus , any of a number of ‘tagged’ variants could underlie the observed association signal . Focusing on the tag SNPs only , without systematically capturing the underlying causal variants within a ‘tagged’ linkage block , limits the functional inferences that can be drawn from GWAS . By incorporating the correlation between SNPs ( linkage disequilibrium; LD ) we expect a stronger and more consistent differentiation of enrichment among genic annotation categories . In the current study , we use an LD-weighted scoring algorithm that allows quantification of the properties of multi-locus LD structure implicitly captured by each tag SNP to our enrichment analysis . These categories can be leveraged to create strata for established sFDR approaches . We employ a model free strategy to identify enriched strata among phenotypes based on GWAS summary statistics . We first calculate the relative enrichment in different genic elements , using the category-specific empirical cumulative distribution function ( cdf ) of the nominal p-values after controlling for estimated genomic inflation . For each nominal p-value threshold an estimate of the category-specific TDR = 1−FDR is obtained from these empirical cdfs . This analysis is implemented on summary p-values from ten published GWAS meta-analyses studying 14 phenotypes . We then use the sub-study GWAS in Crohn's disease to test if the estimated increased TDR translates to improved replication rates , showing that for a given replication rate the nominal p-value threshold is 100 times larger for the most enriched genic category compared to the intergenic category . Finally , using an established sFDR framework we demonstrate the utility of leveraging enriched categories for improving power to detect SNPs likely to replicate , i . e . , to reject more null hypotheses for a fixed FDR .
Under multiple testing paradigms such as GWAS , quantitative estimates of likely true associations can be estimated from the distributions of summary statistics [17] , [18] . A common method for visualizing the enrichment of statistical association relative to that expected under the global null hypothesis is through Q-Q plots of the nominal p-values resulting from GWAS . Under the global null hypothesis the theoretical distribution is uniform on the interval [0 , 1] . Thus , the usual Q-Q curve has as the y-coordinate the nominal p-value , denoted by “p” , and the x-coordinate the value of the empirical cdf at p , which we denote by “q” . As is common in GWAS , we instead plot −log10 p against the −log10 q to emphasize tail probabilities of the theoretical and empirical distributions . In such plots , enrichment results in a leftward shift in the Q-Q curve , corresponding to a larger fraction of SNPs with nominal −log10 p-value greater than or equal to a given threshold ( see Material and Methods ) . The stratified Q-Q plot for height ( Figure 1 ) shows a clear variation in enrichment across genic annotation categories . The separation between the curves for different categories is enhanced when using LD-weighted genic annotation categories in comparison to non LD-weighted positional categories ( Figure S3 ) . The parallel shape of these curves is likely caused by the significant but imperfect correlation among categories due to the non-exclusive nature of the annotation scoring ( Figure S2 ) . An earlier departure from the null line ( leftward shift ) suggests a greater proportion of true associations , for a given nominal p-value . The divergence of the curves for different categories implies that the proportion of non-null effects varies considerably among annotation categories of genic elements . For example , the proportion of SNPs in the 5′UTR category reaching a significance level of −log10 ( p ) >10 is roughly 10 times greater than for all SNPs and 50–100 times greater than for intergenic SNPs . Recently Yang et al [19] demonstrated that an abundance of low p-values beyond what is expected under null hypotheses in GWAS , but not necessarily reaching stringent multiple comparison thresholds , often attributed to ‘spurious inflation , ’ is also consistent with an enrichment of true ‘polygenic’ effects [19] . The prevalence of enrichment below the established genome-wide significance threshold of p<5×10−8 ( −log10 ( p ) >7 . 3; ) in height ( Figure 2A ) is consistent with their hypotheses and strongly suggests that current GWAS do not capture all of the additive ‘tagged variance’ in this phenotype . Importantly , this enrichment varies across genic annotation categories . The enrichment patterns among annotation categories are consistent across phenotypes , including schizophrenia ( SCZ ) and tobacco smoking ( cigarettes per day , CPD; Figure 2B–2C ) . The stratified Q-Q plots for height , SCZ and CPD each demonstrate the largest enrichment for tag SNPs in LD with 5′UTR , and exonic variation , showing nearly tenfold increases in terms of the proportion of p-values expected below a given threshold under the null hypothesis . SNPs that tag intergenic regions show nearly tenfold depletions in comparison to all tag SNPs , although not when compared to the expected null . SNPs tagging intronic variation show minimal enrichment over all tag SNPs , despite making up the largest proportion of genic SNPs ( Table S3 ) . The pattern is consistent for all phenotypes considered ( data not shown ) . Given the log scale of the Q-Q plots , 90% of SNPs fall between 0 and 1 and 99% fall between 0 and 2 on the horizontal axis , and thus it is clear that a majority of enriched SNPs have p-values that do not reach genome-wide significance . We computed significance values for the curves for each annotation category relative to those for intergenic SNPs , using a two-sample Kolmogorov-Smirnov Test . The enrichment for height was highly significant for all categories when compared with the intergenic category , with all p-values less than 2 . 2×10−16 . Nearly every genic category was also significantly enriched for each other phenotype ( Table S5 ) . While the pattern of enrichment is consistent , the shape of the curves varies across phenotypes . In particular , the point at which the curves deviate from the expected null line occurs earliest for height , followed by SCZ , and finally CPD ( Figure 2A–2C ) , consistent with different proportions of SNPs that are likely associated with each trait ( i . e . , different levels of ‘polygenicity’ ) . These findings are consistent with results obtained using an established mixture-modeling framework [17] ( Text S1 and Figures S8 , S9 , S10 , S17 , S18 , S19 ) . The relative absence of enrichment in intergenic SNPs as we have defined them , suggests minimal inflation due to polygenic effects and a more robust estimate of the global null . This fact can be exploited for better estimation of variance inflation due to stratification [20] that is minimally confounded by true polygenic effects [19] . We confined the estimation of the genomic inflation factor [20] , λGC , to only intergenic SNPs ( Table S4 ) and adjusted summary statistics for all phenotypes according to this “intergenic inflation control” procedure . The stratified Q-Q plots for height with and without intergenic inflation control are shown in Figure S4 . Since specific tag SNP categories are significantly more likely to be associated with common phenotypes , while intergenic ones are less likely , all tag SNPs should not be treated as exchangeable . Variation in enrichment across diverse genic categories is expected to be associated with corresponding variation in TDR for a given nominal p-value threshold . A conservative estimate of the TDR for each nominal p-value is equivalent to 1− ( p/q ) as plotted on the Q-Q plots ( see Online Methods ) . This relationship is shown for height , SCZ and CPD ( Figure 2D–2E ) . Similar category-specific TDR plots were calculated for each of the 14 phenotypes ( data not shown ) . For a given TDR the corresponding estimated nominal p-value threshold varies with a factor of 100 from the most enriched genic category to the intergenic category , and the pattern is consistent across phenotypes . Since TDR is theoretically related to predicted replication rate , it is expected that for a given p-value threshold the replication rate will be higher for SNPs in genic categories with high TDR . The high estimates of TDR at significance levels below genome-wide significance is consistent with recent work in Schizophrenia that demonstrates a high proportion of likely true associations at reduced thresholds , but without the needed power to reach genome-wide significance [21] . While the TDR provides a quantification of enrichment for a given nominal p-value threshold ( equivalently , SNP z-score threshold ) , we also provide a single number quantification of enrichment for each LD-weighted annotation category within each phenotype , computed as the sample mean ( z2 ) −1 . The sample mean , taken over all SNPs in a given category , provides an estimate of the variance due to null and non-null SNPs; by subtracting one we obtain a conservative estimate of the variance in effect sizes attributable to non-null SNPs alone . Both TDR and mean ( z2 ) −1 are justified based on a standard mixture model formulation ( see Text S1 ) . These enrichment scores , normalized by the maximum value across categories within each phenotype , are presented in Figure 3 . The 5′UTR annotation category was the most enriched category across all fourteen phenotypes ( Table S6 ) . Additionally , the exon category is consistently more enriched than the intron category . Categories where each SNP tags more reference SNPs on average or represents a larger total amount of LD could spuriously appear enriched . We do note categorical differences in the number of SNPs and total summed LD captured by each SNP ( Tables S7 and S8 ) but multiple regression analyses show the effect of these variables is negligible and independent categorical effects persist ( Table S10 ) despite the significant correlation among categories ( Figure S2 ) . Likewise , systematic deviations in minor allele frequency ( MAF ) across categories could bias annotation category effects as MAF acts multiplicatively with effect size to explain variance . We found minimal categorical stratification for MAF that is inconsistent with this effect driving our enrichment findings ( Table S9 and Figure S6 ) . To further address the possibility that some of the differential enrichment of categories could be due to category-specific genomic inflation from the above factors , we performed null-GWAS simulations based on genotypes from the 1000 Genome Project . The results suggest that such effects are non-existent or negligible ( Table S11 ) . To further address the possibility that the observed pattern of differential enrichment results from spurious ( i . e . , non-generalizable ) associations due to category-specific confounding effects or statistical modeling errors , we also studied the empirical replication rate across independent sub-studies for one phenotype ( CD ) where the required sub-study summary statistics were available . Figure 4A shows the estimated TDR curves for different annotation categories in CD , with a similar pattern as that described for in height , SCZ and CPD , above . TDR is an estimate of the expected replication rate for a sufficiently large replication sample . We hypothesized that strata with higher TDR for a given nominal p-value would also show higher empirical replication rate . Figure 4B shows the empirical cumulative replication rate plots as a function of nominal p-value for the same categories as for the stratified TDR plot in Figure 4A . Consistent with the category-specific TDR pattern , we found that the nominal p-value corresponding to a wide range of replication rates was 100 times higher for intergenic relative to the most enriched genic category ( 5′UTR ) . Similarly , SNPs from genic annotation categories showing the greatest enrichments replicated at higher rates , up to five times higher than intergenic for 5′UTR SNPs , independent of p-value thresholds . The increase in replication rate was found to be greatest for SNPs that do not meet genome-wide significance , suggesting that adjusting p-value thresholds according to the estimated category-specific TDR could greatly improve the discovery of replicating SNP associations . In order to demonstrate the utility of the enriched category information for improved discovery , we leveraged an established method for computing stratified False Discovery Rates [9] . The sFDR method improves the power of traditional methods for FDR control [22] by taking advantage of pre-defined , differentially enriched strata among multiple hypothesis testing p-values . Here , we define an increase in power from using stratified ( vs . unstratified ) methods as a decreased Non-Discovery Rate ( NDR ) for a given level of FDR control α , where NDR is the proportion of false negatives among all non-null tests [23] . Specifically , the ratio of 1-NDR from stratified FDR control to 1-NDR from unstratified FDR control captures the relative power of the two approaches . This ratio can be shown to be equivalent to the ratio of the number of SNPs rejected by sFDR to the number rejected by unstratified FDR for a common level α . For each phenotype we divided the SNPs into independent strata according to its predicted tagged variance ( z2 ) . Tagged variance was predicted using on a linear model with regression weights for each annotation category trained using the height GWAS summary statistics . The enrichment of these strata is presented in Figure S11 . In Figure 5 ( and Table S12 ) we show an increase in the number of discovered SNPs . For example , for α = . 05 the increased proportion of declared non-null SNPs using sFDR ranges from 20% in height to 300% in schizophrenia . Leveraging our genic annotation categories in the sFDR framework provides one possible avenue for improving the output of likely non-null SNPs in GWAS by taking advantage of the non-exchangeability of SNPs demonstrated by our enrichment analyses . Other formulations of strata and continued investigations into enrichment are likely to further improve the power of this approach .
Our results show a significant and consistent pattern of enrichment among genic elements , particularly the 5′UTR , exon and 3′UTR categories , for association with diverse complex traits and disorders . Intergenic SNPs were depleted more than tenfold . This has important analytical and conceptual implications . The results suggest that all tag SNPs should not be treated as exchangeable , but rather functional annotations of the underlying tagged SNPs can be leveraged in SNP discovery . Moreover , the results point to a common functional nature of the polygenetic architecture across diverse complex phenotypes . GWAS have traditionally treated all SNPs as exchangeable , implicitly assigning all SNPs equal a priori probability of association . The current findings suggest that this assumption of exchangeability is not valid , and that the traditional statistical approaches to GWAS are highly suboptimal . Sun et al [13] have laid the groundwork for incorporating this non-exchangeability into Hypothesis Driven Genome-Wide Association Studies ( HD-GWAS ) , and further applications and development of such methods is likely to prove fruitful . To illustrate the utility of these approaches we used our LD-weighted tag SNP annotations , combined with an established method for computing stratified False Discovery Rates , to demonstrate improved discovery of SNPs in GWAS under this testing framework . Annotation categories were chosen based on previous literature suggesting that SNPs in and near genes are likely to harbor many true polygenic effects . We provide a proof of principle , using intuitive categories , that a priori information about SNPs , irrespective of phenotype , can improve discovery of likely non-null SNPs . Given the wealth of information available for SNP variants , it is likely that other annotation schemes will potentially yield even greater enrichment and further increase the gains of our basic approach . We expect this approach will be of particular importance in polygenic complex phenotypes . Only a small fraction of the heritability is explained by currently discovered variants but converging evidence suggests much more remains buried in GWAS ( i . e . , traits with a large “missing heritability” ) [4] for these traits . Moreover , the non-exchangeability of SNPs based on LD-weighted genic categories has important implications for the generalizability of estimated SNP effect sizes . In particular , SNPs in highly enriched categories will have effect size estimates that replicate strongly in independent samples , whereas SNPs in impoverished categories will have effect sizes that replicate weakly in independent samples [17] ( Figure S10 ) . Identifying SNPs with generalizable effects is crucial to improving the predictive power of polygenic risk scores that combine SNP effects to predict variation in complex traits and diseases in new samples [24] . Properly assessing the generalization performance of SNP effect sizes will be of high importance for personalized medicine based on polygenic risk scores . While knowing which SNP categories are enriched for true associations can guide gene discovery , knowing which SNPs are unlikely to have an effect is also important and can guide control of spurious inflation through improved genomic inflation correction [19] , [20] . We show how the genic enrichment pattern can be used for genomic inflation control in GWAS . By estimating genomic control from intergenic tag SNPs , we can minimize the contamination of inflation estimates from true polygenic associations . While emerging studies have suggested that polygenic effects are detectable in GWAS data [5] , [6] , particularly in and around genes [6] , [16] , and the presence of these effects is consistent with a skewed distribution of p-values [17] , [25] that resembles spurious inflation [19] , differential confounds among our categories could persist . We provided a null GWAS ( Table S11 ) and found no indications of spurious enrichment due to differential MAF and LD structure among our categories . Further , if our findings were due to spurious category-specific inflation , including differential population stratification , one would not expect a mirroring increase in replication across independent samples ( Figure 4 ) , except under the extreme condition where the population structure of the discovery sample was mirrored in the replication sample . In addition , the variable shape of the enrichment , deviating at different points along the expected line for different phenotypes , is inconsistent with spurious variance inflation . It is also of importance to note that the presence of spurious category-specific inflation would imply that any GWAS association within an enriched category ( i . e . , tagging an exonic or UTR SNP ) should be considered less reliable . Our findings are consistent with the presence of true polygenic effects , however , we cannot entirely rule out contributions of potential confounding effects or alternate hypotheses . It is highly implausible , however , that they would explain away both the described enrichment and increased replication . Conceptually , our results support findings that aspects of the genetic architecture are consistent across phenotypes [26] , and previous suggestions from both model organisms [27] and humans [6] , [16] that polygenic contributions are greater from variants in and around genes . Our findings agree with emerging trends in model organisms [26] and post-hoc GWAS analyses [28]–[30] suggesting that quantitative traits are affected by a large but quantifiable number of polymorphisms , inconsistent with ‘infinitesimal’ models [31] , but notably polygenic . We also show evidence that tagged variance is proportional to genotypic variance ( Figures S6 and S7 ) , which supports the notion that common variation explains an important part of the variability in common diseases and traits [32] . Our findings also suggest that regulatory genic elements may be particularly enriched for polygenic effects . This is in line with the most strongly associated SNPs from GWAS , which mainly tag regulatory genic elements [3] . The 5′UTR , specifically , is important for the regulation of gene expression [33] making it a compelling candidate for playing a causal role in complex trait variation [1] . Also , 5′UTRs are less conserved evolutionarily than coding regions [34] , despite their noted functionality , pointing to a potential source for regulatory variation thought to drive evolutionary differences among primates [35] and other species . We also found a stronger enrichment of SNPs tagging exons compared to introns . However , because we are considering tag SNPs we can only speculate about the functional consequences of the underlying causal variants . Exome sequencing studies have identified causal variants for Mendelian disorders by leveraging hypotheses about the genetic architecture of these traits and thus focusing on protein changing variants [36] . While methods are continually improving for predicting the functional consequences of coding changes , predicting regulatory function has remained a challenge . Future target capture methods , deep sequencing efforts and custom SNP array designs , as well as functional prediction efforts in complex traits may improve power and utility by adding focus to regulatory elements , in particular 5′UTRs . As other potential annotation categories , such as transcription factor binding sites , methylation targets , conservation/selection , and gene expression patterns , become better characterized the current analyses could be extended to include these .
Fourteen phenotypes , body mass index ( BMI ) [37] , height , waist to hip ratio [38] ( WHR ) , Crohn's disease [39] ( CD ) , ulcerative colitis [40] ( UC ) , schizophrenia [41] ( SCZ ) , bipolar disorder [42] ( BD ) , smoking behavior as measured by cigarettes per day [43] ( CPD ) , systolic and diastolic blood pressure [44] ( SBP , DBP ) , and plasma lipids [45] ( triglycerides , TG , total cholesterol , TC , high density lipoprotein , HDL , low density lipoprotein , LDL ) , were considered . Genome-wide association study ( GWAS ) results were obtained as summary statistics ( p-values or z-scores ) from public access websites ( BMI , Height , WHR , TC , TG , HDL , LDL ) , published supplementary material ( SBP , DBP ) , or through collaborations with investigators ( CD , UC , SCZ , BD ) . For CD , pre-meta-analysis , sub-study specific p-values and effect sizes ( z-scores ) were obtained from the study principal investigators . In total these studies considered more than 1 . 3 million phenotypic observations , but considerable sample overlap makes the number of unique individuals much less . For details , see Text S1 and Table S1 . The summary statistics from the respective GWAS meta-analyses , derived according to best practices , were used as-is . No further processing was performed , with the exception of intergenic inflation control ( described below ) . Results from SNPs with reference SNP ( rs ) numbers that did not map to our 1000 genomes project ( 1KGP ) reference panel were excluded . Bi-allelic SNP genotypes from the European reference sample provided by the November 2010 release of Phase 1 of the 1KGP were obtained in pre-processed form from http://www . sph . umich . edu/csg/abecasis/MACH/download/ . Using Plink version 1 . 07 [46] , [47] 1KGP SNPs with a minor allele frequency less than 1% , missing in more than 5% of individuals and/or violating Hardy-Weinberg equilibrium ( p<1×10−6 ) were excluded from the reference panel . Individuals missing more than 10% of genotypes were excluded . Each remaining 1KGP SNP was assigned a single , mutually exclusive genic annotation category based on its genomic position ( hg19 ) . Genic annotation categories were: 1 ) 10 , 000 to 1 , 001 base pairs upstream ( 10 k Up ) ; 2 ) 1 , 000 to 1 base pair upstream ( 1 k Up ) ; 3 ) 5′ untranslated region ( 5′UTR ) ; 4 ) exon; 5 ) intron; 6 ) 3′ untranslated region ( 3′UTR ) ; 7 ) 1 to 1 , 000 base pairs downstream ( 1 k Down ) ; 8 ) 1 , 001 to 10 , 000 base pairs downstream ( 10 k Down ) , all with reference to protein coding genes only . Annotations were assigned based on the first gene transcript listed in the UCSC known genes database [48] . In total 9 , 078 , 405 1KGP SNPs were assigned positional categories . All positional categories were scored 0 or 1 . For further details see Text S1 . For each GWAS tag SNP a pairwise correlation coefficient approximation to LD ( r2 ) was calculated for all 1KGP SNPs within 1 , 000 , 000 base pairs ( 1 Mb ) of the tag SNP using Plink version 1 . 07 [46] , [47] . LD scores were thresholded providing continuous valued estimates from 0 . 2 to 1 . 0; r2 values <0 . 2 were set to 0 and each SNP was assigned an r2 value of 1 . 0 with itself . LD-weighted annotation scores were computed as the sum of r2 LD between the tag SNP and all 1KGP SNPs positioned in a particular category . Each tag SNP was assigned to every LD-weighted annotation category for which its annotation score was greater than or equal to 1 . 0 . The resulting LD-weighted annotation categories were not mutually exclusive such that each GWAS tag SNP could be annotated with multiple categories . Summary statistics describing the distribution of scores in each category for the 2 , 558 , 411 SNPs , representing the union of all GWAS considered , are provided in Tables S2 and S3 . Figure S1 provides a schematic of our scoring algorithm . All analyses were repeated using a second set of LD thresholding parameters and found to be robust ( Text S1 and Figures S13 , S14 , S15 , S16 ) . Intergenic SNPs were determined after LD-weighted scoring and defined as having LD-weighted annotations scores for each of the eight categories equal to zero . In addition they were defined to not be in LD with any SNPs in the 1KGP reference panel located within 100 , 000 base pairs of a protein coding gene , within a noncoding RNA , within a transcription factor binding site nor within a microRNA binding site . SNPs labeled intergenic were defined to be a specific collection of non-genic SNPs chosen to not represent any functional elements within the genome ( including through LD ) . Because of how they are defined these SNPs are hypothesized to represent a collection of null associations . Other non-genic categories ( 1 k up , 10 k up , 1 k down and 10 k down ) were included in our analyses to ensure SNPs not too far away from genes , but not within protein coding genes , were represented by non-genic categories and enrichment due to these SNPs was not solely attributed to LD with genic categories . Q-Q plots compare two probability distributions . For each phenotype , for all SNPs and for each categorical subset , −log10 nominal p-values were plotted against −log10 empirical p-values . Leftward deflections of the observed distribution from the projected null line reflect increased tail probabilities in the distribution of test statistics ( z-scores ) and consequently an over-abundance of low p-values compared to that expected by chance . We qualitatively refer to this deflection as “enrichment” ( Figure 1 and Figure 2 , Figure S3 ) . We estimated the significance of the annotation enrichment using two sample Kolmogorov-Smirnov ( KS ) Tests to compare the distribution of test statistics in each genic annotation category to the distribution of the intergenic category , for each phenotype . SNPs were pruned randomly to approximate independence ( r2<0 . 2 ) ten times and Table S5 reports the p-value corresponding to the median KS statistics from the ten comparisons . The empirical null distribution in GWAS is affected by global variance inflation due to factors including population stratification and cryptic relatedness [20] and deflation due to over-correction of test statistics for polygenic traits by standard genomic control methods [19] . We applied a control method leveraging only intergenic SNPs that are likely depleted for true associations . All p-values were converted into z-scores and , for each phenotype , the genomic inflation factor [20] , λGC , was estimated for intergenic SNPs . All test statistics were divided by λ GC . The inflation factor λGC was computed as the median z-score squared divided by the expected median of a chi-square distribution with one degree of freedom or all phenotypes except CPD , where the . 95 quantile was used in place of the median . For correction statistics see Table S4 . For each phenotype , enrichment was measured as the mean ( z-score2−1 ) for each category and normalized by the largest value per phenotype . The mean ( z-score2−1 ) is a conservative estimate of the variance attributable to non-null SNPs , given a standard normal null distribution and a non-null distribution symmetric around zero ( see Text S1 ) . Enrichment seen in the conditional Q-Q plots can be directly interpreted in terms of the FDR . Specifically , for a given p-value cutoff , the Bayes FDR [17] is defined as ( 1 ) where π0 is the proportion of null SNPs , F0 is the null cdf , and F is the cdf of all SNPs , both null and non-null . Under the null hypothesis , F0 is the cdf of the uniform distribution on the unit interval [0 , 1] , so that Eq . [1] reduces to ( 2 ) The cdf F can be estimated by the empirical cdf q = Np/N/cdf F p is the number of SNPs with p-values less than or equal to p , and N is the total number of SNPs . Replacing F by q and replacing π0 with unity in Eq . [2] , we get ( 3 ) This is upwardly biased , and hence p/q is conservative estimate of the FDR , and 1−p/q is a conservative estimate of the Bayes TDR [17] . If π0 is close to one , as is likely true for most GWAS , the increase in bias from setting π0 to one in Eq . [3] is minimal . Referring to the formulation of the Q-Q plots , we see that FDR ( p ) is equivalent to the nominal p-value under the null hypothesis divided by the empirical quantile of the p-values . Given the −log10 transformation applied to the Q-Q plots , we can easily read off ( 4 ) demonstrating that the ( conservatively ) estimated FDR is directly related to the horizontal shift of the curves in the stratified Q-Q plots from the expected line x = y , with a larger shift corresponding to a smaller FDR . For the TDR plots in Figure 2 , we estimated the TDR for each genic category according to Eq . [4] . Eq . [3] is the Empirical Bayes point estimate of the Bayes FDR given in Efron ( 2010 ) . Using Eq . [3] to control FDR ( i . e . , the expected proportion of falsely rejected null hypotheses ) [22] is closely related to the “fixed rejection region” approach of Storey [49] , [50] . Specifically , Storey [49] showed , for a given FDR α , rejecting all null hypotheses such that p/q<α is equivalent to the Benjamini-Hochberg procedure and provides asymptotic control of the FDR to α if the true null p-values are independent and uniformly distributed . Storey [49] also noted that asymptotic control is preserved under positive blockwise dependence , whereas Schwartzman and Lin [51] showed that Eq . [3] is a consistent estimator of FDR for asymptotically sparse dependence ( i . e . the proportion of correlated pairs of p-values goes to zero as the number of hypothesis tests becomes large ) . Sparse dependence is a good description of the dependence present in GWAS data; for example , based on a threshold of r2> . 05 within 1 , 000 , 000 basepairs , we estimate the ratio of correlated pairs ( r2> . 05 ) to total pairs of p-values at 0 . 000128 . For each of eight sub-studies contributing to the final meta-analysis in the CD report we independently adjusted z-scores using intergenic inflation control . For each of 70 ( 8 choose 4 ) possible combinations of four-study discovery and four-study replication sets , we calculated the four-study combined discovery z-score and four-study combined replication z-score for each SNP as the average z-score across the four studies , multiplied by two ( the square root of the number of studies ) . For discovery samples the z-scores were converted to two-tailed p-values , while replication samples were converted to one-tailed p-values preserving the direction of effect in the discovery sample . For each of the 70 discovery-replication pairs cumulative rates of replication were calculated over 1000 equally-spaced bins spanning the range of negative log10 ( p-values ) observed in the discovery samples . The cumulative replication rate for any bin was calculated as the proportion of SNPs with a −log10 ( discovery p-value ) greater than the lower bound of the bin with a replication p-value< . 05 . Cumulative replication rates were calculated independently for each of the eight genic annotation categories as well as intergenic SNPs and all SNPs . For each category , the cumulative replication rate for each bin was averaged across the 70 discovery-replication pairs and the results are reported in Figure 4 . The vertical intercept is the overall replication rate . A multiple linear regression was used to predict the tagged variance ( z2 ) for each SNP in the height GWAS from the unthresholded LD-weighted annotation scores . Using the category weights determined from this ‘training’ regression on the height GWAS , the tagged variance for each SNP was predicted from its annotation vector for each other phenotype . For each phenotype , SNPs were grouped into strata according to the rank of this predicted tagged variance . Enrichment for each stratum was demonstrated using Q-Q plots as described above ( Figure S11 ) . We note that for Figure 5A the height data serves as both our ‘test’ ( for creating strata ) and ‘training’ ( for detecting enrichment ) data , but for each other GWAS the training and test data is independent . Sun et al [9] described a stratified False Discovery Rate ( sFDR ) procedure which can result in improved statistical power over traditional FDR methods [22] when a collection of statistical tests can be grouped into disjoint strata with different levels of enrichment . In order to demonstrate the utility of using genic annotation categories in combination with sFDR for increasing power , we computed the number of SNPs deemed significant at a given FDR threshold using both traditional [22] and stratified FDR , where the strata were determined by the predicted tagged variance for each SNP based on regression weights determined from the height GWAS summary statistics ( Figure 5 ) . The increase in rejections for a common threshold α when using sFDR is equivalent to increased power demonstrated by a ratio of one minus Non-Discovery Rates ( 1-NDRs ) for sFDR to FDR greater than 1 [23] . We also computed the average proportion of SNPs above a given rank ( i . e . top 1000 ) that replicated based on unadjusted and strata adjusted ranks ( determined from the sFDR procedure ) across the 70 permutations of four study discovery and four study replication samples possible in the eight study CD meta-analysis GWAS ( Figure S12 ) . These results demonstrate that for a given threshold , SNPs ranked via genic category-informed sFDR replicate in higher numbers than SNPs ranked via traditional FDR .
|
Modern genome-wide association studies ( GWAS ) have failed to identify large portions of the genetic basis of common , complex traits . Recent work suggested this could be because many genetic variants , each with individually small effects , compose their genetic architecture , limiting the power of GWAS . Moreover , these variants appear more abundantly in and near genes . Using genome annotations , summary statistics from several of the largest GWAS , and established statistical methods for quantifying distributions of test statistics , we show a consistency across studies . Namely , we show that , across all assessed traits , the test statistics resulting from SNPs that are related to the 5′ UTR of genes show the largest abundance of associations , while SNPs related to exons and the 3′UTR are also enriched . SNPs related to introns are only moderately enriched , and intergenic SNPs show a depletion of associations relative to the average SNP . This enrichment corresponds directly to increased replication across independent samples and can be incorporated a priori into statistical methods to improve discovery and prediction . Our results contribute to on-going debates about the functional nature of the genetic architecture of complex traits and point to avenues for leveraging existing GWAS data for discovery in future GWA and sequencing studies .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genome-wide",
"association",
"studies",
"genome",
"analysis",
"tools",
"functional",
"genomics",
"mathematics",
"genome",
"scans",
"genetic",
"association",
"studies",
"statistics",
"genetics",
"statistical",
"methods",
"biology",
"genomics",
"biostatistics",
"computational",
"biology",
"human",
"genetics"
] |
2013
|
All SNPs Are Not Created Equal: Genome-Wide Association Studies Reveal a Consistent Pattern of Enrichment among Functionally Annotated SNPs
|
Cerebrovascular dysfunction plays a key role in the pathogenesis of cerebral malaria . In experimental cerebral malaria ( ECM ) induced by Plasmodium berghei ANKA , cerebrovascular dysfunction characterized by vascular constriction , occlusion and damage results in impaired perfusion and reduced cerebral blood flow and oxygenation , and has been linked to low nitric oxide ( NO ) bioavailability . Here , we directly assessed cerebrovascular function in ECM using a novel cranial window method for intravital microscopy of the pial microcirculation and probed the role of NOS isoforms and phosphorylation patterns in the impaired vascular responses . We show that pial arteriolar responses to endothelial NOS ( eNOS ) and neuronal NOS ( nNOS ) agonists ( Acetylcholine ( ACh ) and N-Methyl-D-Aspartate ( NMDA ) ) were blunted in mice with ECM , and could be partially recovered by exogenous supplementation of tetrahydrobiopterin ( BH4 ) . Pial arterioles in non-ECM mice infected by Plasmodium berghei NK65 remained relatively responsive to the agonists and were not significantly affected by BH4 treatment . These findings , together with the observed blunting of NO production upon stimulation by the agonists , decrease in total NOS activity , augmentation of lipid peroxidation levels , upregulation of eNOS protein expression , and increase in eNOS and nNOS monomerization in the brain during ECM development strongly indicate a state of eNOS/nNOS uncoupling likely mediated by oxidative stress . Furthermore , the downregulation of Serine 1176 ( S1176 ) phosphorylation of eNOS , which correlated with a decrease in cerebrovascular wall shear stress , implicates hemorheological disturbances in eNOS dysfunction in ECM . Finally , pial arterioles responded to superfusion with the NO donor , S-Nitroso-L-glutathione ( GSNO ) , but with decreased intensity , indicating that not only NO production but also signaling is perturbed during ECM . Therefore , the pathological impairment of eNOS and nNOS functions contribute importantly to cerebrovascular dysfunction in ECM and the recovery of intrinsic functionality of NOS to increase NO bioavailability and restore vascular health represents a target for ECM treatment .
Cerebral malaria ( CM ) is one of the most severe complications of malaria infection by Plasmodium falciparum that causes unacceptably high rates of mortality and morbidity , imposing devastating health and economic burdens especially in tropical countries [1] . The murine model of CM using C57BL/6 mice infected by Plasmodium berghei ANKA ( PbA ) is a well accepted animal model for studying CM as it shares many common pathological features with human CM [2] . The pros and cons of this model have been recently debated [3] , [4] , [5] , [6] . Murine or experimental CM ( ECM ) is associated with a vasculopathy [7] which is distinctively characterized by widespread cerebral arteriolar vasoconstriction , intense microvascular inflammation and markedly reduced cerebral perfusion [8] , suggesting that cerebrovascular function is severely compromised in ECM . Cerebrovascular function is central to blood flow regulation which in turn is critical for maintaining metabolic homeostasis required for proper neurological functions [9] . Since metabolic disturbances and neurological impairment are evident in ECM [10] , [11] , understanding the underlying mechanisms of cerebrovascular dysfunction will be imperative for providing mechanistic insights into pathogenesis of the disease . Nitric oxide ( NO ) is an important physiological messenger in the brain that is implicated in the regulation of cerebrovascular tone , leukocyte adherence , platelets aggregation , synaptic transmission and cellular defense [12] . An emerging body of evidence has suggested that NO deficiency could delineate the pathogenesis of cerebrovascular dysfunction in ECM [13] , [14] . Low cerebral NO levels could be triggered by pathological consequences from the malarial parasitic invasion of flowing erythrocytes in the blood stream [15] which include hypoargininemia , increased NO scavenging by cell-free hemoglobin , depleted nitrite levels and NO quenching by reactive oxygen species ( ROS ) [16] , [17] . However , it is unclear if NO depletion could be directly linked to an intrinsic loss of NO production from its sources besides these extrinsic factors . NO is produced by NO synthases ( NOS ) which in the brain can exist in three different isoforms , namely the neuronal NOS ( Type I nNOS ) and the endothelial NOS ( Type III eNOS ) that are constitutively expressed and the inducible NOS ( Type II iNOS ) that is induced by pathophysiological stimuli [18] . A fully functional NOS exists as homodimers and this stabilized form of the enzyme favors the biochemical production of NO from L-arginine ( substrate ) and oxygen ( cosubstrate ) with tetrahydrobiopterin ( BH4 ) as cofactor [18] . eNOS , nNOS and to a much lesser extent iNOS functions are dependent on intracellular Ca2+ , the increase of which facilitates binding of calmodulin ( CaM ) to the enzymes , stimulating the electron flow involved in NO production [18] . For eNOS and nNOS , their activation can also be induced by changes in phosphorylation states of their amino acid residues , specifically phosphorylation of Serine 1176 ( S1176 ) ( eNOS ) [19] and Serine 1417 ( S1417 ) ( nNOS ) [20] and dephosphorylation of Threonine 495 ( T495 ) ( eNOS ) [19] . In particular , a major extracellular factor influencing eNOS ( S1176 ) phosphorylation is the cerebrovascular wall shear stress , a mechanical stimulus that arises from blood flow in the brain vasculature [21] . Deficiency in BH4 is known to destabilize NOS dimers by promoting the uncoupling of redox activities in their monomers , resulting in ROS such as superoxide ( O2− ) being generated rather than NO [22] . Residual NO may also react with O2− to form peroxynitrite ( ONOO− ) [12] , a highly potent oxidant that is capable of exacerbating NOS uncoupling by enhancing oxidative degradation of BH4 [23] . Since the augmentation of oxidative stress is a well-recognized pathological outcome of malaria infection [24] which can potentially deplete BH4 [23] , it is conceivable that in ECM , NOS uncoupling can contribute to low NO bioavailability , adversely affecting vascular tone in the brain . Although NOS uncoupling had been implicated in the pathogenesis of vascular disease states seen in coronary artery disease , atherosclerosis , ischemia/reperfusion injury , diabetes and hypertension [25] , [26] , [27] , its role in ECM pathogenesis remains poorly understood . It is in consensus that NOS uncoupling in these diseases is intimately associated with BH4 deficiency induced by enhanced oxidative stress . Consistent with this notion , BH4 replenishment aimed at reversing NOS uncoupling and restoring vascular function has been successfully implemented in the treatment of these diseases in both experimental and clinical studies [25] , [28] . In the present study , we investigated the role of NOS isoforms on cerebroarteriolar dysfunction in ECM and identified potential mechanisms responsible for the dysfunction . We revealed that eNOS and nNOS dysfunctions contribute importantly to the impairment of cerebroarteriolar responses in ECM which can be attributed to NOS uncoupling and downregulation of eNOS ( S1176 ) phosphorylation . The findings provide a better understanding of the involvement of NOS in cerebrovascular dysfunction during ECM , helping to design rational therapeutic interventions to more effectively tackle CM .
Parasitemia , rectal temperature and motor score of mice used for superfusion , including those used in the assessment of pial vascular responses and nitrite/nitrate production , are shown in Figs . S1A–C . As expected , ECM mice developed hypothermia and were associated with rising parasitemia and lower motor score as compared to the uninfected mice . Similar findings were obtained for mice ( Uninfected: 0 . 0% , 38 . 4±0 . 4°C & 22 . 8±0 . 4 , n = 25; ECM: 11 . 9±4 . 6% , 30 . 3±1 . 9°C & 6 . 9±4 . 4 , n = 24 ) used in all other studies ( NOS activity , LPO ( lipid hydroperoxide ) , western blot , and wall shear stress analyses ) . To eliminate a possible effect of vessel size on the magnitude of arteriolar responses , we chose vessels of comparable diameters between all superfusion experiments . As shown in Fig . 1A [open bars] , mean baseline arteriolar diameters of the vessels selected for the superfusion studies did not significantly vary among the different groups . We also ensured the return of arteriolar diameters to their baseline level before compound application during the superfusion experiment . Figure 1A illustrates that mean arteriolar diameters before the superfusion of compounds ( NOS agonist+L-NG-monomethyl arginine ( L-NMMA ) [filled bars] or BH4 [hatched bars] and S-Nitroso-L-glutathione ( GSNO ) [grey bars] ) were not significantly different from the baseline level in all superfusion groups . An improved scheme of cranial window preparation for superfusion of the pial arterioles with test compounds was used to assess cerebroarteriolar function by the pial arteriolar reactivity during ECM . We exposed pial arterioles to classical eNOS and nNOS-dependent agonists ( Acetylcholine ( ACh ) and N-Methyl-D-Aspartate ( NMDA ) ) and a NOS-independent agonist in the form of an exogenous NO donor ( GSNO ) , and examined possible changes in their diameter responses during ECM development . To verify the involvement of eNOS and nNOS in mediating the respective ACh and NMDA-induced arteriolar responses , we also make use of L-NMMA , a non-selective inhibitor of NOS . As shown in Fig . 1B , pial arterioles in the uninfected mice dilated prominently in response to ACh and NMDA applications and this effect was significantly attenuated by 78 . 4% ( P<0 . 0001 ) and 67 . 6% ( P<0 . 0001 ) , respectively with L-NMMA treatment . This verified that ACh- and NMDA-induced pial arteriolar dilations were largely mediated by eNOS and nNOS , respectively . Pronounced mean arteriolar dilation was also found after GSNO application . On the other hand , mean dilatory responses to ACh and NMDA in the ECM mice were considerably decreased from those seen in the uninfected mice by 92 . 8% ( P<0 . 0001 ) and 91 . 0% ( P<0 . 0001 ) , respectively . Any residual dilation was completely abolished with L-NMMA treatment . A significant attenuation ( P<0 . 01 ) of mean GSNO-induced arteriolar dilation was also observed during ECM development but its magnitude ( ∼35% ) was ∼2 . 6–2 . 7 folds less than those seen with ACh and NMDA . This implied that a NOS-independent mechanism can in part contribute to the impairment of ACh- and NMDA-elicited arteriolar responses in ECM but by a substantially lesser extent as opposed to relative contributions by eNOS- or nNOS-dependent mechanisms . We further tested the hypothesis that BH4 deficiency , an effector of NOS uncoupling , accounts for the loss of ACh and NMDA-induced arteriolar dilation in ECM by challenging the possibility of BH4 supplementation in restoring these responses . Representative arterioles and changes in their diameters after applications of agonists ( ACh & NMDA ) , agonist +BH4 and GSNO are shown in Fig . 1C . Notably , the application of the agonist together with BH4 was able to stimulate pronounced vasodilation ( ACh: 9 . 5% & NMDA: 11 . 8% ) that was not seen with the application of the agonist alone . Collectively , as presented in Fig . 1D , pial arterioles in ECM were not very responsive to ACh and NMDA applications . BH4 suffusion alone did not significantly alter mean baseline diameters whereas application of BH4 together with ACh or NMDA significantly enhanced dilatory responses to the agonist ( P<0 . 0001 ) or BH4 alone ( P<0 . 01 and P<0 . 0001 , respectively ) . The resultant mean magnitudes of arteriolar dilation represented ∼52% and ∼67% restorations of respective mean ACh and NMDA-elicited responses that were observed in the uninfected mice , indicating that a deficiency of BH4 could in part contribute to the impairment of vascular responses during ECM . In addition , arterioles remained relatively responsive to GSNO , dilating with a mean magnitude of 16 . 5±11 . 9% which verified that vascular smooth muscle relaxation , though reduced , was active in the ECM mice . We also examined if the above observed changes in cerebroarteriolar responses were specific to ECM by repeating the superfusion protocol in C57BL/6 mice infected by Plasmodium berghei NK65 ( PbNK65 ) , a non-ECM-inducing parasite strain . Unlike the ECM mice , pial arterioles in these non-ECM-infected mice remained relatively responsive to both ACh and NMDA treatments ( Figs . S2A & S2B ) . Their mean magnitudes of dilation were only partially blunted by 37% and 40% , respectively , from those seen in the uninfected controls , and were not significantly enhanced by BH4 supplementation . In addition , the extent of GSNO-mediated vasodilation in the PbNK65-infected mice was not significantly different from that in the uninfected mice , indicating that vascular smooth muscle activity remained intact in a non-ECM setting of malaria infection . To verify that the observed changes in cerebroarteriolar responses in ECM were indeed elicited by changes in NO production by NOS , we assessed the extent of NO generation in response to the superfusion of different test compounds by the total amount of NO metabolites ( ; sum of nitrite and nitrate ) in the superfusates . As shown in Fig . 2 , ACh/NMDA superfusion induced a significant increase in level from baseline ( P<0 . 05 for ACh and P<0 . 01 for NMDA ) in the uninfected mice . This change was significantly attenuated ( P<0 . 05 for ACh and P<0 . 01 for NMDA ) in the presence of L-NMMA , confirming that the observed pial vasodilatory responses produced by ACh/NMDA stimulation under physiological conditions were mediated by an increase in NO production . On the other hand , NO generation in response to the application of eNOS/nNOS agonists was found to be significantly blunted in ECM ( P<0 . 05 for ACh and P<0 . 01 for NMDA ) as levels after ACh/NMDA superfusion in the ECM mice were not significantly different from baseline . However , with BH4 supplementation , the produced in response to the agonists was significantly enhanced ( P<0 . 05 for both ACh and NMDA ) , corroborating that the partial recovery of the impaired ACh/NMDA-induced vasodilatory responses in ECM by BH4 was associated with restoration of NOS functionality to produce NO . We verified NOS dysfunction in the brain during ECM using a NOS activity assay . In this assay , cofactors and substrates ( including L-arginine ) required for NOS activation were added to the brain lysate samples and total NOS activity level in each sample was quantified by the rate of L-citrulline formation normalized to the amount of protein in each sample . Figure 3A shows that mean total cerebral NOS activity level of ECM mice was significantly decreased ( P<0 . 001 ) from that of uninfected mice , substantiating a decline in overall cerebral NOS function in ECM . The level of lipid peroxidation in the brain measured using a LPO assay was used to provide a quantitative assessment of oxidative stress . We sought to examine if oxidative stress is enhanced in the brains of ECM mice as compared to uninfected mice . As shown in Fig . 3B , mean LPO level in the ECM mouse brains was significantly augmented ( P<0 . 05 ) by approximately two folds as compared to the uninfected ones , demonstrating enhanced oxidative stress in the brain during ECM development . An increase in oxidative stress was previously found to induce an upregulation of renal and aorta eNOS and iNOS protein expression [29] which can induce NOS uncoupling through perturbation of the balance between BH4 and NOS bioavailability [30] . Hence , we examined if the protein expression of various NOS isoforms in the brain could be augmented under conditions of elevated oxidative stress in ECM . Western blot analysis of the denatured brain lysates indicated significantly increased amount of eNOS ( Fig . 4A; P<0 . 05 ) and iNOS ( Fig . 4B; P<0 . 05 ) but not nNOS isoforms ( Fig . 4C ) in ECM as compared to uninfected mouse brains . To examine possible alterations of eNOS and nNOS activities in ECM via changes in their phosphorylation patterns , we probed brain protein lysates with primary antibodies specifically against the phosphorylated S1176 and the inhibitory T495 sites of mouse eNOS ( Figs . 4D & 4E ) and the phosphorylated S1417 site of mouse nNOS ( Fig . 4F ) . Phosphorylation of S1176 of eNOS was significantly decreased ( P<0 . 05 ) in ECM while no significant difference was observed between uninfected and ECM mice in cases of phosphorylation of eNOS ( T495 ) and nNOS ( S1417 ) . It is noteworthy that although the total amount of eNOS almost doubled ( shown above in Fig . 4A ) , the proportion of phosphorylated eNOS ( 1176 ) decreased by more than half and therefore the total levels of phosphorylated eNOS ( S1176 ) remained lower in the ECM mice . Indeed , when normalized to β-tubulin , a 41% decrease ( 0 . 56±0 . 17→0 . 33±0 . 14 ) in total amount of phosphorylated eNOS ( S1176 ) from uninfected controls was observed in ECM . Functional NOS dimers can become unstable and dissociate into their nonfunctional monomeric forms under conditions of enhanced oxidative stress [31] , [32] . We asked if enhanced conformational inactivation of NOS isoforms by monomerization could contribute to NOS dysfunction in the brain during ECM development . The extent of NOS monomerization was analyzed in terms of the ratio between its monomeric and dimeric forms as well as the total amount of monomers . Undenatured brain lysates used to monitor the presence of SDS-resistant NOS dimers were run on a low temperature SDS-page for western blot analysis . In the case of eNOS , a significant increase in the ratio of monomers over dimers ( Fig . 5A; P<0 . 05 ) was observed . Although there was an almost two-fold increase in its monomeric form ( Fig . 5D ) , the change was not statistically significant ( P value = 0 . 059 ) . However , this observed increase in total monomeric form of eNOS , together with the significant augmentation of the monomers over dimers ratio , strongly indicated an enhanced eNOS monomerization in ECM . Furthermore , significant increases in both the ratio of monomers over dimers and the amount of monomeric forms of nNOS ( Figs . 5C & 5F; P<0 . 05 & P<0 . 05 ) but not iNOS ( Figs . 5B & 5E ) were found in ECM as compared to uninfected mouse brains , supporting an augmentation of nNOS but not iNOS monomerization in ECM . Pathological alterations in the hemorheological properties of cerebral blood flow such as hematocrit , vessel diameter and flow velocity during malaria infection [13] could result in the modulation of cerebrovascular wall shear stress . Since a decrease in eNOS phosphorylation at S1176 was found in ECM , we questioned if this effect could be triggered by a corresponding reduction of wall shear stress during ECM development . By using a closed cranial window for chronic imaging of the pial microcirculation [33] , a time-course study was performed comparing wall shear stresses ( in venules and arterioles ) between day 0 and day 6 ( after ECM development ) of PbA-infection in mice . Uninfected mice used as controls were correspondingly examined on both days . In uninfected mice , mean wall shear stress levels on day 6 were moderately altered in arterioles ( 2 . 81 Pa ( baseline ) →2 . 53 Pa ) and venules ( 1 . 94 Pa→1 . 99 Pa ) from their day 0 baselines ( Fig . 6 ) . In contrast , mean wall shear stress levels in the ECM mice decreased substantially from day 0 baselines ( arterioles: 2 . 76 Pa→1 . 75 Pa & venules: 1 . 95 Pa→1 . 33 Pa ) and these changes were significantly more pronounced ( Fig . 6; P<0 . 0001 ) as compared to those in the uninfected mice .
The present study unveiled novel insights on changes in the status of NOS functions that contributes to cerebrovascular dysfunction in ECM . Substantial blunting of eNOS- and nNOS-elicited pial arteriolar dilatory responses during ECM development would point to a loss of enzymatic eNOS and nNOS activities involved in NO production , reinforcing the notion of low NO bioavailability in ECM that compromises cerebrovascular function . Several lines of evidence lent strong support to NOS uncoupling as a potential mechanism contributing to the NOS dysfunction which includes a state of BH4 deficiency , blunting of NO generation upon stimulation by agonists , decrease in total NOS activity , increase in lipid peroxidation levels , upregulation of eNOS protein expression and enhancement of eNOS and nNOS monomerization . In the case of eNOS , its dysfunction could be aggravated by a downregulation of phosphorylation activity at S1176 and this effect could be induced by the decrease in cerebrovascular wall shear stress during ECM development . In the presence of a sustained increase in intracellular Ca2+ , the dimeric form of NOS facilitates interdomain electron flow between its monomers [18] . The transferred electrons , together with a second electron supplied by BH4 , reduce and activate O2 which in turn oxidizes L-arginine to generate NO and L-citrulline [18] . However , in the absence of BH4 or L-arginine , O2 is reduced to O2− but L-arginine oxidation fails to occur . Therefore , NOS catalytic activity is uncoupled [18] and potent ROS are produced instead of NO . Although direct assessment of BH4 bioavailability was not performed in the present study , the reversible partial recovery of impaired eNOS- and nNOS-mediated pial arteriolar responses after BH4 supplementation would imply its insufficiency in ECM . Moreover , findings on the depletion of glutathione in our previous study [14] together with an increase in lipid peroxidation levels in the brains of ECM mice demonstrated an augmentation of oxidative stress during ECM . Oxidative stress is known to be capable of promoting the degradation of BH4 to biopterin ( BH2 ) which lacks cofactor activity or suppressing the action of dihydrofolate reductase involved in BH4 recycling from BH2 via the salvage pathway [34] . Therefore , it seems likely that oxidative stress may drive NOS uncoupling in ECM by diminishing BH4 bioavailability , forming a vicious positive feedback loop of enhanced oxidation stress , BH4 depletion and NOS uncoupling which intensifies the decrease in NO production and impairs NOS-mediated cerebroarteriolar reactivity . Hypoarginemia may also play a role in NOS uncoupling during ECM [17] . However in the case of eNOS , Bevers and coworkers [35] reported that BH4 but not L-arginine supplementation was capable of ameliorating eNOS uncoupling in microvascular endothelial cells overexpressing the enzyme . In view of eNOS overexpression in ECM , it is probable that BH4 rather than L-arginine deficiency is the key factor contributing to eNOS uncoupling during ECM . This may not be surprising since eNOS uncoupling leading to NO deficiency and vascular dysfunction is a common pathological feature of many diseases associated with oxidative stress-induced BH4 deficiency [27] , [28] , [31] , [34] . The enhancement of oxidative stress as a result of NOS uncoupling could exacerbate NOS dysfunction by destabilizing the quaternary structure of the NOS dimers , causing them to dissociate into their nonfunctional monomeric forms [31] , [32] . In the present study , increased levels of eNOS and nNOS monomerization indicate that enhanced conformational inactivation of eNOS and nNOS under conditions of elevated oxidation stress in the brain could contribute to NOS dysfunction in ECM . The significant upregulation of total eNOS protein expression in the brain during ECM could be triggered by an increase in ROS activity under conditions of low NO bioavailability [29] . Similarly , a concomitant augmentation of eNOS protein and NADPH oxidases , a major source of ROS , had been identified in many vascular diseases and is recognized as a potential mechanism that exacerbates their pathogenesis through enhancement of oxidative stress by eNOS uncoupling [36] . Increased eNOS protein expression accompanied by potentially enhanced NOS uncoupling in ECM due to a lack of BH4 could lead to the generation of more ROS which can aggravate NO depletion . Excessive NO production by increased iNOS expression is known to play an immunological role in many neurodegenerative diseases and can contribute to massive vasodilation in sepsis [9] . Although total iNOS protein expression was similarly found to be significantly upregulated in ECM , its NO synthesizing capability may be impaired by NOS uncoupling mediated by BH4 depletion . However , unlike eNOS and nNOS , the lack of a substantial change in iNOS monomerization during ECM development would rule out enhanced conformational inactivation of iNOS as a possible cause for diminished NO production in ECM . This corroborates a likely scenario of low NO bioavailability in ECM stemming from the predominant decrease in NO synthesis by its dysfunctional counterparts ( eNOS and nNOS ) . It is also conceivable that strong iNOS induction in inflammatory cells might contribute to excess consumption of L-arginine and hence to substrate shortage for eNOS and nNOS , in addition to generating peroxynitrite that would contribute to NOS uncoupling . Further studies , for instance using iNOS-deficient mice , are necessary to address these potential interactions between the different NOS isoforms in vascular dysfunction in ECM . Unlike ECM , pial arteriolar responses to the application of eNOS/nNOS agonists were only partially attenuated from uninfected levels and were not significantly enhanced by BH4 in a non-ECM setting of murine malaria . These findings suggest that eNOS/nNOS dysfunction due to BH4 deficiency is more likely to mediate the impairments in pial arteriolar responses in ECM as opposed to non-ECM cases of malaria infection . Although some degree of eNOS/nNOS dysfunction may still exist in a non-ECM setting , it seems probable that majority of its associated blunted responses could be triggered by NOS-independent mechanisms that scavenge NO produced by NOS . CD8+ T cells have been shown to be key effectors of ECM , and mice without CD8+ T cells fail to develop the disease [37] , [38] . NOS inhibition has been shown to induce endothelial cell adhesion molecule expression and increase leukocyte adhesion [39] , [40] . In our previous studies [13] , [41] , we showed that exogenous NO supplementation decreased vascular inflammation ( leukocyte and platelet adherence and endothelial cell adhesion molecule expression ) and damage ( microhemorrhages and leakage ) . Therefore , in addition to contributing to vasoconstriction and impaired cerebrovascular responses , it is possible that NOS dysfunction may also contribute to vascular inflammation and possibly facilitate CD8+ T cell recruitment in ECM . eNOS can also be activated independent of a sustained increase in intracellular Ca2+ through phosphorylation of S1176 [18] which can be evoked in the brain by wall shear stress exerted by blood flow on cerebrovascular endothelial cells lining the luminal vessel wall [42] . In this study , we provide evidence of a significant decrease in wall shear stress in cerebral microvessels during ECM which resulted from both the reduction of blood viscosity ( Fig . S3A ) due to anemia ( decrease in mean Hsys; Text S1 ) and the lowering of wall shear rate ( Fig . S3B ) by a predominant decrease of blood flow velocity over vessel diameter ( Text S1 ) . Our findings on the significant decrease in phosphorylated S1176 level in ECM lend support to the reduction of cerebrovascular wall shear stress by hemorheological changes as a potential mechanism synergistically contributing to the downregulation of eNOS activity during ECM . On the other hand , a phosphorylated state of T495 in unstimulated cells is known to deactivate eNOS by interfering with binding of CaM to the enzyme . Dephosphorylation of T495 increases eNOS activity and unlike S1176 , this mechanism is less sensitive to changes in wall shear stress [21] . Instead , changes in T495 phosphorylation are generally associated with stimuli ( e . g . bradykinin , histamine and Ca2+ ionophores ) that elevate Ca2+ [19] . In the case of nNOS , phosphorylation at its S1417 is known to activate the enzyme and enhance the neuronal production of NO in the presence of Ca2+- CaM complex [20] . Levels of T495 and S1417 phosphorylation in ECM were not significantly different from those seen in the absence of infection , suggesting that these phosphorylation mechanisms are unlikely to mediate eNOS and nNOS dysfunctions in ECM . Contradictory to the many beneficial effects of BH4 on ameliorating pathological deficits [43] , [44] , [45] , our recent study [46] failed to prevent ECM by systemic BH4 supplementation which could be in part attributed to the rapid oxidation of BH4 in vivo to BH2 and failure to improve the BH4:BH2 ratios in tissues [27] . In addition , our findings revealed that BH4 supplementation in improving cerebroarteriolar responses in ECM was only effective when applied in conjunction with a NOS agonist . This suggested that not only reversing the NOS uncoupling but also restoring a possible pathological depletion of intrinsically produced NOS agonists are essential for tackling the cerebrovascular dysfunction in ECM . The failure of BH4 to fully restore cerebroarteriolar dilatory responses would suggest that NOS uncoupling due to the cofactor deficiency is not entirely responsible for cerebrovascular dysfunction in ECM . Indeed , we observed a significant loss of pial arteriolar responses in ECM even with the application of a NOS independent agonist ( GSNO ) . Therefore , additional mechanisms must be present that are deleterious to cerebrovascular function . One plausible mechanism would be the rapid quenching of the NO by massive amount of cell-free hemoglobin and ROS generated in ECM , which attenuates the effective amount of NO reaching and activating the effector vascular smooth muscle cells . Since biochemical insults in ECM can produce detrimental effects on cellular and tissue functions [24] , another possibility would be the loss of smooth muscle activity by induced cellular death [47] or disruption of signaling pathways ( e . g . decreased soluble guanylate cyclase ( sGC ) sensitivity to NO stimulation ) leading to cyclic guanosine monophosphate ( cGMP ) downregulation [48] . It is also conceivable that NO-induced cerebroarteriolar dilation may be in part counteracted by the upregulated presence of potent vasoconstrictors such as endothelin-1 ( ET-1 ) in ECM [49] . This can be further supported by a clinical study where the balance of vasoactive substances in the blood plasma of malaria patients was found to have shifted to one that favors vasoconstrictory ( ET-1 ) over vasodilatory ( C-type natriuretic peptide ) effects [50] . Since low NO bioavailability and its associated vascular dysfunction are key factors contributing to ECM pathogenesis [13] , [17] , there is a compelling need for the development of new therapeutic approaches targeting the restoration of both NO bioavailability and cerebrovascular function for ECM treatment . Recently , adjunctive interventions that involve exogenous restoration of NO levels had been promising in improving microcirculatory function and decreasing overall incidence of ECM [13] , [14] . Our current findings suggested that the use of BH4 with a NOS agonist could be a potential alternative to exogenous NO donors for ECM treatment by recovering the intrinsic functionality of NOS to produce NO and possibly prevent their undesired generation of potent ROS . These studies will demand the development of strategies to overcome BH4 oxidation occurring upon systemic delivery which has been proven by others to be the bottleneck targeting delivery to the brain [51] . Besides direct administration of BH4 , it may be worthwhile to consider the modulation of endothelial GTPCH , a regulator of BH4 synthesis in vivo [52] , to increase BH4 bioavailability .
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . All experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee of La Jolla Bioengineering Institute ( Permit Number: NO-CM003 ) , and all efforts were made to minimize suffering . Six- to eight-week-old female C57BL/6 mice ( Jackson Laboratories , Bar Harbor , ME ) were intraperitoneally inoculated with 106 PbA expressing the green fluorescent protein or PbNK65 ( donations from the Malaria Research and Reference Reagent Resource Center - MR4 , Manassas , VA; PbA-GFP: deposited by CJ Janse and AP Waters , MR4 number: MRA-865 and PbNK65: deposited by V Nussenzweig , MR4 number: MRA-268 ) for induction of ECM and non-ECM cases of malarial infection , respectively . In PbA-infected mice , parasitemia was determined using flow cytometry by detecting and counting the number of parasitic cells ( pRBCs ) that expresses GFP in relation to 10000 red blood cells ( RBCs ) from a small blood sample ( ∼1 µl ) obtained by a mouse tail end prick . In the case of mice infected by the non-fluorescent PbNK65 , thin blood smears made from a drop of tail blood and stained with Giemsa were examined under a light microscope at ×1 , 000 magnification with an oil immersion lens ( Nikon Eclipse E200; Nikon Instruments Inc , NY , USA ) . Parasitemia was then calculated by counting the number of pRBCs in at least 1 , 000 RBCs . Rectal temperature was measured with a thermocouple probe ( Oakton Acorn; Oakton Instruments , IL , USA ) . Motor behavioral score was determined by a composite scoring system based on six motor behavior tests modified from the SHIRPA protocol [53] . ECM was defined as the presentation of one or more of the following clinical signs of neurological involvement: ataxia , limb paralysis , poor righting reflex , seizures , roll-over and coma . A new cranial window preparation scheme with improved stability was adopted here ( Refer to Text S1 ) . In brief , the scheme involved performing the major surgical procedures ( skin removal and skull drilling ) to create a skull bone flap in the healthy mice which were left to recover for approximately a week . Mice were then randomly assigned to the uninfected and infected groups . Animals in the uninfected group were left for another week before cranial window implantation . On the other hand , for animals in the infected group , window implantation was carried out on day 5–6 of infection after they had developed clinical signs of ECM with hypothermia and low motor behavioral score [8] . Window implantation was performed by first retracting the bone flap and subsequently assembling a prefabricated perfusion chamber to enable superfusion of superficial pial arterioles on the exposed brain cortical surface . Uninfected and ECM mice with implanted cranial windows for superfusion were subjected to intravital microscopy . Pial arterioles were visualized through the cranial window and their vessel diameter responses were tested by sequential superfusion of the cortical surface with soluble test compounds . In the first series of superfusion experiment , responses to eNOS- and nNOS-dependent agonists ( ACh ( 10−5M ) and NMDA ( 10−4M ) ) , the agonists with non-selective NOS inhibitor , L-NMMA ( 3×10−4M ) , and NOS-independent agonist ( GSNO ( 10−3M ) ) were examined . In the event where arteriolar dilatory responses to ACh/NMDA were found to be impaired in the ECM mice , a second series of superfusion study was conducted in the sick animals using the agonist ( ACh/NMDA ) , the agonist with BH4 ( 10−7M ) and GSNO . To test if the vascular responses obtained with these compounds were specific to ECM , the superfusion protocol was repeated in PbNK65-infected animals . For assessment of nitrite/nitrate production upon superfusion of the test compounds , the superfusion protocols in both uninfected and ECM mice were repeated with the compounds dissolved in pure aCSF . Detailed descriptions on the experimental setup and superfusion protocols are provided in Text S1 . Nitrite and nitrate levels in superfusates were determined using an automated NO detector-HPLC system ( ENO-20; Eicom , San Diego , CA , USA ) according to the manufacturer's instructions described in Text S1 . Infected mice were euthanized on day 6 of infection after the animals had developed clinical signs of ECM . After euthanasia , brains were harvested and were immediately flash frozen in liquid nitrogen and stored at −80°C for subsequent analyses of NOS activity , LPO and protein expression levels . Brains from uninfected mice were used as controls . NOS activity levels in the brain samples were analyzed using a NOS activity assay according to the manufacturer's instructions ( Cayman's NOS Activity Assay Kit – Catalog no . 781001 ) described in Text S1 . The protein concentration of each brain lysate was also determined using a Bradford assay ( Cayman's Protein Determination Kit – Catalog no . 704002 ) . NOS activity was quantified by reaction rate of citrulline formation ( pmoles citrulline formed per hour ) normalized to the amount of protein in the lysate . LPO levels in the brain samples were analyzed by a LPO assay kit according to manufacturer's instructions ( Cayman's LPO Assay Kit – Catalog no . 705003 ) described in Text S1 . Brains were homogenized on ice with an Ultra-Turrax ( Ika , Werke ) in RIPA lysis buffer and processed as previously described [54] . To preserve dimer forms of NOS , samples were loaded on prechilled 4–12% gradient gels ( Invitrogen ) and ran at slow voltage in the cold . Fractions of the LDS resuspended lysates were reduced and boiled when used to detect total amounts of each NOS isoform and its respective phosphorylated form ( s ) , where applicable . After transferring onto polyvinylidene difluoride membranes and blocking and incubating with specific primary antibodies as detailed in Text S1 , bound antibodies were detected with horseradish peroxydase-conjugated secondary antibodies ( Cell Signaling Technology ) . Band intensity ( mean optical density integrated for the band area ) was quantified on unsaturated X-ray films and with ImageJ . Levels of total NOS expression and NOS monomers were presented as a ratio over β-tubulin content and normalized to the uninfected group . By using the closed cranial window preparation for chronic imaging of brain microhemodynamics as described in our previous study [33] , vessel diameter ( D ) and centerline flow velocity ( Vc ) were determined in arterioles and venules ( baseline vessel diameter = 19–65 µm ) of both uninfected and ECM mice on day 0 and day 6 of infection . Systemic hematocrits ( Hsys ) of the animals were obtained as described previously [13] . Wall shear stress ( τw ) was calculated by the product of apparent blood viscosity ( μapp ) and wall shear rate ( γw ) in the vessel . μapp which is a function of Hsys was estimated based on an in vivo empirical relationship ( parameters derived from the rat mesentery ) proposed by Pries et al . [55] ( Refer to Text S1 ) . By assuming a parabolic flow velocity profile in the vessel , γw was approximated by 4Vc/D ( Refer to Text S1 ) . Wall shear stress values were normalized to their baseline levels on day 0 . All statistical analyses were performed using a statistical software package ( Prism 5 . 0 , Graphpad ) . In all studies , unpaired or paired two-tailed student t-test was used to determine the statistical significance of the differences between uninfected and ECM groups or the effect of L-NMMA/BH4 treatment ( in the case of arteriolar response and NO metabolites ) . All reported data were in mean ± standard deviation ( SD ) except for those from the western blot experiments which were in mean ± standard error of the mean ( SEM ) . P<0 . 05 was considered statistically significant .
|
Cerebrovascular dysfunction plays a key role in the pathogenesis of murine cerebral malaria ( CM ) . Low nitric oxide ( NO ) bioavailability has been attributed as a cause of the dysfunction . However , it is unclear if the NO deficit could be inflicted by decreased NO production from its enzymatic sources ( NO synthases ( NOS ) ) in the brain . Here , we applied a novel cranial window method to elucidate the role of NOS in cerebrovascular dysfunction in murine CM . We found a state of endothelial NOS ( eNOS ) and neuronal NOS ( nNOS ) dysfunctions that are detrimental to cerebroarteriolar reactivity . Supplementation of tetrahydrobiopterin ( BH4 ) , a cofactor for NOS activation , ameliorates this effect . Supported by experimental findings on the downregulation of total NOS activity as well as the enhancement of oxidative stress , eNOS protein expression and conformational eNOS and nNOS inactivation in brains of mice with CM , we identified eNOS/nNOS uncoupling as a potential mechanism responsible for the NOS dysfunction . In addition , we discovered a downregulation of eNOS phosphorylation at its Serine residue ( S1176 ) that can possibly decrease eNOS activity and exacerbate eNOS dysfunction . Our study provides new insights on the involvement of NOS in cerebrovascular dysfunction during CM and suggests NOS as a therapeutic target for CM .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"infectious",
"diseases",
"cerebrovascular",
"diseases",
"tropical",
"diseases",
"(non-neglected)",
"neurology",
"neurodegenerative",
"diseases",
"malaria",
"parasitic",
"diseases"
] |
2013
|
Nitric Oxide Synthase Dysfunction Contributes to Impaired Cerebroarteriolar Reactivity in Experimental Cerebral Malaria
|
The regulated secretion of peptide hormones , neural peptides and many growth factors depends on their sorting into large dense core vesicles ( LDCVs ) capable of regulated exocytosis . LDCVs form at the trans-Golgi network , but the mechanisms that sort proteins to this regulated secretory pathway and the cytosolic machinery that produces LDCVs remain poorly understood . Recently , we used an RNAi screen to identify a role for heterotetrameric adaptor protein AP-3 in regulated secretion and in particular , LDCV formation . Indeed , mocha mice lacking AP-3 have a severe neurological and behavioral phenotype , but this has been attributed to a role for AP-3 in the endolysosomal rather than biosynthetic pathway . We therefore used mocha mice to determine whether loss of AP-3 also dysregulates peptide release in vivo . We find that adrenal chromaffin cells from mocha animals show increased constitutive exocytosis of both soluble cargo and LDCV membrane proteins , reducing the response to stimulation . We also observe increased basal release of both insulin and glucagon from pancreatic islet cells of mocha mice , suggesting a global disturbance in the release of peptide hormones . AP-3 exists as both ubiquitous and neuronal isoforms , but the analysis of mice lacking each of these isoforms individually and together shows that loss of both is required to reproduce the effect of the mocha mutation on the regulated pathway . In addition , we show that loss of the related adaptor protein AP-1 has a similar effect on regulated secretion but exacerbates the effect of AP-3 RNAi , suggesting distinct roles for the two adaptors in the regulated secretory pathway .
In contrast to most proteins which undergo immediate and unregulated secretion after biosynthesis , proteins destined for regulated release require sorting into LDCVs , but the mechanisms responsible for sorting to LDCVs and indeed LDCV formation remain poorly understood . LDCVs bud from the trans-Golgi network ( TGN ) [1] , [2] , [3] , and previous work has suggested that lumenal interactions such as the aggregation of granulogenic proteins drive their formation [4] , [5] . Indeed , sorting to LDCVs has been suggested to occur by default , with proteins destined for other organelles removed during the well-established process of LDCV maturation [6] , [7] . However , direct analysis of budding from the TGN has demonstrated the sorting of regulated from constitutive cargo at this early step , before maturation [3] . In addition , LDCV membrane proteins such as carboxypeptidase E and sortilin have been proposed to serve as the receptors for soluble cargo [8] , [9] . In contrast to these lumenal and membrane interactions , the cytosolic machinery involved in sorting to LDCVs and LDCV formation has remained poorly understood . Several membrane proteins contain cytosolic sequences that direct them to LDCVs . For example , the neuronal vesicular monoamine transporter VMAT2 , which fills neurosecretory vesicles with monoamine transmitter , depends on a conserved , C-terminal , cytoplasmic dileucine-like motif for sorting to LDCVs [10] , [11] , [12] , and the LDCV membrane protein IA-2β ( phogrin ) relies on a remarkably similar sequence [13] . Since the requirement for a cytoplasmic motif suggested an interaction with cytosolic sorting machinery , we recently used VMAT as a reporter to screen by RNAi in Drosophila S2 cells for proteins involved in biogenesis of the regulated secretory pathway , identifying multiple subunits of the heterotetrameric adaptor protein AP-3 [14] . Loss of AP-3 results in mis-sorting of VMAT in both S2 and mammalian neuroendocrine PC12 cells , dysregulated secretion , a reduction in the number and alteration in the morphology of LDCVs [14] . Indeed , AP-3 RNAi disrupts sorting at the TGN and impairs the concentration of membrane proteins such as synaptotagmin that are required for regulated release [14] . However , most work in mammalian cells has focused on the role of AP-3 within the endolysosomal pathway , in trafficking from early endosome to lysosome . Consistent with a role in the endolysosomal pathway , mocha mice ( Mus musculus ) lacking AP-3 show defects in lysosome-related organelles ( LROs ) such as melanosomes , platelet granules and synaptic vesicles [15] , [16] . In addition to the abnormal coat color and a bleeding diathesis , however , the animals exhibit perinatal lethality , hyperactivity , head tilt , seizures and reduced fertility [17] , [18] , [19] , and it has remained unclear whether a defect in the endolysosomal pathway can fully account for the severe phenotype . Importantly , previous work in S . cerevisiae has indicated a primary role for AP-3 in the biosynthetic pathway [20] , [21] , [22] . We have thus now used mocha mice to investigate the physiological role of mammalian AP-3 in regulated protein secretion .
To determine whether the loss of AP-3 in vivo affects regulated secretion , we cultured adrenal chromaffin cells from control and AP-3-deficient mocha mice , measuring the release of endogenous secretogranin II ( SgII ) in response to the nicotinic agonist DMPP [23] . Western blotting of the medium indicated that DMPP stimulates SgII secretion from control cells , but SgII was undetectable in the medium of mocha cells ( Figure S1 ) . However , the substantial reduction in cellular SgII content of mocha adrenal glands [14] and of cultured mocha adrenal chromaffin cells ( Figure S1 ) made it difficult to determine whether the cells simply do not contain and release enough SgII to detect , or actually exhibit a defect in regulated release . To assess regulated exocytosis by chromaffin granules , we used total internal reflection fluorescence ( TIRF ) microscopy to image neuropeptide and LDCV membrane protein reporters fused to the superecliptic pHluorin [24] , [25] . The pHluorin is a modified form of green fluorescent protein ( GFP ) with increased sensitivity to protons that is quenched at the low internal pH of LDCVs and therefore increases in fluorescence with exposure to the higher external pH on exocytosis . Since neuropeptide Y ( NPY ) -pHluorin has been shown to undergo regulated exocytosis [26] , we used lentiviral transduction to express this fusion protein and monitored individual exocytotic events at the plasma membrane of living chromaffin cells . In the absence of stimulation , control cells showed very few spontaneous fusion events over 90 s of imaging , but AP-3-deficient mocha cells exhibited substantially more ( Figure 1A–1B ) . Both control and mocha cells showed a clear increase in exocytosis in response to stimulation by DMPP , but the extent of stimulation relative to baseline reveals an ∼70% reduction in mocha cells compared to controls ( Figure 1B ) . To extend these findings to an LDCV membrane protein , we transduced chromaffin cells with a virus encoding VMAT2-pHluorin , with the lumenal location of pHluorin enabling detection of release events [27] , [28] . Similar to NPY-pHluorin , VMAT2-pHluorin also showed a clear increase in basal , unstimulated exocytosis in mocha relative to control cells , and this again resulted in an ∼75% reduction in stimulated release ( Figure 1C ) . The loss of AP-3 thus dysregulates the release of LDCVs as monitored using either soluble or membrane cargo . The role of AP-3 in defining LDCV membrane protein composition has suggested that loss of the adaptor results in mixing of constitutive and regulated secretory pathways [14] . The resulting constitutive secretion ( measured biochemically in the case of endogenous SgII [14] ) could thus reflect either the spontaneous release of constitutive secretory vesicles which contain mis-sorted LDCV cargo but no dense core , or the dysregulated release of LDCVs . To distinguish between these possibilities , we again took advantage of TIRF microscopy . Release from constitutive vesicles without a dense core should yield events with reduced amplitude relative to controls , whereas dysregulated LDCV fusion should yield events with a size similar to controls . Analyzing individual exocytotic events , we observed that the basal as well as stimulated events observed in mocha cells have an amplitude similar to those observed in controls ( Figure 1D ) . The heightened basal secretion observed in mocha cells thus apparently results from dysregulated exocytosis of LDCVs rather than the fusion of constitutive secretory vesicles without a dense core . The increased basal exocytosis of NPY and VMAT2 from mocha chromaffin cells is consistent with earlier experiments using RNAi in PC12 cells [14] , but does the loss of AP-3 also dysregulate release from other neuroendocrine tissues ? Pancreatic β cells store insulin in LDCVs morphologically and biochemically similar to chromaffin granules [29] , [30] . To assess the regulated release of insulin in vivo , we measured baseline serum insulin levels while fasting and stimulated levels 15–20 minutes after intraperitoneal injection of glucose ( Figure 2A ) . Before glucose administration , we observed a slight reduction in the serum insulin levels of mocha mice relative to controls , but this did not reach significance . After glucose administration , the control animals showed a clear increase in serum insulin but the mocha mice did not ( Figure 2A ) , suggesting a failure of regulated release . To assess the consequences of dysregulated insulin release for carbohydrate metabolism , we also measured blood glucose . Surprisingly , we observed no clear difference in fasting blood glucose levels between control and mocha mice ( Figure S2A ) . After glucose administration , mocha animals show an increase in glycemia but less than controls ( Figure S2A ) , a surprising effect in light of the lower serum insulin levels which would have been expected to impair glucose tolerance ( Figure 2A ) . The effects of the mocha mutation on blood glucose levels thus do not correlate with the effects on insulin . However , blood glucose levels reflect the combined action of multiple circulating hormones , many of which may be affected by the loss of AP-3 . Indeed , the dysregulated release of other peptide hormones may indirectly affect the release of insulin . To examine insulin release independent of systemic effects , we isolated pancreatic islets and acutely incubated them in medium containing either low or high concentrations of glucose . Figure 2B shows that in contrast to the clear stimulation of insulin release by high concentrations of glucose in control islet cells , mocha cells exhibit increased basal release with little if any stimulation by high glucose . As opposed to the reduced content of SgII in mocha chromaffin cells , cellular insulin levels show no difference between mocha and control islets ( Figure 2C ) , indicating that the change in basal secretion is not secondary to altered expression of the hormone . In the case of insulin , the mocha mutation thus dramatically and selectively impairs regulated secretion . Since the effects of the mocha mutation on release of other peptide hormones may complicate the observations in vivo , we also examined glucagon , a peptide released from pancreatic α cells that opposes the action of insulin: glucagon raises blood glucose levels in response to starvation . Although we observed no effect of the mocha mutation on fasting serum glucagon ( Figure S2B ) , baseline glucagon secretion was increased in acutely isolated islets ( Figure 2D ) . In addition , the glucagon content of mocha islet cells did not differ from controls ( Figure 2E ) , and the morphology of the islets in situ appears unchanged ( Figure S2C ) . Thus , mocha mice show dysregulated release of two peptide hormones with opposing actions , suggesting a global effect on peptide hormone release that makes it difficult to predict the net consequences for glucose homeostasis in vivo . The heterotetrameric AP-3 complex is known to exist in two isoforms , one expressed by all tissues , and another expressed more specifically by neurons and endocrine tissue including the adrenal gland and pancreatic islets [16] , [31] , [32] . In metazoan cells , the ubiquitous isoform contributes to trafficking from early endosomes to the lysosome through a pathway that does not involve multivesicular bodies [15] , [33] . In contrast , the neural isoform has been implicated in the formation of synaptic vesicles from an endosomal intermediate [34] , [35] , suggesting that this isoform may also contribute to the formation of LDCVs . To test this possibility , we used pearl mice lacking the ubiquitous isoform of the β3 subunit ( β3A ) and β3B knockouts lacking the neural isoform of β3 . Since the δ subunit of AP-3 exists as only a single isoform , and the loss of one subunit usually destabilizes the complex [17] , [36] , we first stained cells in culture for δ to assess the effect of the mutations on the complex as a whole . We were surprised to observe no effect of losing either β3 isoform on the levels of immunoreactive δ in chromaffin cells ( Figure 3A ) , particularly considering the abundance of the ubiquitous β3A subunit in most tissues . However , we did observe reduced expression of δ by β3A-deficient , non-chromaffin cells in the culture ( Figure 3B ) , presumably because they do not express the neural isoform and therefore cannot exhibit redundancy . Consistent with the relative abundance of adrenal cortical cells [37] and of the ubiquitous AP-3 isoform , western analysis of adrenal homogenates showed low levels of AP-3 δ in β3A-deficient pearl animals ( Figure 3C ) but normal levels in β3B knockouts ( Figure 3D ) . To determine whether the loss of β3A or β3B influences the formation of LDCVs , we then examined the effects on SgII . We were surprised to observe that in contrast to mocha cells which showed the reduction previously reported [14] , both β3 mutants had normal levels of SgII by immunofluorescence ( Figure 3A ) . By western analysis of adrenal extracts , both β3A-deficient pearl and β3B knockouts also contained normal levels of immunoreactive SgII ( Figure 3C–3D ) . However , loss of both isoforms in the adrenal gland of double mutant mice produced a reduction in SgII comparable to that observed in mocha mice ( Figure 3E ) . With regard to the cellular content of SgII , the two isoforms thus exhibit redundancy . The reduced expression of SgII in mice lacking AP-3 might reflect increased basal secretion or an entirely distinct process . LDCV contents have indeed been shown to undergo transcriptional regulation through a variety of mechanisms [30] , [38] . We therefore measured adrenal SgII and chromogranin A ( CgA ) transcripts from control and mocha adrenals by quantitative reverse transcription ( qRT ) -PCR . Both SgII and CgA mRNA were substantially reduced ( by ∼50% ) in mocha mice , although not to the same extent as the protein [14] ( Figure 4A ) . PC12 cells showed a similar reduction in SgII mRNA after AP-3 RNAi ( Figure 4B ) . Since AP-3 influences trafficking within the endolysosomal pathway , loss of the adaptor may also influence SgII levels through increased degradation in the lysosome . To test this possibility , we inhibited lysosomal proteases after AP-3 knockdown in PC12 cells , but did not observe any increase in the levels of SgII ( Figure 4C ) . On the other hand , the level of lysosomal hydrolase precursor procathepsin D dramatically increased in response to the inhibition of lysosomal degradation , indicating the effectiveness of the inhibitors ( Figure 4C ) . The reduction in cellular SgII observed with loss of AP-3 thus reflects reduced expression as well as increased constitutive release , but not increased degradation . AP-3 resembles AP-1 in terms of sequence , the ability to recognize similar trafficking motifs and subcellular location at endosomes and the Golgi apparatus [39] , [40] , [41] . In addition , AP-1 associates with immature LDCVs and promotes their maturation through the clathrin-dependent removal of proteins destined for other organelles [6] , [42] . In mouse pituitary AtT-20 cells , maturation indeed contributes to regulated release by removing the inhibitory protein synaptotagmin 4 [43] . In PC12 and pancreatic islet cells , however , immature LDCVs can undergo regulated release [44] , [45] . To determine whether silencing of AP-1 impairs regulated release from PC12 cells , we initially targeted AP-1 β-adaptin since this is the only mammalian AP-1 subunit without multiple isoforms and hence with reduced potential for redundancy . Despite highly efficient knockdown of the β1 subunit by RNAi , μ1A levels were not reduced ( data not shown ) , raising the possibility that the β2 subunit ( of AP-2 ) stabilized the complex by replacing β1 [46] , [47] . We therefore targeted the γ subunit of AP-1 , in particular the γ1 isoform which appears to show little redundancy with γ2 [48] . siRNA transfection reduced endogenous γ1 protein by ∼80% ( Figure 5A ) . It also reduced the stimulated secretion of SgII and the intracellular accumulation of SgII , very similar to AP-3 δ RNAi ( Figure 5B , C ) . However , normalizing to the reduced cellular stores of SgII revealed more of an increase in basal SgII secretion than a reduction in stimulated release with AP-1 knockdown , in contrast to AP-3 RNAi ( Figure 5D ) . AP-1 RNAi thus still reduces the extent of stimulation ( stimulated/basal release ) from 19-fold for control to 10-fold for AP-1 RNAi ( p<0 . 005 ) . AP-1 knockdown also potentiates the effect of AP-3 RNAi on stimulated secretion even after normalization to the reduced intracellular stores of SgII ( Figure 5D ) . AP-1 thus contributes to regulated secretion , and its role appears independent at least in part from that of AP-3 .
The results show that mocha mice have a major defect in the regulated secretion of peptide hormones . mocha animals exhibit hyperactivity , poor fertility , seizures and premature lethality [18] , [19] , but this has generally been attributed to a role for AP-3 in the endolysosomal pathway and the formation of LROs [15] , [34] . We now find that mocha animals exhibit dysregulated exocytosis of adrenal chromaffin granules and both insulin- and glucagon-containing granules from pancreatic islet cells . All of the mocha cells examined show increased constitutive release relative to their cellular stores . In addition , they all show a reduced effect of stimulation , with a virtually complete loss of regulation in pancreatic β cells . Consistent with these findings in mocha mice , AP-3 RNAi increases constitutive and reduces stimulated LDCV release in PC12 cells [14] . Considering the parallel effects of AP-3 deficiency on LDCV behavior in chromaffin and pancreatic islet as well as PC12 cells , we infer that the dysregulated release observed in mocha mice reflects a common disturbance in the formation of LDCVs . The dysregulation of release by multiple neuroendocrine populations further suggests that a global defect in regulated protein secretion contributes to the phenotype of mocha mice , although the specific peptides contributing to individual features of the phenotype such as hyperactivity and seizures remain unknown . In previous work , we found that the loss of AP-3 reduces the amount of SgII stored in PC12 cells and the adrenal gland [14] . This reflects the increased baseline exocytosis of LDCVs , but we now find that multiple granin mRNAs down-regulate as well , indicating unanticipated transcriptional effects of AP-3 deficiency . Indeed , the transcriptional down-regulation of certain LDCV cargo may account for the apparently different effects of AP-3 deficiency on different reporters and in different cells . The down-regulation of SgII mRNA in PC12 cells presumably makes it difficult to detect an increase in the absolute amount of SgII released constitutively , but transfection into PC12 cells of pHluorin-based reporters , which are not subject to this down-regulation , reveal the increased basal secretion [14] . We observe a similar increase in basal exocytosis of VMAT2- and NPY-pHluorin expressed in AP-3-deficient chromaffin cells using a lentivirus . In the case of pancreatic islets from mocha mice , cellular levels of insulin do not fall , presumably enabling us to detect the increase in basal insulin release . Similar to AP-3 deficiency , loss of the major LDCV protein IA-2 reduces expression of multiple LDCV cargo [49] , [50] , suggesting that AP-3 may sort IA-2 to LDCVs . In the absence of AP-3 , decreased LDCV IA-2 may thus result in reduced granin gene expression . Although AP-3 has a role in the endolysosomal pathway , we also find that the reduced granin content does not reflect increased degradation . The role for AP-3 in regulated secretion thus appears distinct from its well-established role in trafficking to the lysosome . The analysis of isoform-specific knockouts indicates redundancy between the ubiquitous and neural isoforms of AP-3 with regard to LDCV formation . Using SgII as a reporter for a defect in the regulated secretory pathway , we find that only the loss of both ubiquitous β3A and neural β3B causes a reduction in adrenal SgII levels . However , the loss of these isoform-specific subunits has differential effects on other trafficking phenomena . In neurons , loss of β3B mimics the effect of the full mocha mutation , with reduced presynaptic expression of proteins such as the zinc transporter ZnT3 and the chloride carrier ClC-3 [35] . Loss of β3A , on the other hand , increases presynaptic expression of these proteins [35] . The redundancy of neural and ubiquitous AP-3 forms in LDCV formation thus contrasts with the opposing roles of the two isoforms in delivery of specific proteins to the nerve terminal . How does AP-3 promote regulated secretion ? In PC12 cells , the loss of AP-3 reduces the number of LDCVs and changes their morphology [14] . They appear less dense by gradient fractionation and larger by electron microscopy ( EM ) . Consistent with this , previous work in mocha mice has shown enlarged chromaffin granules by amperometry and EM [31] . In addition , we found that AP-3 deficiency affects the membrane proteins required for regulated exocytosis: the calcium sensor synaptotagmin 1 shifts from LDCVs to lighter membranes [14] . An assay for budding from the TGN further shows that AP-3 deficiency impairs LDCV formation [14] . Despite the importance of AP-3 for LDCV formation , however , its role may be indirect , and previous work has indeed localized AP-3 primarily to endosomes [33] , [41] . Several observations have suggested a role for AP-3 at the Golgi complex . In yeast , AP-3 contributes to a direct pathway from the Golgi to the vacuole [20] , [22] , [51] . Although this has been considered specific to yeast , work in mammalian cells has more recently supported a role for AP-3 in delivery of membrane proteins from the biosynthetic pathway to the lysosome [52] , [53] . Biochemical studies have further demonstrated the specific binding of AP-3 to membranes derived from the Golgi or to artificial membranes containing the Golgi lipid phosphatidylinositol-4-phosphate ( PI4P ) [54] , [55] . Ultrastructural studies with immunogold have also demonstrated a small pool of AP-3 at the Golgi complex [33] , [41] . However , it remains possible that the role of AP-3 in LDCV formation is indirect , helping to recycle critical LDCV membrane proteins to the Golgi , or adding these proteins to LDCVs during their maturation [56] . Despite the complete loss of AP-3 and increased basal secretion , adrenal chromaffin cells from mocha mice still show residual stimulated release , raising the possibility that another system also contributes to LDCV formation . Interestingly , previous work has implicated the related adaptor AP-1 in the formation of secretory granules by other cell types , such as glue granules of the Drosophila exocrine salivary gland , and the Weibel-Palade bodies of mammalian endothelial cells [57] , [58] . AP-1 also has a clear role in LDCV maturation [6] , [42] , but immature LDCVs can undergo release from PC12 cells [45] . We were therefore surprised to find that loss of AP-1 impairs regulated release by PC12 cells . It remains possible that maturation promotes regulated secretion even if it is not absolutely required . Alternatively , AP-1 may promote regulated release independent of LDCV maturation . We also find that the loss of AP-1 exacerbates the dysregulation of release by AP-3 RNAi , suggesting that the two adaptors act through distinct mechanisms . If AP-1 promotes regulated secretion through its role in LDCV maturation , it may indeed act to remove proteins that interfere with regulated release , a process that occurs in AtT-20 cells [43] . We speculate that a proofreading role for AP-1 may become even more important in the absence of AP-3 to concentrate the membrane proteins required for regulated secretion .
All procedures involving animals were approved by the UCSF Institutional Animal Care and Use Committee . The rabbit SgII antibody was obtained from Meridian Life Science , the mouse actin monoclonal antibody from Millipore , the mouse δ SA4 monoclonal antibody from the Developmental Studies Hybridoma Bank , the goat cathepsin D antibody from Santa Cruz , the mouse HA . 11 monoclonal antibody from Covance , the mouse adaptin γ monoclonal antibody from BD Transduction Laboratories , the mouse insulin monoclonal antibody from Sigma , the guinea pig glucagon antibody from Linco , the mouse glucagon monoclonal antibody from Sigma and the rabbit somatostatin antibody from Thermo Scientific . Silencer Select rat Ap3d1 ( sense , 5′-CAUGGAUCAUGACCAAGAA-3′ ) and corresponding non-targeting control siRNAs were from Ambion . ON-TARGETplus rat Ap1g1 ( sense , 5′-CAUAAAUAUUCUUGGUCGA-3′ , 5′-GUGUGGAGAUGCACGCUUA-3′ , 5′-UGUAACAGUGAUAACGAUA-3′ , 5′-GGACUGGAAUUCACGGCAA-3′ ) and corresponding non-targeting pooled control siRNAs were from Dharmacon . The sequences of NPY-pHluorin ( a generous gift of R . Holz , U . Michigan ) and VMAT2- pHluorin were amplified by PCR to add 5′ BamHI and 3′ EcoRI sites , then subcloned into the FUGW lentiviral expression vector , replacing the EGFP coding sequence . PC12 cells were maintained in DMEH-21 medium supplemented with 10% horse serum ( HS ) and 5% cosmic calf serum ( CCS; HyClone ) in 5% CO2 at 37°C . siRNA transfection was performed using Lipofectamine 2000 ( Invitrogen ) according to the manufacturer's instructions . HEK293T cells were maintained in DMEH-21 medium with 10% fetal bovine serum ( FBS ) in 5% CO2 at 37°C . Lentivirus was produced by transfecting HEK293T cells with FUGW , psPAX2 and pVSVG and Fugene HD ( Roche ) according to the manufacturer's instructions [59] . Mouse adrenal chromaffin cells were isolated and cultured as previously described [60] . Briefly , adrenal glands were dissected and placed in cold Ca++- , Mg++-free ( CMF ) Hank's balanced salt solution ( HBSS ) . The surrounding fat and cortex were removed and the medullae transferred to tubes containing 300 U/ml Collagenase I ( Worthington ) in CMF-HBSS . Medullae were dissociated by shaking for 40 min at 37°C . Collagenase solution was then replaced by CMF-HBSS containing 200 µg/ml DNAse I ( Sigma ) and 1% heat-inactivated FBS ( Gibco ) , the tissue triturated first with a P200 pipette tip , then with a 23 gauge needle . The cells were pelleted at 300 g for 8 min at room temperature and resuspended in pre-warmed culture medium . Cells were maintained in DMEH-21 medium supplemented with 10% FBS and antibiotics . For lentiviral transduction , freshly isolated chromaffin cells were plated in viral supernatant , and fresh medium was added the following morning . For TIRF microscopy , control or mocha chromaffin cells were plated onto glass chambers coated with poly-L-lysine , immediately transduced with lentivirus encoding either NPY- or VMAT2-pHluorin and imaged live 4–7 days later . Images were typically collected for 40–50 ms at 10 Hz and room temperature using an inverted TIRF microscope ( TE2000E; Nikon ) with 100× Plan Apo 1 . 49 NA oil objective , a 1 . 5× tube lens and an electron-multiplying charge-coupled device camera ( QuantEM; Photometrics ) . Basal exocytosis was measured in Tyrode's solution containing ( in mM , 119 NaCl , 25 HEPES-NaOH , pH 7 . 4 , 30 glucose , 2 . 5 KCl , 2 CaCl2 , 2 MgCl2 ) over 90 s , and release stimulated for 60 s in Tyrode's containing 5 µM 1 , 1-Dimethyl-4-phenylpiperazinium ( DMPP; Sigma ) . Individual exocytotic events were quantified manually using NIS-Elements software ( Nikon ) . The amplitude of individual exocytotic events was measured by placing 2×2 pixel ROIs manually over the center of events , and the mean ROI intensity prior to an event subtracted from the maximum event intensity . Glucose tolerance was assessed and serum insulin levels measured using 5–15 week-old mocha mice and age-matched controls . Mice were fasted overnight ( ∼16 h ) , weighed the following morning , and blood samples collected for baseline glucose and insulin levels . Mice were then injected intraperitoneally with glucose at 2 mg/g body weight and blood samples collected from the tail vein at the time points indicated . Blood glucose levels were measured using the FreeStyle glucometer ( Abbott ) . To measure serum insulin , the blood was allowed to clot , sedimented at 2000 g and 4°C for 20 min , and insulin levels determined using the Ultra Sensitive Mouse Insulin ELISA kit ( Crystal Chem ) . Islets were isolated as previously described from 9–16 week-old mocha mice and age-matched controls [61] . Briefly , islets were purified on a Ficoll gradient and allowed to recover for 1 h at 37°C . Five islets were aliquoted into tubes containing HEPES-buffered RPMI medium supplemented with either 2 . 8 mM glucose ( basal ) or 28 mM glucose ( regulated ) and incubated for 1 h at 37°C . The islets were then sedimented and the supernatants collected to measure insulin secretion . The pellets were resuspended and sonicated in 2 mM acetic acid , 0 . 25% RIA-grade BSA to extract intracellular insulin . Finally , the nuclei were lysed by additional sonication in 67 mM ammonium hydroxide , 0 . 2% Triton X-100 . Secreted and cellular insulin were quantified by ELISA ( Mercodia ) according to the manufacturer's instructions , and islet DNA quantified to confirm that the amount of islets per tube was similar between conditions . Glucagon was measured by ELISA ( R&D Systems ) . Chromaffin cells were fixed by adding an equal volume of 4% formaldehyde in CMF-PBS to the culture medium and incubating for 20 min at room temperature . Cells were blocked and permeabilized in CMF-PBS containing 2% BSA , 1% fish skin gelatin and 0 . 02% saponin . Primary antibodies were diluted in blocking solution at 1∶500 ( SgII ) and 1∶100 ( δ SA4 ) . The following secondary antibodies conjugated to Alexa Fluor dyes ( Invitrogen ) were used at 1∶1000 in blocking solution: goat anti-rabbit IgG 488 and goat anti-mouse IgG 594 . Images were acquired using a Zeiss LSM 510 confocal microscope and 100× oil objective . pearl and mocha mice were obtained from the Jackson Laboratory , and mocha animals were backcrossed to C57BL/6 to remove grizzled and Pde6brd1 alleles . Ap3b2 KO mice were obtained from V . Faundez ( Emory ) and S . Voglmaier ( UCSF ) . Double mutant pe/pe; Ap3b2−/− mice were generated by crossing pe/pe; Ap3b2+/− males to pe/+; Ap3b2+/− females . Adrenal glands from 3–6 week-old mice were homogenized in 150 mM NaCl , 50 mM Tris-HCl , pH 8 . 0 , 1% NP-40 , 0 . 5% sodium deoxycholate , and Complete protease inhibitors ( Roche ) with 1 mM EGTA and 1 mM PMSF . After sedimentation at 14 , 000 g to remove nuclei and cell debris , 20–40 µg protein was separated by electrophoresis through polyacrylamide , transferred to nitrocellulose , and the membranes immunoblotted for AP-3δ and SgII , with actin as loading control and the appropriate secondary antibodies conjugated to IRDye800 ( Rockland Immunochemicals ) . The immunoreactivity was quantified by imaging with an Odyssey system ( LI-COR Biosciences ) and ImageJ ( National Institutes of Health ) , and the signals normalized to actin . For western analysis of the SgII secreted from chromaffin cells , Tyrode's solution was collected , sedimented at 300 g for 3 min at 4°C , and the supernatant mixed with SDS-PAGE sample buffer before electrophoresis through polyacrylamide . Chromaffin cells were directly lysed by the addition of sample buffer . In this case , SgII and actin were detected using ECL plus ( GE Healthcare ) . Total RNA was isolated from PC12 cells and mouse adrenal glands using TRIzol reagent ( Invitrogen ) according to the manufacturer's instructions . To improve the yield of adrenal RNA , ultrapure glycogen ( Invitrogen ) was added to the TRIzol as carrier . In addition , adrenal RNA was treated with RNAse-free DNAse I ( NEB ) to remove contaminating genomic DNA . cDNA was synthesized using oligo ( dT ) or gene-specific primers and a Transcriptor First Strand cDNA Synthesis kit ( Roche ) . qPCR was performed with SYBR Green ( Applied Biosystems ) on a Stratagene Mx4000 machine . The following primers were used: rat SgII fwd: 5′-ACAATATAAGACAGAGGAAAATTTT-3′ , rev: 5′-TGGATAAGAAGCAGAACTG-3′; rat β-actin fwd: 5′-CCGTGAAAAGATGACCCAGATC-3′ , rev: 5′-CAGGGACAACACAGCCTG-3′; mouse CgA fwd: 5′-CCAACCGCAGAGCAGAG-3′ , rev: 5′-AGCTGGTGGGCCACCTT-3′; mouse SgII fwd: 5′-AAGTGCTGGAGTACCTCAACC-3′ , rev: 5′-TTACATGTTTTCCATGGCCCG-3′; mouse GAPDH fwd: 5′-ATGGTGAAGGTCGGTGTGAAC-3′ , rev: 5′-TCCACTTTGCCACTGCAAATG-3′ . Two days after the second siRNA transfection , PC12 cells were incubated for ∼24 h in complete medium supplemented with vehicle or a cocktail of lysosomal protease inhibitors ( Sigma ) including ( in µM ) 10 antipain , 10 leupeptin and 5 pepstatin A . Cells were washed on ice with cold PBS and lysed by the addition of 50 mM Tris-HCl , pH 8 . 0 , 150 mM NaCl , 1% Triton X-100 , and Complete protease inhibitors ( Roche ) plus 10 mM EDTA and 1 mM PMSF . Samples were analyzed by quantitative fluorescent immunoblotting . Cells were transfected with siRNA ( 100 nM ) on days 1 and 3 after plating , washed 2 days later and incubated in Tyrode's solution containing 2 . 5 mM K+ ( basal ) or 50 mM K+ ( stimulated ) for 30 min at 37°C . The supernatants were collected , cell lysates prepared as described above , and the samples analyzed by quantitative fluorescent immunoblotting . Statistical analysis was performed using the Student's two-tailed t-test unless otherwise indicated .
|
The physiological action of peptide hormones and neural peptides depends on their sorting to vesicles capable of regulated exocytosis in response to stimulation . Despite the diversity and importance of signals released by this pathway , surprisingly little is understood about the molecular mechanisms involved in sorting to and indeed formation of the large dense core vesicles ( LDCVs ) that mediate regulated secretion as opposed to secretory vesicles that undergo constitutive release . We recently used RNA interference in cell lines to identify a requirement for the adaptor protein AP-3 in sorting to the regulated secretory pathway , but the importance of this role in vivo has remained unknown . Using mutant mice lacking various subunits of the AP-3 complex , we now show that AP-3 is indeed required for appropriate , regulated secretion in multiple neuroendocrine cell types . Although AP-3 exists as both ubiquitous and neuronal forms , we also find that either form alone suffices to confer regulated secretion . The results show that AP-3 plays a novel and essential role in regulating the release of peptide hormones and neural peptides .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2013
|
Widespread Dysregulation of Peptide Hormone Release in Mice Lacking Adaptor Protein AP-3
|
GLUT1 facilitates the down-gradient translocation of D-glucose across cell membrane in mammals . XylE , an Escherichia coli homolog of GLUT1 , utilizes proton gradient as an energy source to drive uphill D-xylose transport . Previous studies of XylE and GLUT1 suggest that the variation between an acidic residue ( Asp27 in XylE ) and a neutral one ( Asn29 in GLUT1 ) is a key element for their mechanistic divergence . In this work , we combined computational and biochemical approaches to investigate the mechanism of proton coupling by XylE and the functional divergence between GLUT1 and XylE . Using molecular dynamics simulations , we evaluated the free energy profiles of the transition between inward- and outward-facing conformations for the apo proteins . Our results revealed the correlation between the protonation state and conformational preference in XylE , which is supported by the crystal structures . In addition , our simulations suggested a thermodynamic difference between XylE and GLUT1 that cannot be explained by the single residue variation at the protonation site . To understand the molecular basis , we applied Bayesian network models to analyze the alteration in the architecture of the hydrogen bond networks during conformational transition . The models and subsequent experimental validation suggest that multiple residue substitutions are required to produce the thermodynamic and functional distinction between XylE and GLUT1 . Despite the lack of simulation studies with substrates , these computational and biochemical characterizations provide unprecedented insight into the mechanistic difference between proton symporters and uniporters .
The glucose transporter GLUT1 catalyzes facilitative diffusion of glucose into red blood cells[1] and across the blood-brain barrier[2] . The bacterial homologues of GLUT1 are all proton symporters whereby the transmembrane proton gradient is employed to drive the uphill translocation of the substrate saccharides into the cell[3] . The distinct transport mechanisms are consistent with their working environment . The glucose concentration in blood maintains at around 5 mM and the intake glucose is immediately metabolized to glucose-6-phosphate in the cytosol , thereby creating a constant transmembrane gradient of glucose[4] . A facilitative uniporter is thus sufficient to mediate the uptake of glucose . In contrast , bacteria may have to hunt under stringent conditions . A co-transport mechanism may ensure efficient uptake of the nutrient at low concentration in the environment . Interestingly , GLUT1 and its bacterial homologues share considerable sequence similarities[5] , raising the question of what is the determinant under the mechanistic divergence between the closely related proton symporter vs . uniporter . The xylose:proton symporter XylE from E . coli is one of a number of rigorously characterized GLUT1 homologues . In recent years , crystal structures of GLUTs[6 , 7 , 8] and XylE were determined in multiple conformational states[5 , 9 , 10] , in line with alternating access mechanism of membrane transporters[11] . Structure-guided mutational analysis identified Asp27 of XylE to be the protonation site for symport[12] . The D27N variant of XylE that was designed to mimic the neutral residue Asn29 at the corresponding position of GLUT1 , however , lost transport activity in in vivo experiments , despite fully active in in vitro counter-flow assays[12] . Thus , the behavioral difference between uniporters and symporters exemplified by GLUT1 and XylE cannot be accounted for simply from the perspective of protonation-site residues . Instead , atomic-level description on the transitions between alternating-access states and quantitative evaluation on the thermodynamics of these processes are required to address these questions . Molecular dynamics ( MD ) simulations , which resemble in silico single-molecule experiments at atomic resolution , emerge as a suitable tool for investigating conformational transitions of macromolecules[13] . In addition to the direct observation on large-scale conformational changes , thermodynamics and many physical properties could be rigorously evaluated to elucidate the internal causes of molecular behaviors[14 , 15 , 16 , 17] . In this work , we used MD simulations to study the alternating-access transitions of three apo systems: XylE with Asp27 protonated ( denoted as XylE_H ) , XylE with Asp27 deprotonated ( denoted as XylE_noH ) and GLUT1 . Despite the lack of substrates , these simulations provided the quantitative details for the structural transitions from the inward-facing ( IF ) to outward-facing ( OF ) states , which are informative parts of the complete transport cycle . From the free energy profiles calculated for the transitions , we not only revealed the coupling between Asp27 protonation/deprotonation and conformational transition of XylE , but also identified a remarkable thermodynamic difference between XylE and GLUT1 . Subsequently , to further understand the mechanism of these thermodynamic observations , we developed Bayesian network ( BN ) models to analyze changes of residue interaction networks during conformational transitions . Besides mechanistic illustration , these statistical models predicted a few residues essential for the appropriate conformational preference in XylE , which was then validated by experiments on the corresponding mutants . More importantly , our results suggested that the thermodynamic divergence between XylE and GLUT1 arises from multiple residue substitutions accumulated during evolution . Based on a group of residues inferred from our models , we successfully designed a uniporter-like XylE mutant , which was then confirmed by experimental validation . Conclusively , our computation results provided insight into the mechanistic difference between proton symporters and uniporters in the absence of substrates .
GLUTs and XylE have a classical MFS fold of 12 transmembrane ( TM ) helices plus a unique intracellular helical ( ICH ) domain that comprises of 4 or 5 helices[7] . Structural comparison and accompanying biochemical characterizations of the sugar porter ( SP ) family members suggested that the two TM domains , the N-terminal domain ( NTD ) and C-terminal domain ( CTD ) , undergo concentric rotations relative to each other to accomplish transition between inward- and outward-facing conformations , resulting in the alternating access of the substrate binding site ( s ) from either side of the membrane[18 , 19] . To study the conformational transition , we started from the inward open GLUT1 and inward occluded XylE with Asp27 protonated/deprotonated denoted as XylE_H/XylE_noH hereafter . The conventional MD ( cMD ) simulations ( Sim #1 , #2 and #3 in Fig 1B ) revealed little change with RMSDs of TM domains < 1 . 5 Å ( S1A Fig ) for all three systems . Subsequently , accelerated MD ( aMD ) simulations were initiated to encourage conformational transitions without directional inclinations ( Sim #4 , #5 and #6 ) . Conformational states captured by aMD simulations were illustrated on a 2D-map of Extracellular Gate distance and Intracellular Gate distance between NTD and CTD , which depict the extent of opening towards periplasm and cytoplasm respectively ( Fig 1A , see Free energy calculations in Methods section ) . As shown in Fig 1C , GLUT1 rapidly left the initial state and underwent IF→OF transition ( also in S1B Fig ) , suggesting that the uniporter has a barrier-free energy landscape consistent with facilitative transport . Unlike GLUT1 , aMD trajectories of XylE_H and XylE_noH were confined to inward- and outward-facing conformations respectively , without discernible IF↔OF transitions ( Fig 1C; S1C and S1D Fig ) . Therefore , apo XylE might possess high-energy transition state ( s ) that would require substrate binding to stabilize . We then performed extensive string method with swarms of trajectories ( SMwST ) simulations using conformations selected from the aMD and cMD trajectories for transition path identification ( Sim #7 , #8 and #9; see S2B Fig for convergence ) . Further equilibrations of on-path images ( Sim #10 , #11 and #12 ) permitted path reconstruction in the space of gate distances and subsequent free energy calculations by bias-exchange umbrella sampling ( BEUS ) scheme ( S2A Fig ) . In the BEUS simulations ( Sim #13 , #14 and #15 ) , each window/image represents a conformation of the IF→OF transition pathway . The final free energies were evaluated ( see Free energy calculations in Methods section ) both in the space of gate distances ( S2C Fig ) and along window/image index ( Fig 1D ) . Low RMSDs between structures in individual windows and the crystal structures of various states validate the efficacy of our path-finding protocol ( S1E Fig ) . The three systems exhibit drastically different free energy profiles ( Fig 1D ) . GLUT1 shows a broad energy well which is required for the fast turnover of passive uniporters . The slight preference of OF over IF conformations agrees with the unidirectional conformational shift in the aMD simulation ( S1B Fig ) . Notably , the GLUT1 profile exhibits no favorable energy wells at the window corresponding to the inward-facing crystal structure 4PYP ( S1F Fig ) . The reason at least partially arises from the mutation and detergent introduced to the crystal structure ( see S1 Text ) . In contrast , an energy barrier is present at the image index of 20 in both XylE systems . According to our rough estimation , this barrier of ~5 kcal/mol can effectively forestall rapid IF↔OF transition ( Fig 1D ) and thereby suppressing proton leakage across the membrane in the absence of substrate ( see S1 Text for details ) . The marked difference between XylE_H and GLUT1 at the energy barrier cannot be accounted for purely by the protonation-site residue . In addition , comparison on free energy profiles between XylE_H and XylE_noH indicates that conformational preference of XylE could be altered thermodynamically by protonation/deprotonation . Particularly , once the proton and substrate have been unloaded , inward-facing XylE would spontaneously restore the energetically more stable outward-facing state for another cycle of transport . The systematic difference between XylE and GLUT1 may hinder accurate side-by-side comparison on their free energy profiles . We thus developed an RMSD-scoring based algorithm to align BEUS windows for the three systems , which can effectively eliminate the cross-system window shift along the path ( see Window alignment in Methods section; S3 Fig ) . From the post-alignment free energy profiles of the three systems , the IF and OF states as well as the transition state ( TS ) can be unanimously represented by windows renumbered as A3 , A16 and A12 respectively ( S2C Fig and S3C Fig ) . Comparing aligned states of the three systems , the most noteworthy motion that can be visualized during transition is the domain rotation ( S1 Movie and S4B Fig ) . Hence , we superimposed all structures upon their NTDs and investigated the global conformational changes using principal component analysis ( PCA ) on TM helices ( see Analysis techniques in Methods section ) . The top 3 principal components explain ~85% of the structural variations in all systems and show strong inter-system correlations , which reflects consensus rocker-switch movements for both uniporters and symporters ( S4 Fig ) . Local changes accompanying the conformational transition were characterized by monitoring the per-residue structural fluctuations among all aligned states ( S5 Fig ) . As expected , NTD and CTD remain nearly rigid in majority of the TM helices ( RMSF < 1 Å ) . In addition to the loop regions , TM7b exhibits flexibility within all three systems , and thus may be a local gating element coupled with the global conformational transition . The TMs show different packing patterns in the extracellular and intracellular gates . Unlike side-by-side helix anchoring in the former , TM5 and TM11 insert into the interfaces between TM8/10 and TM2/4 by partial twisting in the latter ( S4C Fig ) . The relationship between global conformational transition and gating can be seen from the pore radius analysis ( S6C Fig ) . Interestingly , we identified a unique gate around Tyr298 in XylE that confines the periplasmic entrance ( pointed by arrows in S6C Fig ) . Although this gate becomes fully open in a transient state ( window A18 ) to allow substrate binding , its constriction tendency may cause considerably lower rate of substrate dissociation in XylE than in GLUT1 . Apart from the global movement , TM7b bending contributes to the extracellular gating ( see the definition of kinking angle in S6A Fig ) . In GLUT1 , the severe kinking of TM7b diminishes in the outward-facing state that exhibits striking structural resemblance to the outward-open GLUT3 ( PDB ID: 4ZWC ) . Conversely , the helix remains bended in the outward-facing conformations of both XylE_H and XylE_noH , thus creating the unique gate ( S6 Fig ) . A proline ( Pro301 ) residing in the middle of TM7b indicates why this helix is reluctant to adopt the straight conformation in XylE . The above conformational changes cannot explain the functional and thermodynamic differences between uniporter GLUT1 and symporter XylE despite that they are consistent with the structural studies . We then switched our focus to the interactions around the meaningful protonation-site residue , i . e . Asp27-Arg133 bonding in XylE and the corresponding Asn29-Arg126 interaction in GLUT1 , as reported[5 , 10] . Analysis on hydrogen bonds ( H-bonds ) reveals that side-chains of the Asn29-Arg126 pair in GLUT1 never form favorable interactions in any conformational states , whereas those of Asp27-Arg133 in XylE_H preserve moderate H-bonds in TS and OF states ( Fig 2; S1 Movie ) . This observation and thermodynamic distinction between GLUT1 and XylE_H jointly negate the proposition that simple neutralization of Asp27 in XylE could eliminate their functional gap . The Asp27-Arg133 bonding in XylE_noH system is well maintained in all conformations due to electrostatic attraction , which is consistent with structural observations[10] . In contrast , this interaction is greatly impaired in the inward-facing state of XylE_H . Considering that XylE_H and XylE_noH only differs in the protonation state of Asp27 , changed bonding strength of Asp27-Arg133 could be the origin of their different thermodynamics . To understand the molecular basis for the thermodynamic difference between XylE and GLUT1 observed in simulations , we made side-by-side comparison on the changes in residue interaction networks during their conformational transitions . Conventional modeling strategies of residue interaction network usually construct undirected graphs with nodes representing residues or Cα atoms , and with edges reflecting correlated motions and/or physical contacts[20] . Although it has been reported that causality could be extracted from such networks[21] , these mutual-information based approaches show limited predictive power on network tuning and had a poor performance in our case . We designed a novel method of H-bond network modeling , named as the Interaction Regulation with Bayesian Networks ( IRBN ) , to infer the causal relationship between all interacting residue pairs and to search for factors responsible for the evolutionary divergence using the Bayesian network , which is widely used for causality inference in omics data analysis[22] . Instead of abstracting residues as nodes , inter-residue H-bonds were symbolized as basic units to construct network models , since they can bridge network intervention and pairwise interaction energies , the latter of which jointly correlate with free energy ( see Modeling of hydrogen bond networks in Methods section for details ) . To reduce the network complexity , H-bonds formed by identical moieties of side-chain ( s ) ( SC ) and/or backbone ( s ) ( BB ) are treated as a whole ( Fig 3A; also see S7 Fig for comprehensive illustration of network learning and inference ) . Intuitively , the aggregate number of H-bonds in a network ( defined as HB ) negatively correlates with free energy , and therefore the change of HB ( ΔHB ) could be used to assess perturbation in free energy ( ΔG ) for a specific conformational state induced by certain mutations . To decipher the mutational effect on the free energy change during the transition between two conformation states ( ΔΔG ) , we extended the concept to ΔΔHB by mimicking a thermodynamic cycle ( Fig 3B ) . To clarify distinct conformational preferences of XylE systems , we first constructed the network models for aligned IF , TS and OF conformations , and then disabled specific H-bonds of a polar/charged side-chain ( called mutation here ) in the models to calculate its ΔΔHB of OF→IF transition . Positive ΔΔHBOF→IF signifies that the mutation introduces an additional bias to favor the inward-facing state ( in respect of HB ) , relative to the outward-facing state . Fig 3C shows the ΔΔHBOF→IF values of the tested mutations in XylE systems . The diagonal distribution demonstrated that the network interventions generally led to equivalent outcomes in XylE_H and XylE_noH systems . The mutations that gave rise to substantial positive or negative ΔΔHBOF→IF values in both XylE systems were categorized as IF-favoring and OF-favoring respectively ( Fig 3D ) . Mutation on either Asp27 or Arg133 favored inward-facing conformation , supporting our hypothesis that proton-coupled conformational preference originates from the modulation on Asp27-Arg133 bonding strength . Mutations that weakened or destroyed the Asp27-Arg133 interaction , i . e . D27N and D27L , were tested by semi-quantitative PEGylation assays ( see PEGylation assay in Methods section; S8 Fig ) . As shown in S1 Text , results of PEGylation for certain mutation plus L65C or V412C should be interpreted by comparison with L65C or V412C , respectively . Thus , we can conclude that the population of outward-facing states dramatically declined in D27N and completely diminished in D27L , comparing to WT XylE ( S8C Fig ) . The results support that Asp27 protonation modulates conformational preference in XylE through adjusting the local H-bond network . Other residues whose mutations showed significant conformational preference were mainly involved in inter-domain interactions , including residues located at the NTD-CTD interface and conserved cytoplasmic motifs in SP family ( Fig 3D ) . PEGylation assays showed that notable changes could only be detected in the mutants with markedly perturbed salt bridges , such as D337L , D27L and K305M , possibly because salt bridges contribute more to free energy than regular H-bonds . Similar to D27L , the D337L mutant strongly prefers inward-facing state ( S8C Fig ) . Despite that mutation on Lys305 ( K305M ) was expected to favor outward-facing conformations , its impact is weaker than D27L and D337L , considering the smaller magnitude of its ΔΔHBOF→IF value ( less than cutoff , see Fig 3C and S8C Fig ) . Since Lys305 is the only salt-bridge forming residue predicted to favor outward-facing state , we constructed double mutants combining K305M and mutations in OF-favoring region of Fig 3C including Y298F and Y179F . Consistent with model prediction , in comparison to the WT XylE , the inward-facing conformations almost disappear in the tested double mutants ( S8C Fig ) . We thus confirm the predictive power of the newly developed methodology . Likewise , the same analysis on GLUT1 predicted multiple potential mutations for stabilizing typical conformations , which awaits further experimental validation ( S8B Fig ) . To explore detailed mechanisms for the mutational perturbation on HB values , we evaluated the structure of Bayesian network models by model averaging ( S7C Fig ) and presented the individual nodes that were intensely perturbed upon network tuning ( see Modeling of hydrogen bond networks in Methods section ) . Considering the coincidence of Asp27-Arg133 bonding and free energy discrepancy between XylE_H and XylE_noH ( see Fig 1D and Fig 2B ) , we disrupted the Asp27-Arg133 side-chain H-bonds in network models but retained other interactions concerning Asp27/Arg133 ( e . g . , Asp27-Glu206 ) ( Fig 4A ) . Following this protocol , we investigated the change in the architecture of H-bond networks upon Asp27-Arg133 H-bond dismissal in XylE systems . When forcing the number of Asp27SC-Arg133SC H-bonds to zero for Bayesian network inference , the overall HB value of IF state is considerably less weakened than those of TS and OF states in both systems , suggesting that breaking this interaction triggers inward-facing preference ( Fig 4B ) . Specifically , in the XylE_H system ( Fig 4C ) , Asp27SC-Arg133SC disruption introduces no changes in IF state , whereas numerous inter-helix H-bonds are weakened in TS and OF states . Moreover , the upheaval of network architecture in diverse states revealed drastic variation of interaction patterns . In the XylE_noH system ( Fig 4D ) , unlike the TS and OF states , H-bond loss regarding Asp27 in IF state ( i . e . Asp27SC-Arg133SC and Asp27SC-Glu206SC ) can be partially compensated by multiple connected interactions ( i . e . Ser102SC-Arg133SC and several TM4 backbone H-bonds ) . We speculate that the sacrifice of these compensated H-bonds upon Asp27SC-Arg133SC fortification may be the cause of unfavorable inward-facing conformation in deprotonated XylE . Complete removal of the H-bonding capability of Asp27 triggers more intricate network perturbations , yet generates a comparable ΔHB pattern destabilizing TS and OF states more than the IF state ( S9 Fig ) . However , similar treatment with Asn29 side-chain in GLUT1 does not induce strong IF or OF preference ( S9D Fig ) , possibly because of the dramatic rearrangement of local H-bond network as exemplified by the accumulation of hydrophobic residues surrounding Asn29 in GLUT1 ( S9A Fig ) . Therefore , multi-fold evolutionary events may have occurred to modulate the local environment surrounding Asp27 in XylE and Asn29 in GLUT1 in divergent directions . As shown in the free energy landscapes , the most crucial discrepancy between XylE and GLUT1 is the presence/absence of energy barrier at transition state . To elucidate its molecular basis , we sought for side-chain mutations in XylE_H and GLUT1 systems that would alter the energetics towards each other based on Bayesian network inference ( Fig 5A ) . Considering the potential negative correlation between free energy and total amount of H-bonds ( HB ) , a GLUT1 mutation that devastates H-bonds at transition state more than IF/OF states is likely to convert its free energy profile towards XylE_H pattern , and vice versa . Possible determinants for evolutionary divergence can be predicted from the values of ΔΔHBIF→TS and ΔΔHBOF→TS , thusly . Since it was impractical to consider all possible combinations , we only focused on screening single side-chain mutations of GLUT1 and XylE . The mutations that would generate a barrier in GLUT1 ( located in the blue region of Fig 5B ) and vice versa in XylE_H ( located in the red region of Fig 5B ) , were itemized in Fig 5C . After disposal of unaligned , conserved and binding-site residues in both transporters , a total of 7 residues were identified in GLUT1 to account for its thermodynamic divergence from XylE . These residues are H-bonding donors/acceptors in GLUT1 but are replaced by nonpolar residues with comparable volume in XylE ( Fig 5C ) . To verify their joint performance , we mimicked a combination of 7 point mutations to replace the XylE residues with the corresponding ones in GLUT1 ( T45V , T60A , D236L , S294A , T295P , S313V and Y424M ) in the Bayesian network models of GLUT1 and re-evaluated the HB loss in various states ( Fig 5D ) . Just as expected , the HB value of the mutant decreases dramatically at TS state ( by > 3 ) , thus supposedly converting the flat energy landscape of GLUT1 into a XylE_H-like fashion and prohibiting rapid IF↔OF transition . Unfortunately , we could not obtain well-behaved protein of these GLUT1 mutants required for biochemistry characterization . As a compensation , we tested the effect of these residues in the reverse direction , trying to convert XylE into a uniporter . Besides the 7 candidate residues , D27N was also introduced into XylE , and the mutants were evaluated by in vivo transport assays ( Fig 5E ) . The comparable transport activity between D27N variant with negative control indicates that D27N mutation alone is insufficient for the symporter-to-uniporter conversion ( see S11A Fig and S1 Text ) . Similarly , incorporating a single mutation ( V43T , A52T , A300S , P301T , L248D , V321T , or M428Y ) with D27N presented little to no activity increase . However , the combination of all mutations led to a significantly higher transport efficiency and thus indeed converted a symporter to a functional uniporter . In vitro counterflow assays also confirm the functionality of this 7-mutation variant comparing to uniporters GLUT1 and GLUT3 ( S11B Fig ) . Notably , the identified residues are not confined to a sub-region of the structure ( Fig 5D ) , implying that elaborate adjustment of interactions rather than simply altering protonation-site residues may be essential for driving uniporters and symporters to divergent directions .
We hereby present multiplex analyses for the sugar porter family members XylE and GLUT1 using MD simulations and a novel approach based on Bayesian networks for residue interaction analysis . The free energy calculations on the 3 systems discriminate XylE from GLUT1 , and highlight the protonation state of Asp27 as the molecular basis for conformational preference alteration in XylE . Inspection of IF↔OF transitions reveals numerous details that are consistent with structural and biochemical studies , supporting the reliability of aMD+SMwST path-finding scheme . The newly developed modeling framework IRBN plays a pivotal role in mechanism illustration . For instance , from the network models , we can predict that H-bond loss following the disruption of Asp27-Arg133 interaction in the inward-facing state of XylE_noH will be compensated by three contiguous interactions ( Fig 4C ) . By artificially manipulating residue interactions , IRBN helped disclose molecular determinants for the uniporter/symporter divergence , hence implying that complex variations instead of mere Asp↔Asn substitution are required for the functional interconversion . As for GLUT1 , we could also explain the mechanism of some disease-related mutations . As shown in S10A–S10F Fig , these mutations unanimously perturb the energetic balance among states by disrupting local H-bond network , and consequently impair fast turnover required for facilitators . Moreover , our approach of combining MD simulations and IRBN modeling could be extended to the fields of protein engineering and drug development , and could facilitate rational design as well as allosteric regulation studies for pharmaceutically important targets . Under the guidance of computer simulation and statistical modeling , we successfully transformed a symporter ( XylE ) into a uniporter by neutralizing protonation site residue and reducing the energy barrier . In contrast , converting GLUT1 to a proton symporter would presumably need more intricate changes to meet following requirements: ( 1 ) ability to load and unload proton , ( 2 ) no proton leak in the absence of substrate , and ( 3 ) no substrate leak without change in the protonation state . Here , we provided several necessary tactics for the uniporter-to-symporter design , which were generalized from our calculations on thermodynamics: ( 1 ) possess a titratable residue as protonation site , ( 2 ) create an energy barrier in apo state , and ( 3 ) destabilize inward-facing state in deprotonated state . Notwithstanding these advances , it is noteworthy that relative free energies between IF and OF states in XylE_H and GLUT1 systems slightly disagree with experimental observations[23] for two possible reasons: ( 1 ) the applied force field did not consider polarizable dipole-dipole interactions that were supposed to fasten the extracellular gate[7] , and ( 2 ) boost energies in aMD simulations partially destroyed the integrity of ICH domain in the outward-facing conformations , which should be intact as exemplified by outward-facing GLUT3 structures ( PDB ID: 4ZW9 , 4ZWB , and 4ZWC ) [6] . In summary , we thoroughly investigated crucial reactions of the transport cycles emphasized in Fig 5F . Comparison of three apo systems provides detailed understanding of transporter mechanisms . Since sugar porters may recognize both α- and β-anomers as substrates , the uncharacterized processes in Fig 5F still await extensive computational and biophysical research .
Using the plugins in VMD[24] , we established three systems , i . e . Asp27 protonated XylE ( XylE_H ) , Asp27 deprotonated XylE ( XylE_noH ) and GLUT1 . The structure of inward-open GLUT1 mutant ( N45T & E329Q , PDB ID: 4PYP ) were preprocessed for MD simulations with 3 modifications: ( 1 ) Gln329 was mutated back to Glu as in WT GLUT1 , ( 2 ) the detergent β-NG whose head group occupies the binding pocket was removed , and ( 3 ) the missing ICH5 ( residue 459 to 468 ) was constructed as α-helix referring to template structures of the outward-facing XylE ( PDB ID: 4GBY ) and GLUT3 ( PDB ID: 4ZWC and 4ZW9 ) using Modeller 9v12[25] . For XylE , we selected the inward-occluded structure ( PDB ID: 4JA3 ) , and modeled all missing loops using the outward-occluded conformation ( PDB ID: 4GBY ) as the template . PROPKA 3 . 1[26 , 27] was used to determine the protonation states of titratable residues other than Asp27 in XylE at pH 7 . 0 . In specific , Glu206 was neutralized in both XylE systems . XylE was inserted into a palmitoyl-oleoyl-phosphatidyl-ethanolamine ( POPE ) bilayer , given the fact that the majority of E . coli . lipids ( ~75% ) belong to the PE class[28] , . To mimic the physiological condition for GLUT1 , a palmitoyl-oleoyl-phosphatidyl-choline ( POPC ) membrane was adopted since it has been reported to restore transport activity of purified GLUT1[29] . After solvation and neutralization in 150 mM NaCl , the total number of atoms reached ~86 , 000 for each system . We generated input files and performed simulations using MD simulation suites AMBER12[30] and AMBER14[31] . The transporters were parameterized by ff12SB force field , and were surrounded by LIPID11 phospholipids[32] and TIP3P water molecules[33] . With the protein and ligand fixed , the systems first underwent a 5000-step minimization . Then , a 1-ns melting of lipid tails was simulated in an NVT ensemble at 310 K with the rest of the system constrained with a large force constant k = 100 kcal/mol/Å2 . Afterwards , one heating procedure ( k = 10 kcal/mol/Å2 for protein ) was carried out from 0 K to 310 K under constant volume condition for 1 ns . Next , the value of k was set to 1 kcal/mol/Å2 for another 1-ns pre-equilibration . To further naturalize the lipid bilayer , two 5-ns runs in the NPγT ensemble ( 1 atm of pressure ) were performed with k = 0 . 1 kcal/mol/Å2 on Cα atoms and no constraint at all , sequentially . We deliberately selected surface tension γ for bilayers ( γ = 17 dyn/cm for POPC and γ = 26 dyn/cm for POPE membrane ) suggested by the reported tests of LIPID11 force field[32] . Under periodic boundary conditions ( PBC ) , all pre-equilibrations were performed with the time step of 1 fs and the van der Waals cutoff of 10 Å , using the Particle Mesh Ewald ( PME ) method to estimate electrostatics[34] . Initially for each system , we performed an NPγT cMD simulation for 100 ns ( Sim #1 , #2 and #3 ) using the time step of 2 fs and with the SHAKE algorithm[35] applied . Tiny fluctuations of the system volume indicated that membrane and solvent molecules were well equilibrated , and therefore we fixed volume and used the GPU implementation of PMEMD[36] for the subsequent simulations . It is usually unrealistic to sample the large-scale conformational change of transporters by cMD simulations , because of the long autocorrelation time ranging from micro- to milliseconds or even longer . Therefore , we conducted aMD simulations ( Sim #4 , #5 and #6 ) to sample the conformational transition , based on a reported study of GPCR[37] . Preserving the shape of energy surface , aMD adds a boost potential ΔV ( r ) to adjust total potential and/or dihedral potential[14]: V ( r ) *=V ( r ) +ΔV ( r ) , ΔV ( r ) = ( Ep−V ( r ) ) 2αp+Ep−V ( r ) + ( Ed−Vd ( r ) ) 2αd+Ed−Vd ( r ) . Here Ep and Ed denote the reference values for the total and dihedral potentials respectively , while V ( r ) and Vd ( r ) denote the total and dihedral potentials calculated for the current state of the system . The boost energy that is tuned by the parameters αp and αd is applied only in the situation of V < E . We set Ep and Ed as the average potentials of the second halves of the 100-ns cMD trajectories , and expressed the parameters via: Ed=Vd⋅ ( 1+λd ) , αd=λd5⋅Vd , Ep=V+λp⋅Natom , αp=λp⋅Natom . After testing some combinations of parameters , we found that enhanced sampling could be visualized within 100 ns for λd = 0 . 3 and λp = 0 . 2 , without perturbing the system stability ( < 5% loss of helical contents ) . Subsequently , we extended the aMD simulations to 500 ns for all systems . Flattening energy barriers on the path , aMD allowed the protein structure to evolve faster and presented an overview of the conformational transitions . For MFS transporters , identification of the pathway for the IF↔OF conformational transition typically requires an a priori reaction path that in principle should not deviate much from the minimum free energy path ( MFEP ) , since otherwise even extensive iterative optimization can hardly guarantee the path convergence , particularly when an energy barrier is present between the a priori path and MFEP . In this work , the a priori paths were collected from the unbiased aMD trajectories , and therefore should be close to the MFEP in principle . This protocol simplifies the path selection process , which traditionally requires the tedious evaluation on enormous candidate trajectories produced by permuting the applying sequence of all relevant artificial collective variables . The a priori paths were then refined iteratively using the SMwST method until convergence . We have utilized the GPU code of AMBER12 to run the SMwST[38] simulations in order to relax and optimize paths ( Sim #7 , #8 and #9 ) . In total , 51 discretized conformations were selected for each system to compose an a priori pathway , including 2 end-point structures . The images were picked from previous cMD and aMD trajectories . Unlike the aMD trajectory of GLUT1 , XylE_H and XylE_noH systems sampled either IF or OF conformations with slight overlapping . Hence , we combined the trajectories of Sim #4 and #5 to construct the IF↔OF transitions of XylE , and unified protonation state for each pathway . To obtain stable end points of the a priori path , IF or OF snapshots with the largest Intracellular/Extracellular Gate distances were chosen and were then equilibrated for 5 ns following a 10 , 000-step minimization . The initial path connecting 2 end points should meet three criteria described as follows: To ensure continuity in high-dimensional space , we used Cartesian coordinates of the Cα atoms as the collective variables for SMwST calculation . The iterative method mainly follows a 4-step procedure as published[39 , 40]: The above procedure was repeated for 60 iterations to iteratively refine the path . Distance between two paths could be evaluated by the summation of pairwise RMSDs between corresponding images . In the convergence tests , the path of the current iteration was compared with the initial path ( named as std 1 ) and that of 4 iterations before ( named as std 2 ) . The path-finding algorithm is thought to reach convergence if both distances become nearly constant , independent of the iteration index . By comparison with the convergence scenario in [40] , the average drifts in RMS space not only reach plateaus more rapidly in our cases , but also hold lower values ( <1 . 5Å vs . >3Å ) . The observation indicates that initial paths found by aMD are sufficiently close to MFEP and are likely to outperform those sampled from TMD used in [40] . The final SMwST paths were used to initiate follow-up calculations . We introduced 2D reaction coordinates for PMF calculations based on a priori knowledge . As NTD and CTD undergo nearly rigid-body rotation in line with the alternating access mechanism , we simplified the metric of IF↔OF transitions to Extracellular and Intracellular Gate distances , both of which quantify the gate opening using separation between the centers of mass ( COM ) of two Cα groups . Specifically , residues of the two groups describing XylE Extracellular Gate , XylE Intracellular Gate , GLUT1 Extracellular Gate and GLUT1 Intracellular Gate are listed below: ( 1 ) XylE Extracellular Gate: residue group {28–34 , 58–63 , 178–183} vs . residue group {295–301 , 315–320 , 423–428} , ( 2 ) XylE Intracellular Gate: residue group {75–80 , 149–154 , 160–166} vs . residue group {332–337 , 391–397 , 404–410} , ( 3 ) GLUT1 Extracellular Gate: residue group {30–36 , 66–71 , 171–176} vs . residue group {289–295 , 307–312 , 419–424} , and ( 4 ) GLUT1 Intracellular Gate: residue group {83–88 , 141–147 , 153–159} vs . residue group {324–329 , 387–393 , 400–406} , respectively . The free energy calculations were performed in the BEUS scheme[41] . The method has a feature of swapping adjacent windows/replicas with moderate acceptance ratio to improve sampling continuity . Ideally , images of a convergent SMwST path can serve as the window centers for BEUS simulations . For complex systems with high dimensionality , however , hundreds of swarmed trajectories in the SMwST calculations are far from sufficiency to yield a drift that depicts the real gradient . Therefore , SMwST images could stray from MFEP . To overcome this problem , we performed a 5-ns equilibrium simulation for each image to generate an ensemble of structures around the SMwST path , and then used a curve reconstruction algorithm[42] to regenerate the path in the 2D space of gate distances ( S2A Fig ) . The 5-ns equilibrium simulations ( Sim #10 , #11 and #12 ) can not only facilitate path reconstruction , but also produce sufficient conformations to shed light on the positions of putative energy barriers and basins . To estimate the PMF profile of an IF↔OF transition , the reconstructed path was divided into 30 windows , in each of which the simulated structures were restrained in the space of gate distances by a biasing potential U ( ζ ) =k2⋅ ( ζ−ζ ( s ) ) 2 , where ζ is the space coordinate , ζ ( s ) represents the center of window and k is the force constant . The window exchange was initiated every 20 ps following a deterministic odd-even scheme[43] . For each window exchange , two neighboring windows swapped their centers following an acceptance probability as described below: Pij=min{1 , exp ( −Δ ) } , Δ=1kBT{[Ui ( Xj ) +Uj ( Xi ) ]−[Ui ( Xi ) +Uj ( Xj ) ]} , where Pij is the acceptance probability for the swapping between window i and j , kB is the Boltzmann constant , T is the absolute temperature , U is the potential energy and X is the Cartesian coordinate . Test simulations were conducted on the conformations collected from trajectories of Sim #10 , #11 and #12 to optimize the window centers and corresponding force constants , so as to guarantee moderate exchange rates of 20–40% between all neighboring windows . Ultimately , we conducted 30 replicas × 20 ns/replica = 600 ns of BEUS simulations for each path . Due to limited computing resources , we produced the final PMF profile using the last 16 ns trajectories in all BEUS windows , which included 24 , 000 structures ( 30 windows × 800 snapshots/window ) in total . Bayesian block bootstrapping scheme in combination with the weighted histogram analysis method ( WHAM ) was used for the estimation of PMF and error bars[44 , 45] . For a BEUS replica , we measured its autocorrelation times in dimensions of the reaction coordinates and window index . The maximal autocorrelation time τ is less than 2 ns . We thus divided the BEUS time series within one window into 4-ns non-overlapping blocks . Notably , sizes of blocks were at least twice of τ , thus avoiding inter-block correlation . The block sampling technique resolves the violation of independent and identical distribution case by standard bootstrapping , and provides better uncertainty estimations as discussed in previous literatures[16 , 17 , 41] . The BEUS windows of XylE_H , XylE_noH and GLUT1 need to be properly aligned before pathway comparison . Following superimposition , the structures in each BEUS window were averaged to generate a representative conformation for the window . Therefore , a total of 30 representative structures were generated for each path . Distances between representative structures were evaluated by RMSD . To estimate RMSD between XylE and GLUT1 , the structures should be superimposed on their matching residues in the TM domain . In specific , residue {12–35 , 66–84 , 91–111 , 122–174 , 186–207 , 272–297 , 307–357 , 367–401 , 411–428 , 434–451} in GLUT1 matches the residue {10–33 , 59–77 , 84–104 , 130–182 , 200–221 , 279–304 , 316–366 , 372–398 , 408–425 , 446–463} in XylE . Ahead of multi-path alignment , we first performed pairwise alignment by calculating pairwise RMSDs between the representative conformations of the two paths ( S3A Fig ) . Here , we recorded the results in a 30 by 30 matrix denoted as A , where A[i , j] is the RMSD between window i of path 1 and window j of path 2 . We then generated legitimate routes between the end points A[0 , 0] and A[29 , 29] , requiring that as long as A[i , j] is on a route , either A[i+1 , j] ( i∈{0 , 1 , … , 28} , ( j∈{0 , 1 , … , 29} ) or A[i , j+1] ( i∈{0 , 1 , … , 29} , ( j∈{0 , 1 , … , 28} ) should stay on the route . By this way , any route including ( i , j ) would have the form of { ( 0 , 0 ) , … , ( i , j ) , ( i+1 , j ) , … , ( 29 , 29 ) } or { ( 0 , 0 ) , … , ( i , j ) , ( i , j+1 ) , … , ( 29 , 29 ) } . We assume that the optimal route should have the minimal summation of RMSDs ( ∑ ( i , j ) ∈routeA[i , j] ) . Note that this definition does not infer a one-to-one mapping between representative structures . In the multi-path alignment , the windows from 3 paths were regarded as matched , only when every pair of them was connected by the above route finding algorithm ( S3B Fig ) . Among the one-to-many mapping of the 3-window sets representing the same conformation , the one with the lowest summation of pairwise RMSDs was chosen to guarantee the uniqueness of state matching . To evaluate the conformational changes , we conducted principal component analysis for GLUT1 , XylE_H and XylE_noH systems using structural snapshots in the aligned BEUS windows ( S4A Fig ) . Ahead of principal component analysis , the Cα atoms of residue {12–35 , 66–84 , 91–111 , 122–174 , 186–207} in GLUT1 and residue {10–33 , 59–77 , 84–104 , 130–182 , 200–221} in XylE systems were used to align the structures upon the NTDs . The same TM atom set as adopted in the preceding section ( “Window alignment” ) was used here to estimate the principal components . To estimate the kinking angle of TM7b , we first split this helix into two 7-residue short helices , i . e . residue 287–293 and 294–300 in GLUT1 , residue 285–291 and 292–298 in GLUT3 , as well as residue 293–299 and 300–306 in XylE . We then constructed an ideal α-helix using ( Ala ) 7 and superimposed this ideal helix to the two halves of TM7b respectively . The axes of the two superposed ideal helices could be used to estimate the kinking angle ( S6 Fig ) . Pore radii were calculated using the program HOLE2[46 , 47] with default parameters . H-bonds were evaluated using the “Hydrogen bonds” plugin of VMD , with the angle and distance cutoffs chosen as 45° and 3 . 0 Å respectively . The networks of residue interactions imply crucial details in the IF↔OF transitions . Among all kinds of interactions , we studied H-bonds explicitly , considering the easiness of mutation and experimental verification . We developed a modeling procedure named IRBN , which constructs Bayesian network models for the aligned BEUS windows . Any system can be regarded as a set of interdependent component , which are described by random variables X1 , X2 , … , Xn . Using Bayesian networks , causality relationship between the components can be inferred and the joint probability distribution of all random variables can be estimated , given sufficient observation data . Ideally , one may attempt to fully describe a high-dimensional problem using the joint probability distribution P ( X1 , X2 , … , Xn ) obtained from the observations ( i . e . data set ) . However , this is impractical to handle computationally for large n because the number of parameters increases exponentially . Taking each variable as a node , Bayesian networks factorize the joint distribution and assign inter-node arcs according to conditional dependencies , thereby reducing the complexity and revealing causality between nodes . The obtained Bayesian network models provide mass knowledge on information propagation through nodes and supports probabilistic reasoning at the given evidence . Here , we only investigated the H-bond networks for the IF , TS and OF states . In total , 3 window ( representing IF , TS , OF states ) × 3 path = 9 windows , each of which has a sample size of 1000 , were studied . The Bayesian network models were trained using BEUS trajectories , where the conformations were saved every 20 ps for 20 ns . Note that 20 ps is sufficiently longer than the autocorrelation time of H-bond rearrangement . We determined the H-bonds using simple geometric criteria with the angle and distance cutoffs chosen as 45° and 3 . 0 Å respectively . For a protein structure , each residue was decomposed into side-chains ( SC ) and backbones ( BB ) , either of which could act as a moiety to form H-bond interaction . Logically , we articulate a random variable ( node ) as the number of H-bond between two such moieties . For example , the number of H-bonds formed between the backbone of Asn29 and the side chain of Arg126 in GLUT1 is defined as a discrete random variable called Asn29BB-Arg126SC , which has a value range of VAL ( Asn29BB-Arg126SC ) = {0 , 1 , 2} . The H-bonds that occur in less than 10% of the structural snapshots were considered trivial for a particular window , thus were not involved in the network . Also , H-bonds involving certain loop regions that have RMSFs larger than 2 . 0 Å ( XylE: residue 106–123 , 430–442 , 465–480 , GLUT1: residue 454–469 ) as shown in S5A and S5B Fig , were not taken into account , because the 20-ns BEUS simulations cannot guarantee complete sampling in the conformation space of these loops . Following the above requirement , each network contains 400–500 nodes . To speed up structure learning of the networks , it is reasonable to assume that two nodes are disconnected if their moieties do not contact . By considering physical interactions , an arc blacklist was constructed , and only node pairs outside the blacklist could be connected by arcs ( called candidate arcs ) in the following Bayesian network learning . Here , physical contact is counted either if on average > 1 pairs of heavy atom proximate within the cutoff of 4 . 0 Å , or if nontrivial H-bonds are found . As a result , we implicitly included the van der Waals interactions into network modeling . The regime of IRBN is thoroughly illustrated in S7 Fig . We used the bnlearn[48] and gRbase[49] packages in R for the learning and inference of Bayesian network . Model construction and parameter estimation followed a two-step process: ( 1 ) structure learning to generate the structure of the directed acyclic graph ( DAG ) , and ( 2 ) parameter learning to determine the local probability distributions . For discrete Bayesian networks , we performed structure learning using a hybrid algorithm named general 2-Phase Restricted Maximization ( RSMAX2 ) , which is a combination of constraint- and score-based algorithms . We chose the Semi-Interleaved HITON-PC for the restrict step ( phase 1 ) , which is optimal for high-dimensional data sets[50] . In detail , the conditional independences of nodes were evaluated by the mutual information tests using semi-parametric χ2 distribution[51] ( α = 0 . 01 as the type I error threshold ) . After learning the skeleton of the DAG by Semi-Interleaved HITON-PC , the structure underwent fine-tuning by a score-based tabu search ( length of tabu list = 50 ) . The score of the Bayesian Dirichlet equivalent uniform ( BDe ) posterior density ( imaginary sample size = 5 ) was maximized in tabu search in the structural tuning process[52] . Once the highest score structure was learned from data , the step of parameter learning was executed by the commonly used Bayesian estimation with imaginary sample size of 5 . In rare cases , the nodes representing nontrivial H-bonding that comprises of constant number of H-bonds in all snapshots were added back to the network manually after model learning . It is worth mentioning that the parameters and algorithms described above were chosen based on 5-fold cross-validation results in one dataset ( the window of IF state in GLUT1 ) . Setting log-likelihood loss ( also known as negative entropy ) as our loss function , we evaluated the combinations of following items: ( 1 ) cutoff for physical contacts ( 0 . 5 , 1 , or 2 pairs of contiguous heavy atoms ) , ( 2 ) constraint-based algorithms named Max-Min Parents and Children ( MMPC ) , or Semi-Interleaved HITON-PC for phase 1 , ( 3 ) score-based algorithms denoted as hill-climbing or tabu search for phase 2 , and ( 4 ) the significance level of dependency test ( α = 0 . 01 , 0 . 02 , or 0 . 05 ) . According to the testing results , the loss function is insensitive to the choice of parameters/algorithms . We thus selected aforesaid combination that has relatively small loss and moderate number of arcs . The features of a Bayesian network model can be estimated by resampling techniques . Here , we performed 10 times of 5-fold cross-validation on each dataset , thereby generating a total of 50 Bayesian network models ( each constructed from 80% of the dataset ) for feature estimation and model averaging ( S7C Fig ) . Under the assumption that net change of H-bonds correlates with the free energy change ΔG , we defined a network feature as HB to depict the expectation of the total number of H-bonds: HB=E ( ∑iXi ) =∑iE ( Xi ) . Clearly , HB can be estimated as the summation of E ( Xi ) , which can be computed from the marginal distribution of random variable Xi at each node . To measure the HB change , i . e . ΔHB of a network caused by specific mutations , one can perform Bayesian network inference at a given evidence Evi={Xj1=e1 , Xj2=e2 , … , Xjk=ek} , j1 , j1 , … , jk∈{1 , 2 , … , n} , which quantitatively describes random variables affected by the mutations , and calculate conditional distribution P ( Xi | Evi ) . Consequently , the expression of ΔHB is written as: ΔHB=HBMut−HBOri=∑i[E ( Xi|Evi ) −E ( Xi ) ] , where HBOri denotes the HB of original unperturbed Bayesian network model , while HBMut denotes the HB of Bayesian network model after mutational intervention . Furthermore , we can investigate the efficacy of some mutations on the transition between two states , such as ΔΔHBOF→IF = ΔHBIF - ΔHBOF , which imitates ΔΔG of a thermodynamic cycle . In the investigation on information propagation , we applied model averaging using 50 Bayesian network models to guarantee the robust network structures . An arc was included in the averaged DAG only if it emerged in > 90% of the 50 models . In addition , the arc was considered directed if the network connection shows evident directionality ( proportion of same direction > 80% ) . In the graphical representation of the robust network ( see Fig 4C and 4D and S9B and S9C Fig ) , the highly oriented arcs representing strong causality are shown in arrows , while the other arcs are represented as dashed lines . Methoxypolyethylene glycol maleimide 5 , 000 ( mPEG-Mal-5K; Sigma-Aldrich ) , serving as a membrane-impermeable PEGylation reagent , can be covalently attached to solvent accessible thiol groups , thus change the mobility of molecules in electrophoresis . In spite of its native cysteines , WT XylE embedded in E . coli membranes shows no shifting on gel after mPEG-Mal-5K treatment , suggesting that mPEG can access none of the native thiol groups . We introduced a cysteine at Leu65 or Val412 in combination with certain mutations for inspection . The specific sites of L65 and V412 were selected based on solvent accessibility analysis of simulated conformations , as L65 or V412 is only exposed in OF or IF state , respectively . All XylE mutants were subcloned into pET15b ( Novagen ) with an N-terminal 6 × His tag . E . coli BL21 ( DE3 ) cells were used as expression system and were incubated in shakers at 220 r . p . m . at 37°C . When OD600 of Luria-Bertani medium reached 1 . 5 , overexpression was induced by 250 μM isopropyl β-d-thiogalactoside ( IPTG ) for 4 h . The cells were harvested and washed twice before resuspension by PEGylation buffer containing 25 mM HEPES-Na pH 7 . 5 , 150 mM NaCl and 10% ( v/v ) glycerol . Reaction system comprised of 0 . 1 g/ml cells with or without sonication and 10 mM mPEG-Mal-5K in PEGylation buffer . The mixtures were gently shaken for 1 h at 20°C . Then the reaction was quenched by 20 mM dithiothreitol ( DTT ) followed by SDS-PAGE and western blotting analysis ( S8C Fig ) . XylE variants were subcloned into pET15b ( Novagen ) and transformed into E . coli BL21 ( DE3 ) cells for transport activity measurements in vivo . E . coli cells were induced with 250 μM IPTG for 1 h , when OD600 of Luria-Bertani medium reached about 1 . 5 . Ice-cold MK buffer ( 150 mM KCl , 5 mM MES-K pH 6 . 5 ) was used to wash the harvested cells twice and subsequent resuspension . To energize the cells , we added glycerol to a final concentration of 10 mM , and left the reaction system at room temperature ( 25°C ) for 2 min right before sugar uptake . Cell density of the 100 μl reaction system should be adjusted to OD600 = 2 . 0 . The transport reaction triggered by 0 . 83 μM extracellular D-[3H]xylose addition was allowed for 30 seconds at room temperature , and was stopped by rapid dilution in 2 ml of ice-cold MK buffer followed by filtration through 0 . 22-μm cellulose acetate filter ( Sartorius ) and 2 ml MK buffer washing . The filter membranes were taken for liquid scintillation counting . Another MFS transporter FucP , which is unable to transport D-xylose , served as negative control . All the XylE mutants were expressed on similar levels quantified by western blotting . We mainly followed previous protocols for protein purification and liposome preparation[5 , 6 , 7] with minor modifications . N-his tagged 7-mutation variant was extracted from E . coli BL21 ( DE3 ) cells with the addition of 20 mM D-xylose throughout the purification procedure . GLUT1 was tagged with N- terminal FLAG for better behavior on gel filtration . The liposomes were reconstructed in KPM 6 . 5 buffer ( 50 mM potassium phosphate , 2 mM MgSO4 pH 6 . 5 ) , with 20 mg / ml E . coli polar lipids ( Avanti ) , 1% β-OG ( Anatrace ) , 200 μg / ml protein , and 20 mM D-xylose or D-glucose . Liposomes containing no protein served as negative control . After detergent removal , extrusion and ultracentrifugation , proteoliposomes were re-suspended with ligand-free KPM buffer right before transport assays . Counterflow assays were performed at room temperature ( 25°C ) . The external concentration of hot substrates was 0 . 83 μM for both D-[3H]xylose ( 12 Ci / mmol ) and D-[3H]glucose ( 20 Ci / mmol ) . The uptake was allowed for 30 s , followed by rapid filtration through 0 . 22-μm filters ( Millipore ) and liquid scintillation counting .
|
We seek to address one intriguing question , the mechanistic distinction between active proton-coupled symporters and passive uniporters that are related in evolution . Proton-coupled symporters harness the transmembrane proton gradient to drive the substrate transport , while uniporters can only facilitate the passive substrate translocation . In this work , we focus on two sugar transporters GLUT1 and XylE , which belong to symporters and uniporters respectively but have high sequence similarity . We first applied molecular dynamics simulations to characterize the thermodynamic behaviors of apo GLUT1 and XylE , which are supposed to provide prominent details of mechanisms . From the identified difference in thermodynamics , we concluded that neutralizing protonation site in XylE is insufficient for its conversion to GLUT1 analog . To pinpoint extra elements contributing to their evolutionary divergence , we developed a novel network modeling scheme based on Bayesian network which shows impressive predictive power on residue mutations . Our models suggested the detailed mechanism of proton coupling in XylE and molecular basis of symporter/uniporter discrepancy . Furthermore , the modeling scheme could help to guide the design of biomolecules for desired functions .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"protons",
"crystal",
"structure",
"condensed",
"matter",
"physics",
"network",
"analysis",
"crystallography",
"thermodynamics",
"computer",
"and",
"information",
"sciences",
"solid",
"state",
"physics",
"proteins",
"pegylation",
"biophysics",
"nucleons",
"free",
"energy",
"physics",
"biochemistry",
"biochemical",
"simulations",
"nuclear",
"physics",
"post-translational",
"modification",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"computational",
"biology",
"biophysical",
"simulations"
] |
2017
|
Molecular determinants for the thermodynamic and functional divergence of uniporter GLUT1 and proton symporter XylE
|
Martinotti cells are the most prominent distal dendrite–targeting interneurons in the cortex , but their role in controlling pyramidal cell ( PC ) activity is largely unknown . Here , we show that the nicotinic acetylcholine receptor α2 subunit ( Chrna2 ) specifically marks layer 5 ( L5 ) Martinotti cells projecting to layer 1 . Furthermore , we confirm that Chrna2-expressing Martinotti cells selectively target L5 thick-tufted type A PCs but not thin-tufted type B PCs . Using optogenetic activation and inhibition , we demonstrate how Chrna2-Martinotti cells robustly reset and synchronize type A PCs via slow rhythmic burst activity and rebound excitation . Moreover , using optical feedback inhibition , in which PC spikes controlled the firing of surrounding Chrna2-Martinotti cells , we found that neighboring PC spike trains became synchronized by Martinotti cell inhibition . Together , our results show that L5 Martinotti cells participate in defined cortical circuits and can synchronize PCs in a frequency-dependent manner . These findings suggest that Martinotti cells are pivotal for coordinated PC activity , which is involved in cortical information processing and cognitive control .
Martinotti cells , ubiquitous to the cortex [1] , are the most prominent cross-laminar interneuron subtype forming synapses in layer 1 onto the distal dendrites of cortical pyramidal cells ( PCs ) [1–3] . Despite this close structural relationship , the role of Martinotti cell inhibition is not clear . Studies identifying Martinotti cells by various markers have found different morphologies and microcircuit connectivity depending on the cortical layer in which their cell bodies reside [2] . In general , the division of neocortical interneurons into either parvalbumin- , somatostatin ( SOM ) - , or 5HT3aR-expressing cells [4–6] has been helpful for dissecting neural functionality; yet , these groups can be further subdivided and show partial overlap between interneuron markers . Martinotti cells are a subclass of SOM+ cells [7 , 2 , 4] , and several combinations of transgenic lines have been created to try to genetically and morphologically isolate Martinotti cells [8–10] . For example , the SOM-cyclization recombinase ( Cre ) mouse line marks layer 1–projecting Martinotti cells with cell bodies in both layer 5 ( L5 ) and layer 2/3 ( infragranular and supragranular layers ) but also labels non-Martinotti cells in layer 4 [10] . Although electrophysiologically , SOM+ Martinotti cells are often referred to as low-threshold spiking ( LTS ) neurons [3] or slow-inhibitory interneurons [11] , early studies have shown up to four different firing patterns for Martinotti cells [2 , 9] . Functionally , cortical SOM+ interneurons have been suggested to provide a “blanket of inhibition” [12] , a dense and nonspecific spread of inhibition on nearby PCs . Whether Martinotti cells are capable of generating such indiscriminate inhibition when firing simultaneously in large groups has not been tested . Martinotti cells that reside in the main cortical output L5 provide frequency-dependent disynaptic inhibition ( FDDI ) on neighboring PCs [13 , 14] , an inhibitory mechanism that synchronizes two or more PCs by one or a few Martinotti cells [15] . Synchronized activities in the cortex have been reported in vivo [16] as well as in vitro , where slow oscillations appear to be initiated in L5 [17] . Moreover , computational studies suggest that Martinotti cell activity can synchronize L5 PC spiking through distal inhibition [18]; however , this has not been tested experimentally . It is intriguing that distal dendrite–targeting interneurons , generating attenuated inhibitory currents , can affect PC spike time output . Here , we genetically targeted the L5 Martinotti cell population using a nicotinic acetylcholine receptor α2 subunit ( Chrna2 ) -Cre mouse line to investigate how Martinotti cell inhibition can synchronize L5 PC firing . Our results show that Chrna2-Cre–labeled L5 Martinotti cells were preferentially and reciprocally connected with thick-tufted PCs . Furthermore , we found that short burst firing of L5 Martinotti cells was able to reset L5 PC spiking and that controlling Martinotti cell activity to rapid bursts repeated in a slow rhythm was the most efficient inhibition to synchronize unconnected PCs . Finally , we show that L5 PC microcircuits could synchronize their own action potentials ( APs ) when coupled by L5 Martinotti cells and that inhibition was crucial for PC synchronization over prolonged periods .
To test whether Chrna2 can be used as a marker of Martinotti cells , we crossed Chrna2-Cre mice with a tdTomato reporter line ( R26tom , Fig 1 and S1 Fig ) [19 , 20] . L5 Chrna2-Cre/R26tom cells were found in all cortices ( S1A–S1C Fig , S1 Movie ) . Only very few Chrna2-Cre/R26tom cells were detected in layer 2/3 ( supragranular: 27 [2 . 4%] versus infragranular: 1102 [97 . 6%] cells in an 800-μm-thick section ) , suggesting that Chrna2-Cre/R26tom specifically labels L5 neurons . Reconstructions of patched biocytin-filled Chrna2-Cre/R26tom cells showed that 36 out of 37 ( 97 . 3% ) cells met the criteria of deep layer Martinotti cells by having an ovoid cell body in L5 , bipolar dendritic morphology , axons emerging from the main dendrite , proximal axonal arborizations , and long axonal projections to layer 1 with a dense arborization around PC distal dendrites ( Fig 1B–1F and S2 Fig ) [2 , 3 , 10 , 13] . The excluded cell had its cell body outside of L5 . At higher magnifications , the axonal plexus of L1 is highlighted by red fluorescent signal ( tdTomato ) of Chrna2-Cre/R26tom cell axonal ramifications ( see star in Fig 1E , 1F and S1C , S1D and S2B Figs ) . Immunohistochemistry revealed that 30 . 3% of Chrna2-Cre/R26tom cells were SOM+ ( n = 3 mice , 2–3 mo old , 8 sections of 35-μm thickness , of a total of 792 cells ( L1–L6 ) ; 297 cells were Chrna2+ , 495 cells were SOM+ , and 90 of these were double labelled for both genetically expressed tdTomato and SOM antibodies; S1D and S1E Fig ) . Counting cells only in L5 , we found a total of 549 cells , of which 292 cells were Chrna2+ , 257 were SOM+ , and 85 of these were double labelled for both Chrna2 and SOM ( 29 . 1% of L5 Chrna2+ cells were SOM+; S1D and S1E Fig ) . Single-cell reverse transcription PCR ( RT-PCR ) of individually picked Chrna2-Cre/R26tom cells ( n = 7 cells , n = 2 animals ) found 6/7 collected neurons to be positive for Glutamate decarboxylase 1 ( GAD1; S1F Fig ) , while no cell was positive for the vesicular glutamate transporter subtype 1 or 2 , indicating an inhibitory nature of Chrna2+ neurons . Membrane properties of Chrna2-Cre/R26tom cells measured by whole-cell patch clamp revealed a mean input resistance of 337 . 28 ± 11 . 42 MΩ and a mean resting membrane potential of −63 . 69 ± 1 . 02 mV ( n = 36 cells , S1 Data ) . The first AP generated upon 500 ms depolarizing current injections with 1-pA increments ( on average , the first spike was reached in response to 18 . 11 ± 1 . 97 pA ) had a mean AP amplitude of 72 . 61 ± 2 . 20 mV , an AP threshold of −43 . 28 ± 0 . 53 mV , an AP half-width of 1 . 92 ± 0 . 08 ms , and a first spike latency of 254 . 75 ± 24 . 24 ms ( n = 36 cells , S1 Data ) . Afterhyperpolarizations ( AHPs ) measured after the first AP had a mean magnitude of −8 . 28 ± 0 . 72 mV with a gradual depolarization over repeated spikes , and each AHP displayed both a fast and a slow component ( Fig 1G , S1 Data ) [2] . Hyperpolarizing currents generated rebound afterdepolarizations ( ADPs ) and , on average , 2 . 50 ± 0 . 25 rebound APs ( at −80 pA , 500 ms ) upon termination of current steps but did not produce a sizable membrane “sag , ” suggesting that these cells have a minimal hyperpolarization-activated current ( Ih , Fig 1G , top , S1 Data ) . The Chrna2-Cre/R26tom cell firing frequency versus current relationship showed a linear increase of average firing rate with increasing current towards a frequency of 22 . 4 ± 2 . 49 Hz ( at 200 pA , n = 36 cells; Fig 1H , left , S1 Data ) , indicating that Chrna2-Cre/R26tom cells are slow spiking interneurons . The relationship between maximum frequency ( 52 . 87 ± 2 . 44 Hz , n = 36 cells , S1 Data ) and steady-state frequency ( 21 . 87 ± 1 . 02 Hz , n = 36 cells , S1 Data ) revealed a spike-frequency adaptation ratio ( see Materials and Methods ) of ~59% in response to a 200-pA step ( Fig 1G , bottom and Fig 1H , middle ) . The spike-frequency adaption shows how firing frequency of Chrna2-Cre/R26tom cells decreases as a function of time ( 20 . 16 ± 2 . 44 Hz at 416 ms , n = 36 cells; Fig 1H , right ) . In summary , both morphological and electrophysiological characteristics of infragranular Chrna2-Cre/R26tom cells ( ~97% ) are similar to those reported in previous studies of Martinotti cells [2 , 13 , 21] , thus we hereafter refer to these L5-specific Chrna2-Martinotti cells as MCsα2 . Previous studies have speculated that SOM+ cells ( including Martinotti cells ) predominantly contact specific subpopulations of PCs [21 , 22] . Thus , we patched pairs of a MCα2 and its neighboring PC in L5 ( ≤60 μm ) . Next , we categorized PCs into type A and type B cells based on morphological and electrophysiological criteria [23] . Cells with a large cell body , thick-tufted basal dendrites with apical dendrites extensively branching in layer 1 , burst-regular spiking , responding with large AHPs , prominent hyperpolarization sags , and pronounced rebound ADP were classified as type A PCs ( Fig 2A , left ) . Cells with small soma , thin-tufted basal dendrites with limited spreading apical dendrites , absence of AHP or ADP , and small hyperpolarization sags were classified as type B PCs ( Fig 2A , right ) . We expected L5-specific MCsα2 to be locally connected to PCs [2 , 13] and found , amongst morphologically reconstructed pairs of patched cells , that 77% of type A PCs–MCsα2 were connected ( n = 7/9 ) , while none of the patched type B PCs–MCsα2 were connected ( n = 0/9 ) . Out of the paired MCα2–type A PC recordings , 55% ( n = 5/9 ) of pairs were reciprocally connected . Paired recordings of PCs and MCsα2 revealed that high-frequency stimulation ( 70 Hz ) of type A PCs generated excitatory postsynaptic potentials ( EPSPs ) , or an occasional spike , in MCsα2 ( example shows 12 repetitions from one patched type A PC–MCα2 pair; Fig 2B , left ) , whereas type B PC stimulation did not result in EPSPs in local MCsα2 ( 12 repetitions; Fig 2B , right ) . Additionally , inhibitory postsynaptic potentials ( IPSPs ) were generated in type A PCs ( IPSP amplitude: −1 . 08 ± 0 . 12 mV; example shows 12 repetitions from one patched MCα2–type A PC pair; Fig 2C , left ) after MCα2 stimulation , whereas no inhibitory responses were observed in type B PCs ( 12 repetitions; Fig 2C , right ) . Thus , our data suggest that MCsα2 connect with type A PCs and not type B PCs . We next investigated the influence of MC inhibition on PCs when simultaneously activating a large group of MCsα2 in Chrna2-Cre mice ( 1–2 mo old ) previously injected with floxed Channelrhodopsin-2 ( ChR2; Fig 3A ) . Compared to electrical stimulation of single MCsα2 , light activation of MCα2 groups produced IPSPs in type A PCs with a higher mean amplitude ( from −0 . 96 ± 0 . 05 mV to −1 . 41 ± 0 . 04 mV ) , a smaller mean time to peak ( from 29 . 53 ± 1 . 24 ms to 20 . 54 ± 0 . 97 ms ) , and a decreased mean half decay time ( from 63 . 46 ± 2 . 04 ms 51 . 20 ± 2 . 08 ms [all comparisons: n = 12 cells; a total of 54 IPSPs , p < 0 . 001 , Fig 3B , left and S3A Fig , S2 Data] ) . We also recorded from type B PCs ( n = 12 cells ) , but no IPSPs were observed in type B PCs in response to blue light stimulation of ChR2+ MCsα2 ( Fig 3B , right ) . We next tested different stimulation frequencies ( 2 , 5 , 15 , 25 , 40 , and 70 Hz ) [13 , 18] for ChR2+ MCsα2 to investigate the role of MCα2 firing frequency on IPSP amplitude in type A PCs ( n = 12 cells; Fig 3C ) . We found a nonlinear relationship between IPSP amplitude and MCα2 stimulation frequency in which , at higher frequencies ( >15 Hz ) , IPSPs summed into smooth compound IPSPs , probably due to the depressing synaptic properties of the MCα2-to-PC connection [13] . To further characterize the frequency-dependency of the MCα2–PC IPSPs , we stimulated MCsα2 with continuous light , which generated accommodating firing in MCsα2 ( Fig 3D , bottom and S3B–S3D Fig ) but still large IPSPs in PCs ( n = 12 cells , Fig 3D and 3E , S3 Data ) . This suggests that large compound IPSPs can be generated in type A PCs when MCsα2 fire at high frequencies and that the compound IPSP amplitude mainly depends on the firing frequency during the first 300 ms ( Fig 3D and 3E ) . Martinotti cells have been shown to provide FDDI [13 , 24 , 15] onto PCs . To examine whether L5-specific MCsα2 generate FDDI , we patched pairs of type A PCs with cell bodies next to MCsα2 and provided high-frequency ( 70 Hz ) current injection to one PC ( Fig 4A , top ) . This led to an early EPSP in the other PC , presumably due to monosynaptic PC–PC connections , followed by a delayed inhibition ( FDDI , amplitude: −1 . 02 ± 0 . 23 mV , time to peak: 57 . 16 ± 2 . 43 ms , half decay time: 95 . 34 ± 5 . 64 ms , n = 12 cells; Fig 4A , bottom , S4 Data ) . In some cases , patched type A PC pairs were not monosynaptically connected and only the FDDI was observed ( amplitude: −1 . 05 ± 0 . 21 mV , time to peak: 91 . 33 ± 2 . 83 ms , half decay time: 123 . 34 ± 5 . 87 ms , n = 12 cells; Fig 4B , top and Fig 4C , S4 Data ) . To confirm that the FDDI response was mediated by MCsα2 , we silenced MCsα2 in slices from Chrna2-Cre/Halorhodopsin ( HaloR ) -floxed mice ( 1–2 mo old ) with green light . Indeed , green light abolished the delayed inhibition ( n = 12 cells; Fig 4B , bottom and Fig 4C , S4 Data ) . However , we also noted that large IPSPs occurred in both patched type A PCs upon termination of the light pulse ( amplitude: −1 . 66 ± 0 . 16 mV , time to peak: 53 . 45 ± 2 . 23 ms , half decay time: 88 . 45 ± 4 . 53 ms , n = 24 IPSPs , all comparisons: p < 0 . 001; Fig 4B , bottom and Fig 4C , S4 Data ) . Current clamp recordings from MCsα2 showed that the green light generated strong hyperpolarization of MCsα2 and that , subsequently , bursts of rebound spikes were generated in HaloR-expressing MCsα2 upon light termination ( Fig 4D , top and S4A and S4B Fig , S5 Data ) . HaloR-activation for 500 ms consistently evoked one or more rebound APs with varying frequency in MCsα2 that could not be blocked by the Ih blocker ZD7288 ( 20 μM , S4C Fig ) , similar to Ih-independent rebound APs in distal dendrite–targeting X98 cells [9] . Additionally , MCsα2 and type A PCs were patched in the presence of carbachol ( 10 μM ) to further examine how MCsα2 could modulate L5 PC spontaneous firing . Carbachol depolarized PCs and MCsα2 by 10 to 15 mV and did not result in a specific oscillatory frequency as seen with high concentrations of carbachol; instead , it showed broad peaks in the power spectral density plots ( S5 Fig ) . Two firing patterns could be distinguished for type A PCs [25–28]: single-spiking ( n = 42 cells; Fig 4D , middle ) and burst-spiking ( n = 18 cells; Fig 4D , middle ) . Independent of the type A PC firing type , bursts of HaloR-induced rebound spikes from MCsα2 caused large , compound IPSPs ( S6 Fig , S6 Data ) that resulted in a resetting of type A PC firing ( Fig 4D , middle ) . We define resetting as temporally aligning spiking after a period of inhibition . APs from type A PCs aligned around 500 ms ( single-spiking: 556 . 19 ± 9 . 42 ms , n = 42 cells; burst-spiking: 524 . 72 ± 10 . 17 ms , n = 18 cells , S4 Data ) after light-off for HaloR-inhibition of MCsα2 . The shape of the summed type A PC IPSP trace corresponded well to the first 3–4 rebound APs of MCsα2 ( S6A Fig ) and steadily evoked subsequent post-inhibitory rebound APs in the type A PCs ( n = 60 cells; Fig 4D , middle ) but not type B PCs ( n = 12 cells; Fig 4D , bottom ) . Note that the IPSP amplitude more than doubled with carbachol present ( Vm = −60 mV; IPSP amplitude = −1 . 66 ± 0 . 26 mV , Fig 4B , bottom; Vm = −48 mV [single-spiking PC]; IPSP amplitude = −3 . 51 ± 0 . 48; Vm = −48 mV [burst-spiking PC]; IPSP amplitude = −6 . 30 ± 0 . 91 mV , all comparisons: p < 0 . 0001; S6B and S6C Fig , S6 Data ) . Voltage clamp experiments ( holding at −60 mV ) further highlighted the presence of large IPSCs in type A PCs ( 107 . 40 ± 3 . 54 pA , n = 3 cells; Fig 4E , top , S4 Data ) and the absence of IPSCs in type B PCs ( n = 3 cells; Fig 4E , bottom ) . Together , these results show that a rapid burst of MCα2 APs can abruptly halt the firing of type A PCs and also reset PC firing by temporal coupling rebound APs of PCs while leaving type B PC firing unaffected . Therefore , we only aimed for type A PCs for the remainder of the study . It is unknown how PCs synchronize their firing , although computational studies have suggested a role for distal dendrite–targeting interneurons in synchronizations [11 , 18 , 29] . Thus , we only patched unconnected type A PCs and recorded spontaneous firing , in the presence of carbachol , from two PCs simultaneously ( n = 24 cells ) while optogenetically stimulating the MCα2 population at various frequencies ( 2 , 5 , 15 , 25 , 40 , and 70 Hz ) . To identify the frequency of MCα2 activity that best temporally aligns unconnected , randomly firing , type A PCs , we recorded repeats of 4-s sweeps ( 2 s light-off , 2 s light-on ) . When pairwise superimposing simultaneous recordings from two type A PCs , we observed that light stimulation of 2 Hz or 15 Hz ( n = 24 cells; 12 black and 12 grey PC spike trains; Fig 5A ) created a rhythmical firing pattern of type A PCs highlighted by kernel density estimates , which show the distribution of APs over time ( orange traces ) . Although mean power spectral density plots from both 2 Hz and 15 Hz MCα2 stimulation revealed peaks around 2 Hz ( 1 . 99 ± 0 . 09 versus 1 . 87 ± 0 . 14 Hz , n = 24 cells ) , other frequencies tested ( 5 , 25 , 40 , and 70 Hz ) did not result in any clear peaks ( Fig 5B , top and S7 Fig , S7 Data ) . This indicates that MCsα2 preferentially give rise to slow frequencies in a group of type A PCs . However , flat mean coherence plots of pairwise analyzed PCs did not show any correlation between simultaneously recorded type A PCs in specific frequency bands plotted up to 20 Hz . This suggests that type A PCs as a population can produce an oscillatory firing rhythm , but individual cells are mostly out of phase and not synchronized with each other ( n = 24 cells; Fig 5B , bottom ) . To generate complete synchrony between type A PCs , we hypothesized that a rapid burst of MCα2 activity could reset/align type A PC spiking ( 3–4 APs as seen in Fig 4D and S6A Fig ) and a slow rhythm could maintain in-phase synchronous firing . To test this , we patched two unconnected type A PCs and stimulated MCsα2 with 15-Hz bursts every 500 ms ( 2 Hz ) . This stimulation protocol resulted in high AP synchronization ( Fig 5C ) directly after MCsα2 were paused . Mean power spectral density ( peak at 2 . 02 ± 0 . 04 Hz , n = 24 cells ) and mean coherence examination showed that type A PCs followed MCα2 stimulation frequency and were pairwise aligned in that frequency ( Fig 5D , S7 Data ) , suggesting synchronized firing of type A PCs at slow frequencies . This shows that MCsα2 have means to both initiate and maintain prolonged type A PC synchronous firing . Next , we sought to quantify the synchrony ( provide a synchrony index [30] ) between type A PC firing when MCsα2 were activated in bursts of 15 Hz . A representative recording of two simultaneously captured type A PCs is shown in Fig 6A , where initially unsynchronized type A PCs aligned during MCα2 stimulation ( orange rectangles highlight synchronized APs ) . In the absence of MCα2 stimulation , the mean cross-correlograms of pairwise analyzed recordings showed only low magnitude peaks , while light stimulation organized firing of both type A PCs in cohorts every 500 ms ( n = 24 cells; Fig 6B ) . A 3-fold increase in the synchrony index could be extracted from the cross-correlograms when MCsα2 were light stimulated ( control: 0 . 21 ± 0 . 03 , burst stimulation: 0 . 61 ± 0 . 04 , n = 12 dual recordings , n = 24 cells , p < 0 . 0001; Fig 6C , S8 Data ) . Thus , we conclude that delivery of MCα2 inhibition in bursts of 15 Hz indeed synchronized type A PC firing . This is likely due to burst firing repeated in slow frequency , causing inhibition in type A PCs with little depression compared to continuous 15 Hz stimulation of MCsα2 showing apparent synaptic depression ( examples in grey , mean in black , red dashed lines for visual guidance , Fig 6D ) . Interestingly , the 15-Hz continuous light stimulation revealed that bursting PCs can switch firing patterns from burst-spiking into single-spiking [25 , 18] during continuous low-magnitude inhibition ( S8A Fig ) . This change in firing was observed only at near-threshold potentials; the physiological role remains to be studied ( S8B Fig ) . To test if type A PC circuits can self-synchronize their firing through Martinotti cell activation , we designed a closed-loop system ( optical feedback inhibition [31] ) for paired recordings that delivered four blue light pulses ( 15 Hz ) to the MCsα2 when one PC fired in the presence of carbachol ( n = 24 cells; 12 black and 12 grey PC spike trains; Fig 7A , inset ) in tissue from Chrna2-Cre mice previously injected with floxed ChR2 . Kernel density estimates showed increased co-occurrences of type A PC APs during optical feedback inhibition ( Fig 7A ) . Moreover , because these experiments involved a leading and a following type A PC , it was possible to calculate the statistical dependency between the spike trains of two PCs and to express this with a mutual information index ( see Materials and Methods ) . This index gives an estimate of how well one signal can predict the other and is helpful to interpret to what extent one PC can drive another , e . g . , via recurrent or feedforward inhibition . In controls , the mutual information index was low ( 2 ± 1 , n = 12 dual recordings ) , whereas turning on the optical feedback inhibition resulted in immediate auto-alignment of type A PC APs with an increased index ( 11 ± 5 , n = 12 dual recordings , p < 0 . 05; Fig 7B , S9 Data ) . Venn diagrams show the mutual information as the degree of overlap between two circles , representing each PC train as entropy ( Fig 7B , inset ) . The overlap demonstrates the predictive value ( mutual dependency ) between a known PC train and a following PC train . When shifting one spike train relative to the other , the incremental mutual information index plot ( mutual information index as a function of time lag , Fig 7B ) showed that the mutual dependency was largest around 0-ms lag , suggesting high synchronization of the two PCs directly when coupled by optical feedback inhibition . Peaks around ± 400–600 ms indicate that this activity-dependent inhibition causes repeated synchronization every 400–600 ms .
We found that MCsα2 were exclusively synaptically connected to large , thick-tufted PCs , often referred to as L5B PCs [32] or type A PCs [23] . Different PC morphologies seem to be associated with different connectivity patterns in the brain , e . g . , large , thick-tufted PCs are usually synonymously named subcerebral projection neurons or pyramidal tract neurons , whereas thin-tufted PCs , or type B PCs [23] are callosal projection neurons or intratelencephalic neurons [26–28] . Our in vitro preparation could not define PCs according to connectivity patterns; however , based on the extensive branching of the distal dendrites and the large triangular-shaped cell bodies , we find it likely that the type A PCs correspond to the thick-tufted PCs [22] and are probably subcortically projecting [26–28] . Thick-tufted type A PCs can further be described by firing properties as single-spiking or burst-spiking [25–28] . Typically , at near-threshold potentials , bursting cells respond with two or more bursts , of two or more APs , generated in quick succession with short interspike intervals [25] . Burst properties of PCs disappear with increasing current injections [25] and may be dependent on the size of the dendritic tree [33] . In addition to morphological variances , such as a smaller soma compared to type A PCs , type B PCs had characteristic electrophysiological differences . Our pair-recordings between type A PCs and MCsα2 confirmed that type A PCs provided facilitating synaptic responses in MCsα2 . We also found depressing synaptic connections from MCsα2 to type A PCs [13] , while no type B PC connectivity with MCsα2 was observed . Lack of IPSPs in type B PC was not likely due to shunting of inhibition , as voltage clamp recordings also failed to find synaptic connectivity between MCsα2 and type B PCs . However , due to difference in thin- versus thick-tufted morphology , it is possible that the internal solution creates less dialysis of the chloride ion Cl- concentration in type B PCs compared to type A PCs . Therefore , perforated patch recordings would be needed to firmly rule out the possibility of shunting of inhibition . Still , we found that FDDI , a Martinotti cell–dependent feature [13 , 24 , 15] was relayed by MCsα2 and consequently was specific for type A PCs . This is in agreement with a previous study showing FDDI between thick-tufted PCs but not between corticocallosally projecting cells [22] . Distal inhibition by individual MCs is important for shaping local dendritic voltage–activated responses . FDDI combined with dendritic depolarization has shown that MCs can attenuate back-propagating AP-activated Ca2+ spike firing and thereby reduce burst firing of PCs [34] . On the network level , collective and precisely timed Martinotti cell activity can further be potent enough to affect somatic spike generation . Here , we first used HaloR to examine if blocking MCα2 activity could eliminate the appearance of FDDI in thick-tufted PCs [13 , 24 , 15] . This led to the observation that on the termination of green light MCsα2 fired bursts of HaloR-induced rebound spikes , inhibiting PCs and subsequently causing hyperpolarization-induced rebound APs that could reset PC firing . The HaloR-induced rebound in MCsα2 is a methodological artifact and has little physiological relevance; however , it is interesting to speculate whether Martinotti cells receive inhibition that could generate rebound spikes . Recently , the vasoactive intestinal peptide ( VIP ) interneuron has been shown to densely inhibit Martinotti cells [35] . The high connection probabilities between VIP cells and Martinotti cells [36] suggest that VIP cells could provide strong hyperpolarization in Martinotti cells for the possible generation of rebound excitation . Rebound spikes have been previously demonstrated to occur in entorhinal cortex neurons in vivo and are attributed to play a role in generating grid cell fields that usually arises when grid cells fire synchronized [37 , 38] . A similar role could be applicable to rebound spikes in the neocortex , where a “blanket of inhibition” [12] evolves through the synchronized spread of inhibition , serving to coordinate PC firing . Thereby , VIP cells appear to make "holes in the blanket of inhibition" [35] by inhibiting Martinotti cells [35 , 36] . In other words , VIP cell activity might regionally disrupt coordinated PC firing while local Martinotti cell activity could reset and rescue PC synchronous firing . Second , focusing on the combined activity of ChR2-expressing MCsα2 , our data show that bursts of MCsα2 were able to reset type A PC firing and , if repeated , could synchronize PC activity . In a computer model , oscillatory inhibition of the distal PC dendrite at 10–20 Hz , presumably by LTS SOM+ Martinotti cells , was shown to control L5 PC firing [18] . Our findings support that 15 Hz firing of MCsα2 can align type A PC firing but also show that 15 Hz firing in short bursts more reliably synchronizes PCs compared to continuous 15 Hz firing . In other computational models , the importance of a beta rhythm in regulating gamma oscillations and intercortical signaling has been demonstrated and , furthermore , that the beta frequency is regulated by cholinergic modulators [11 , 29 , 39] . In this respect , the exclusive expression of the alpha 2 cholinergic receptors in MCsα2 is noteworthy and may suggest a specific role for MCsα2 in transmitting the modulatory action of cholinergic signaling . Cholinergic modulation of LTS cells has been suggested to generate beta oscillatory activity ( beta2 ) in L5 of the primary auditory cortex [40] . These oscillations were insensitive to the muscarinic antagonist atropine but sensitive to the nicotinic receptor antagonist d-Tubocurarine [40] . Thus , computational and experimental studies indicate that the beta rhythm is important for network properties [40 , 41]; however , beta activity in bursts repeated in slow frequency has not been reported previously . At this slow frequency MCα2–PC inhibition shows minimal depression , similar to the minimal depression of slow firing SOM+ interneurons defined by their green fluorescent protein expression in a transgenic mouse ( GIN-cells ) [42] , and therefore , a combination of rapid bursting and slow rhythmical inhibition seems most effective to synchronize PCs . Genetic targeting and optical feedback inhibition are a potent technique to study how PCs can drive a population of interneurons by their innate rhythm . A previous study used a closed-loop system to optogenetically produce feedback inhibition onto PCs from parvalbumin+ interneurons [31] . Sohal et al . used synthetic excitatory post synaptic currents ( EPSCs , dynamic clamp ) in a single PC triggering parvalbumin+ interneuron excitation with light [31] . Differently , in our study , we depolarized optically stimulated MCsα2 in the presence of carbachol and measured synchronization of simultaneously recorded PCs using mutual information . Analogously to our optogenetic stimulation , gap junctions could provide a physiological mechanism for the synchronization of interneuron populations [42–44] . Berger et al . have shown the existence of electrical coupling between L5 MCs [15] . It will be interesting in the future to explore the existence of gap junctions between MCsα2 . The Chrna2-Cre/R26tom mouse line simplifies identification and characterization of L5 Martinotti cells . MCsα2 are morphologically and electrophysiologically homogenous , further evincing the specificity of our marker . The dense axonal plexus observed in L1 and the near absence of Cre+ cell bodies in L2 ( 2 . 4% ) in Chrna2-Cre/R26tom mice also indicate that Cre+ cells are , in fact , L5 MCs . Still , we found a high proportion of Cre+ cells in Chrna2-Cre/R26tom mice that were not labelled with the antibody against SOM , and this could be due to extra-somatic location of the peptide . So far , SOM-Cre is the most widely used transgenic mouse line for targeting MCs together with the GIN mouse [8 , 45] , but still SOM-Cre has been shown to label all cell layers [46] . Here , we provide a layer-specific , single genetic marker for MCs across the cortex and confirmed their inhibitory nature using single cell RT-PCR . Although we did not explicitly block optogenetically evoked , inhibitory postsynaptic currents of MCsα2 ( e . g . , with Gabazine ) , Martinotti cell dendritic inhibition in vivo has been shown to be GABAA-mediated [34] . The specific expression of Chrna2 in inhibitory L5 MCsα2 raises questions of how important the α2 subunit is for cholinergic inputs . Several cortical interneurons express nicotinic acetylcholine receptors ( nAChRs ) [47–49] , suggesting cholinergic modulation of inhibition in the cortex , most likely from the basal forebrain [50] . Cortical LTS cells , such as Martinotti cells [51 , 52] , are excited by acetylcholine via nicotinic receptors and alter cortical circuit processing [53] . Cholinergic input is most likely mediated by additional nicotinic subunits that together form high affinity receptors for acetylcholine [54] . Several candidate subunits exist , but perhaps the more promising ones , judged from their specific expression in cortical L5 , response to nicotine , and known co-expression with α2 subunits , include α6-nAChRs , β2-nAChRs , and β4-nAChRs [55–57] . The focus of our work has been on the functionality of Martinotti cells , not the nAChR subunits; however , earlier studies of cholinergic subunits can provide potential clues to Martinotti cell function . A deletion of α2-nAChRs has shown a normal phenotype but altered responses during nicotine-associated behaviors [58] . Deletion of α2-nAChRs has also shown reduced nicotine-induced hippocampal LTP in the temperoammonic path , most likely via oriens-lacunosum moleculare ( OLM ) interneurons [59 , 19] . Interestingly , Chrna2 is expressed in OLM cells , which target the distal dendrites of hippocampal PCs in a comparable manner as MCsα2 target the distal dendrites of cortical PCs . In some similarity to the suggested role for LTS cells in directing the flow of information in the cortex [53] , OLM cells have been suggested to gate internal and external signals to the hippocampus [19] . In addition , MCsα2 might modulate cortical states , because SOM+ interneurons have recently been implied to be involved in transitions between UP and DOWN states [60] . Furthermore , studies of β2-nAChR KO mice have suggested a role for β2-containing nAChR in restricting cortical UP states and might be interesting for future studies on how nAChR are distributed in cortical interneurons such as Martinotti cells [13 , 61 , 62] . Our preparation did not examine cortical UP and DOWN states; instead , we depolarized neurons with a low concentration of the cholinergic agonist carbachol . As the IPSP amplitude generated by MCsα2 is dependent on the membrane potential of the postsynaptic PCs , this illustrates how MCα2 inhibition ( amplitude of IPSPs ) could alter in a state-dependent manner , thereby exerting a state-dependent modulation of PC excitability . In summary , we report the identification of a marker specific for L5 Martinotti cells projecting to layer 1 . These Martinotti cells were synaptically connected to large , thick-tufted PCs with prominent AHP and ADP , demonstrating a distinctive microcircuit between one type of interneuron and one subtype of PCs . Furthermore , we demonstrate that Martinotti cell–mediated inhibition can initiate and also maintain synchronous firing between PCs . We also show that this inhibition is frequency dependent and , when repeated in beta bursts , can continuously align firing of PCs . Lastly , using a closed-loop system in which PCs auto-synchronized their firing , we show that Martinotti cells were able to bridge the communication between unconnected PCs via activity-dependent inhibition . Thus , via their feedback and feedforward connections , Martinotti cells are important for regulating thick-tufted type A PC output in L5 , most likely altering voltage-dependent dendritic properties and actively influencing somatic spike generation and synchronization .
All experiments were approved by the Swedish Animal Welfare authorities and followed Uppsala University guidelines for the care and usage of laboratory animals ( ethics permits C132/13 and C135/14 ) . Efforts were made to minimize the numbers of animals used . In this study , we used transgenic mice ( both males and females ) , with Chrna2-Cre [19 , 20] that were crossed with a tdTomato fluorescence reporter line Gt ( ROSA ) 26Sortm14 ( CAG-tdTomato ) Hze ( R26tom; Allen Brain Institute ) or with a HaloR-expressing line Rosa26-eNphR-EYFP ( HaloR; Jackson Laboratory Stock No . 014539 ) . Cre-negative littermates were routinely used as controls . The CLARITY procedure followed a standard protocol [63] . In summary , 2–3-mo-old Chrna2-Cre/R26tom mice ( n = 2 ) were transcardially perfused with 20 ml of ice-cold 1x PBS solution followed by 20 ml of a hydrogel monomer solution consisting of 4% acrylamide , 0 . 05% bis-acrylamide , 0 . 25% VA-044 initiator , and 4% paraformaldehyde in PBS . Brains were quickly dissected and placed in hydrogel monomer solution for 3 d at 4°C . Prior to polymerization of the hydrogel monomer solution , samples were placed in a desiccation chamber attached to a vacuum pump . With the sample lid ajar , air was removed from the chamber for 10 min and replaced with nitrogen gas , after which the sample lid was tightly shut . The hydrogel monomer solution was polymerized by heating the samples to 37°C for 3 h in a water bath whilst shaking . Embedded tissue was extracted from the gel , and brains were sliced to 3-mm coronal sections using a brain matrix . Passive clearing of slices was achieved by repeated , 3-d washes in a 4% Sodium Dodecyl Sulphate ( SDS ) sodium borate buffer ( 200 mM , pH 8 . 5 ) solution at 45°C on a shaker plate for 6 wk . SDS was removed from the samples by incubating in PBST0 . 1 ( 1x PBS and 0 . 1% Triton X-100 ) on a shaker plate for two consecutive 1-d washes . Clear tissue was refractive index-matched through serial , 1-d incubations in 20% , 40% , and 63% 2 , 2′-Thiodiethanol ( TDE , Sigma-Aldrich ) in 1x PBS solution . Light sheet fluorescence images were acquired using Zeiss light sheet Z1 with a 5x/0 . 16 objective . Individual image tiles were 3D-stitched using Arivis Vision4D ( Arivis ) . Imaris 8 . 1 ( Bitplane ) was used for analysis , volume rendering , soma detection , and data visualization . Chrna2-Cre/R26tom cells were counted using Imaris 8 . 1 ( Bitplane ) and Matlab ( version 2013a , MathWorks ) in an 800-μM section ( AP: −2 . 40 to −3 . 20 mm , ML: 2 . 00 to 5 . 00 mm , and DV: 0 . 50 to 3 . 50 mm ) . Cells were divided into infragranular ( roughly corresponding to L5/6 ) and supragranular ( roughly corresponding to L2/3 ) cells by fitting a curve between ML: 4 . 25 mm , DV: 3 . 50 mm and ML: 2 . 00 mm , DV: 0 . 75 mm , roughly dorsolateral to L5 . Chrna2-Cre/R26tom mice ( 2–3 mo , n = 3 ) were anesthetized with isoflurane and decapitated before dissection . Immunohistochemistry ( IHC ) was performed as previously described in [19] . The following dilutions of antibodies were used: SOM antibody 1:150 ( MAB354 , Anti-SOM Antibody , clone YC7 [Merck Millipore Corporation] ) , Anti-rat Cy5 , 1:500 ( Invitrogen ) . Following whole-cell recordings , we used strong negative pressure to suck the cytoplasm and organelles of the cells into the recording pipette tip , similar to [19] . Buffers and cDNA conversion are further described in [19] . A two-round PCR ( nested ) to detect GAD1 , Vglut1 , or Vglut2 cDNA was done . The following mix was used for the first and second round of PCR: 1 . 5 mM MgCl2 , 10 pmol of each primer , 1 . 0 U of platinum Taq-DNA polymerase ( Invitrogen ) , 20 mM Tris·HCl , and 50 mM KCl pH 8 . 4; thermal cycle: 94°C/2-min denaturation step followed by 35 cycles of 94°C/50 s , 55°C/45 s , and 72°C/45 s . In the second round , instead of mixing the original template , 10% of the first PCR reaction as template was used . Second-round PCR products were visualized on 2% agarose gels . Primers were designed based upon sequences deposited in the GenBank database ( www . ncbi . nlm . nih . gov/nucleotide ) . The primers used were as follows: first round: GAD1: CCAATAGCCTGGAAGAGAAGAG ( forward ) , TCCCATCACCATCTTTATTTGA ( reverse ) ; Vglut1: CGCTACATCATCGCCATCATGAG ( forward ) , GGAGGGGCCCATTTGCTCCA ( reverse ) ; Vglut2: GCCGCTACATCATAGCCATC ( forward ) , GCTCTCTCCAATGCTCTCCTC ( reverse ) ; second round: GAD1: CCAATAGCCTGGAAGAGAAGAG ( forward ) , TCCCATCACCATCTTTATTTGA ( reverse ) ; Vglut1: CTGGAGGATTTATCTGCCAAAAAT ( forward ) , GGTATGTGACCCCCTCCACCAAT ( reverse ) ; Vglut2: ACATGGTCAACAACAGCACTATC ( forward ) , ATAAGACACCAGAAGCCAGAACA ( reverse ) . Coronal slices from Chrna2-Cre/R26tom transgenic mice ( P19–29 , n = 12 ) were obtained similar to [24] . In summary , brains were rapidly removed and placed in ice-cold sucrose/artificial cerebrospinal fluid ( ACSF ) consisting of the following ( in mM ) : KCl , 2 . 49; NaH2PO4 , 1 . 43; NaHCO3 , 26; glucose , 10; sucrose , 252; CaCl2 , 1; MgCl2 , 4 . Coronal 300-μm-thick slices containing the primary auditory cortex were cut using a vibratome ( VT1200 , Leica , Microsystems ) and were subsequently moved to a submerged holding chamber containing normal ACSF ( in mM ) : NaCl , 124; KCl , 3 . 5; NaH2PO4 , 1 . 25; MgCl2 , 1 . 5; CaCl2 , 1 . 5; NaHCO3 , 30; glucose , 10 , constantly bubbled with 95% O2 and 5% CO2 and kept at 35°C for 1 h then maintained at room temperature . The slices were transferred to submerged chamber under an upright microscope equipped with DIC optics ( Olympus ) and perfused with oxygenated ASCF ( 1–1 . 25 ml/min ) at 30°C . For some experiments , carbachol ( 10 μM , Sigma-Aldrich ) and ZD7288 ( 20 μM , Tocris Cookson Inc . ) were added to the perfusate . Patch pipettes from borosilicate glass capillaries ( GC150F-10 Harvard Apparatus ) were pulled on a vertical puller ( Narishige , Japan ) with resistance around 7 MΩ . Pipettes were filled with internal solution containing ( in mM ) the following: K-gluconate , 130; NaCl , 7; MgCl2 , 2; ATP , 2; GTP , 0 . 5; HEPES , 10; EGTA , 0 . 1 ( pH was adjusted to 7 . 2 using KOH ) . Whole-cell current clamp recordings were acquired using a Multiclamp 700B amplifier ( Axon Instruments , CA , USA ) and digitized with a Digidata 1440A data acquisition card ( Axon Instruments , CA , USA ) . WinWCP and WinEDR softwares implemented by Dr . J . Dempster ( University of Strathclyde , Glasgow , UK ) were used to record electrophysiological signals . Patch-clamp data were analyzed with custom routines in MATLAB . APs were triggered by 500-ms depolarizing current injections from 10–100 pA . The first fired AP in response to minimal current injection was analyzed for AP amplitude ( peak to AHP voltage ) , threshold ( where the change in membrane potential exceeds 20 mV/ms ) , half-width ( halfway between threshold voltage and peak ) , and first spike latency ( time between stimulus onset and the AP threshold of the first spike ) . AHPs were analyzed for magnitude ( AP threshold—minimum of voltage trough between the first and the second AP in a spike train ) . Spike rate was calculated as the number of APs per 1 , 000 ms . Spike-frequency adaptation was measured as the inverse of the mean of the last three interspike intervals ( steady-state frequency ) divided by the inverse of the first interspike interval ( maximum frequency ) in response to 100-pA current injections and subtracted from 100% ( no adaptation ) . In spike-frequency adaptation plots , the reciprocal of consecutive interspike intervals is shown for each AP versus the time after onset of the current pulse . ChR2 stimulation frequencies ( 2 , 5 , 15 , 25 , 40 , and 70 Hz ) and HaloR-evoked hyperpolarization ( 5 , 10 , 100 , 250 , 500 , and 1000 ms ) were applied in randomized order to avoid statistical dependencies between cases . ChR2-triggered IPSPs and FDDI were measured as amplitude , time to peak , and half decay time , for which the onset was defined as the time at which the potential exceeded three times the standard deviation of the preceding baseline . HaloR-evoked hyperpolarization amplitudes were quantified as the difference between resting membrane potential and the peak of hyperpolarization . Rebound APs were quantified by number , maximum frequency , and duration ( time of last minus time of first rebound AP ) . Burst-spiking PCs were distinguished from single-spiking PCs by obtaining the interspike intervals , at which burst-spiking PCs showed an increased amount of short ( ≤20 ms ) interspike intervals at near-threshold potentials . MCsα2 were identified by cortical tdTomato-expression in Chrna2-Cre/R26tom mice that were perfused as previously described [19] , and 20-μm- and 60-μm-thick coronal slices were imaged using a Zeiss LSM 510 Meta confocal microscope . Cells targeted by electrophysiology experiments were routinely filled with biocytin and stained with streptavidin-488 nm for post-hoc analysis . Images were collected on a Zeiss LSM 510 Meta confocal microscope and stacked and 2D-stichted using ImageJ 1 . 50a ( NIH ) , where the color palette was adjusted for consistency ( tdTomato—red , biocytin/ChR2—green ) . Soma detection and Neurite tracings were done semi-automated with NeuronJ 1 . 4 . 2 ( ImageJ Plugin ) or fully automated with Imaris 8 . 1 FilamentTracer ( Bitplane ) using “Autopath” in the algorithm settings and the threshold mechanism to correct for over-/under-sampled tracings following the image intensity . Chrna2-Cre/R26tom mice ( 1–2 mo old , n = 33 ) were anesthetized with Isofluran ( 1%–4% ) and placed on a heat pad , with the head fixed with a nose holder and ear bars in a stereotaxic frame ( Stoelting Co . ) . The skin was cleaned with iodine and opened with a straight incision , and the bregma was identified using small amount of peroxide . The coordinates for bilateral virus injection were as follows: AP: −2 . 46 mm , ML: +/−4 . 00 mm , and DV: 2 . 00/2 . 50 mm . We used bilateral injections to obtain the maximum number of slices containing the primary auditory cortex , with preserved dendritic trees of the type A PCs in layer 1 ( on average two slices , 300 μm thick , per hemisphere ) per animal . A small hole was drilled in the skull using a dental micro drill , causing minimal bleeding during the process . Viral vectors ( pAAV-EF1a-double floxed-hChR2 ( H134R ) -EYFP-WPRE-HGHpA , University of Pennsylvania Vector Core Facility ) in solution ( 6 . 2 x1012 / 1 . 6×1013 particles/ml ) of 0 . 50–1 . 00 μl were slowly infused ( 0 . 10 μl/min ) into the auditory cortex at two depths ( 2 . 00 and 2 . 50 mm ) using a Hamilton 10-μl syringe and an infusion/withdraw pump ( World Precision Instruments/KD Scientific ) . After infusion , the needle was left in place for 1–5 min to allow complete diffusion of the virus . Next , the scalp was rehydrated with saline and sutured with 4–5 stitches and local anesthesia ( a drop of Lidocaine/Marcaine ) applied onto the sutured skin before the mouse was allowed to wake up . The animal was monitored and kept warm until fully awake ( moving and starting to eat and drink water ) . Mice were killed after approximately 3–6 wk for in vitro electrophysiological experiments and/or histological procedures . We modified our dynamic clamp system [30] running the Real Time Application Interface for Linux-based ( RTAI ) from the Politecnico di Milano Institute-Dipartimento di Ingegneria Aerospaziale ( Mantegazza , http://www . rtai . org/ ) on a Dell Precision T1500 with a Quad-Core Intel Core i7 with 2 . 80 Ghz , 5 . 8 Gigabyte memory , and a National Instruments DAQ card ( NI PCI-6251 ) . Routines for data acquisition were programmed in GNU-C using the Linux Control and Measurement Device Interface ( COMEDI ) . The membrane potentials of the patched cells were acquired in 20 kHz , and APs were detected based on threshold ( >−20 mV ) in real time , triggering the LED ( CoolLED pE-1 ) via 3-ms ( ChR2 experiments ) or 500-ms ( HaloR experiments ) TTL pulses . Matlab ( version 2013a , MathWorks ) was used for data analysis . APs from MCsα2 and PCs were detected based on threshold ( >−20 mV ) . PC spike trains were transformed into a series of 0 ( no spike ) and 1 ( spike ) , with 0 . 1-ms precision ( binning ) . Accordingly , kernel density estimates are probability densities and were computed on population data ( all patched PCs ) with 0 . 1 “bandwidth” ( “ksdensity” command in Matlab ) . The kernel density estimates show the distribution of APs over time and highlight increased ( peaks ) and decreased ( valleys ) co-occurrences of spikes . Power spectral density analysis of the binary spike series was made using Welch’s method ( “pwelch” command in Matlab ) to find the frequency components with highest power . Coherence was calculated pairwise ( i . e . , for simultaneously patched PCs , the similarity between the binary sequences of the two PCs was calculated ) and plotted as mean over all recordings ( “cohere” command in Matlab ) to investigate the dependence of two cells as a function of frequency . Cross-correlograms were calculated using the “coeff” option of the cross-correlation command in Matlab ( to scale the cross-correlation values from −1 to 1 and prevent dependency of the cross-correlation on the number of spikes ) and then smoothed by a moving average filter with a span of 10 ms to find the functional dependence between the APs of simultaneously recorded cells over time . We displayed cross-correlations over a lag range of ±1 s . The synchrony index was defined as the maximum peak of the normalized cross-correlograms between −50 ms and 50 ms , as previously described [30] , with 0 ≙ no synchronization and 1 ≙ full synchronization . To investigate statistical dependence , mutual information MI ( PCleft , PCright ) was determined on the binary spike data ( binning = 20 ms ) by calculating the distributions of the spike trains ( univariate distribution of each spike train separately as well as bivariate distribution of both spike trains ) and expressing them as entropies H ( PCleft ) and H ( PCright ) , meaning how “diverse” the spike trains were . The sum of the two entropies H ( PCleft ) and H ( PCright ) minus the joint entropy H ( PCleft , PCright ) quantified the conditional entropy , i . e . , the mutual information MI ( PCleft , PCright ) . The MI was formulated as a mutual information index with a scaling factor of 1 , 000 [30] . All statistical analysis was performed using R version 3 . 2 . 3 ( Foundation for Statistical Computing , Vienna , Austria ) . Data are reported as mean ± standard error of the mean ( SEM ) and plotted as bar plots or box plots . Data larger than q3 + 1 . 5* ( q3–q1 ) or smaller than q1–1 . 5* ( q3–q1 ) , with q1 and q3 denoting the 25th and 75th percentiles ( see box plots ) , were considered as outlier and discarded . Statistical comparisons were determined using two-tailed Student’s paired t test , and to account for multiple comparisons , the data were analyzed using ANOVA and post-hoc test with Tukey correction ( * ≙ p < 0 . 05 , ** ≙ p < 0 . 01 , *** ≙ p < 0 . 001 , and **** ≙ p < 0 . 0001 ) . The order of stimulations of different frequencies ( e . g . 2 , 5 , 15 , 25 , 40 , 70 Hz ) was systematically varied to avoid statistical dependencies between the timing of recordings and the frequency investigated .
|
Cognitive functions and information processing are linked to the coordination of neuronal events and activities . This coordination is achieved through the synchronization of neuronal signals within subnetworks . Local networks contain different types of nerve cells , each of them playing distinct roles in the synchronization mechanism . To understand how synchronization is initiated and maintained , we have identified one of the key players using genetic strategies; we have identified a subtype of nicotine receptors uniquely expressed in cortical Martinotti cells . Because of their architecture and connection properties , Martinotti cells are able to synchronize ongoing activity of unconnected pyramidal cells ( PCs ) . We show that this mechanism only applies to one subtype of PCs , thereby demonstrating that Martinotti cell inhibition is not spread randomly . By testing optimal firing patterns of Martinotti cells , we are able to coordinate the firing of this specific PC subtype over longer periods of time , showing how one unique interneuron is contributing to information processing .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"carbachol",
"action",
"potentials",
"medicine",
"and",
"health",
"sciences",
"drugs",
"membrane",
"potential",
"electrophysiology",
"light",
"neuroscience",
"electromagnetic",
"radiation",
"interneurons",
"pharmacology",
"neuronal",
"dendrites",
"nicotinic",
"acetylcholine",
"receptors",
"animal",
"cells",
"proteins",
"transmembrane",
"receptors",
"light",
"pulses",
"hyperpolarization",
"physics",
"acetylcholine",
"receptors",
"biochemistry",
"signal",
"transduction",
"cellular",
"neuroscience",
"cell",
"biology",
"physiology",
"neurons",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"physical",
"sciences",
"neurophysiology"
] |
2017
|
Chrna2-Martinotti Cells Synchronize Layer 5 Type A Pyramidal Cells via Rebound Excitation
|
Acquisition of cis-regulatory elements is a major driving force of evolution , and there are several examples of developmental enhancers derived from transposable elements ( TEs ) . However , it remains unclear whether one enhancer element could have been produced via cooperation among multiple , yet distinct , TEs during evolution . Here we show that an evolutionarily conserved genomic region named AS3_9 comprises three TEs ( AmnSINE1 , X6b_DNA and MER117 ) , inserted side-by-side , and functions as a distal enhancer for wnt5a expression during morphogenesis of the mammalian secondary palate . Functional analysis of each TE revealed step-by-step retroposition/transposition and co-option together with acquisition of a binding site for Msx1 for its full enhancer function during mammalian evolution . The present study provides a new perspective suggesting that a huge variety of TEs , in combination , could have accelerated the diversity of cis-regulatory elements involved in morphological evolution .
Morphogenesis is generally controlled by spatiotemporal expression of a number of specific gene sets [1] . The acquisition of novel phenotypic traits during mammalian evolution has been posited to result from changes in gene expression patterns , which are mediated by gain of new cis-regulatory elements such as enhancers [2] . Mammalian genomes contain hundreds of thousands of conserved non-coding elements ( CNEs ) , which , in humans , occupy 3–8% of the genome [3] . Because CNEs are expected to include a number of transcriptional enhancers [4] , they are recognized as highly important clues to understanding the key gene regulatory mechanisms involved in mammalian evolution [5–8] . Mammals have acquired a variety of morphological features during evolution . One of the striking evolutionary events is the development of the bony secondary palate [9] . Complete closure of the mammalian secondary palate during development ( palatogenesis ) separates the oral cavity from the nasal cavity , which allows breathing while eating and efficient suckling . This closure begins with the formation of bilateral palatal shelves ( PS ) in the embryonic maxillary prominences , and then the PS grow horizontally to fuse with each other at the midline [10] ( S1 Fig ) . Dozens of genes such as the msx1 and wnt family are known to be involved in palatogenesis [11 , 12] . Agenesis of the secondary palate , known as cleft lip/palate , is one of the most common congenital defects in humans , occurring once in every 700 newborns [11] . In this regard , wnt5a is one of the possible responsible genes identified by genetic association studies of human cleft lip/palate [13] . Correspondingly , mice lacking wnt5a or its non-canonical receptor Ror2 are born with cleft palate [14 , 15] . Therefore , revealing the molecular regulatory mechanisms of such genes , which remain largely unknown , is essential to our understanding of the molecular basis of mammalian-specific morphological evolution as well as that of the cleft lip/palate defect in humans . Transposable elements ( TEs ) , i . e . , retroposons and DNA transposons , occupy nearly half of mammalian genomes . Retroposons such as SINEs propagate their copies via reverse-transcription of their RNA intermediates , with reintegration of the copied DNA , whereas DNA transposons simply directly relocate within the genome [16–18] . Although TEs are , in general , regarded as genomic parasites or sometimes as harmful dynamic mutagens , we for the first time proposed , together with the Bejerano’s group , that some TEs are involved in macro-evolution by showing that they overlap with CNEs [7 , 19] . This fact implies that TEs under purifying selection acquired functions during evolution [20 , 21] , which is called exaptation [22] or co-option , and that many types of TEs such as SINEs might have contributed to various morphological innovations during mammalian evolution [8] . Indeed , we previously demonstrated that hundreds of AmnSINE1 sequences are evolutionarily conserved among mammals [7 , 23 , 24] . One AmnSINE1 is an enhancer of fgf8 in the diencephalon , and another acts as an enhancer of satb2 expression in the deep layer of the neocortex , especially in callosal projection neurons [23 , 25 , 26] . Further , the LF-SINE locus , which is shared among tetrapods , serves as a distal enhancer of the neurodevelopmental gene Isl1 [19] , and Pomc has two neuronal enhancers derived from CORE-SINE and MaLR [27 , 28] . Thus , it has been clearly established that TEs are one of the main sources of cis-regulatory elements [29] . Mammalian genomes harbor a variety of TEs as exemplified by the human genome , which has >1 , 100 types ( subfamilies ) that occupy >45% of the genome . These facts prompted us to consider whether multiple TEs of different origin and sequence , being located proximal to one another , could be co-opted/exapted as a single enhancer element . If so , a huge diversity of developmental enhancers could have been generated by combining different TE types during evolution . This possibility has never been examined , however , and in all the known cases of co-option/exaptation , a developmental enhancer was found to consist of only a single TE [19 , 23 , 27 , 28] , although one interesting case in which the promoter region of decidual prolactin was reported to be derived from two TEs [30] . Here , we report that a CNE containing three TEs , including AmnSINE1 , acts as a distal enhancer of wnt5a during palatogenesis . This is an unprecedented example , to our knowledge , in which three different TEs inserted side-by-side play a cooperative role in the distal enhancer function within a CNE . TEs located proximal to one another may have potential as genetic sources of diversity of regulatory elements and that such cooperative enhancers might have contributed to mammalian morphological evolution through controlling spatiotemporally diverse expression of various genes .
The 1 . 2-kb AS3_9 locus is located at chr3:54916774–54917973 of the human genome ( GRCh38/hg38 ) ( Fig 1A and 1B ) . This locus is one of the hundreds of AmnSINE1-derived CNEs in mammals identified by our group [24] . In this locus , the AmnSINE1-related region is 71 . 4% identical to nucleotide positions 391–501 of its original consensus sequence ( Fig 1B , S2 Fig ) [7] . To test whether the evolutionarily conserved AS3_9 locus possesses an enhancer function—as is the case for other AmnSINE1-derived CNEs [23 , 25 , 26]—we performed a transgenic mouse enhancer analysis using a construct containing AS3_9 and a lacZ reporter gene ( Fig 1B ) . The transgenic mice ( AS3_9-lacZ ) consistently displayed strong lacZ expression in the frontonasal region at embryonic day 13 . 5 ( E13 . 5 ) ( Fig 1C , S3A Fig ) . Especially , lacZ was expressed in the frontonasal prominence including the medial and lateral nasal processes , the maxillary processes that give rise to the upper lip and PS , and mandibular processes that form the lower lip . The lacZ expression patterns in the frontonasal prominence during embryogenesis were consistent among the three AS3_9-lacZ mouse lines we established in this study ( S3B Fig ) . We expected that this TE-derived CNE serves as a distal enhancer of a gene responsible for the development of the frontonasal region in mammals . The ~2-Mb region surrounding AS3_9 contains eight genes ( Fig 1A ) . We carried out in situ hybridization ( ISH ) for each of the eight candidate genes by using the respective mRNA as a probe; this revealed that only wnt5a is expressed in the frontonasal region ( S4 Fig ) . This is consistent with a previous report that wnt5a is expressed in the frontonasal prominence and anterior side of PS [14] and is responsible for secondary palate development [12 , 15] . We also found that lacZ expression in the AS3_9-lacZ embryos coincided exactly with wnt5a expression in the frontonasal prominence at E10 . 5 ( S5A Fig ) as well as in the anterior side of PS at E13 . 5–14 . 5 ( Fig 1C–1E; S5B Fig ) . These results suggested that AS3_9 is a distal enhancer for wnt5a expression in the frontonasal region including PS . We next assessed whether AS3_9 serves as an enhancer of wnt5a expression during secondary palate formation . Using AS3_9-ko mouse embryos in which the TE-derived 800-bp region of AS3_9 was targeted ( see S6 Fig ) , we carried out both ISH ( wnt5a mRNA probe ) and a histological analysis . AS3_9 homozygous mutant mice established from two lines were viable and fertile . Expression of wnt5a in the frontonasal region of E14 . 5 homozygous AS3_9-ko embryos was weak and/or irregular compared with wild-type littermates ( Fig 2 ) . For example , one of the AS3_9-ko mice ( #1 in Fig 2 ) showed no wnt5a expression on the anterior side of PS ( Fig 2B ) and very weak expression in the PS and mandibular processes in the intermediate region ( Fig 2F ) . Another mouse ( #2 ) displayed moderate wnt5a expression in mandibular processes but little in the PS ( Fig 2C and 2G ) . Therefore , none of the AS3_9-ko embryos showed strong wnt5a expression , i . e . , equivalent to that of wild type . This result demonstrated that AS3_9 is indeed an enhancer of wnt5a expression . To investigate whether reduced wnt5a expression could affect PS development , we performed a histological analysis of E14 . 5 wild-type and AS3_9-ko littermates; notably , the knockout embryos did not exhibit any distinguishable agenesis or delayed palatogenesis ( S7A Fig ) . Moreover , at E15 . 5 , the PS of AS3_9-ko embryos were completely closed as was observed in the wild-type embryos ( S7B Fig ) . Therefore , even when wnt5a expression was unstable or weak in the AS3_9 mutants , palatogenesis progressed essentially normally , presumably owing to a compensation mechanism involving other cis-elements ( see Discussion ) . The conservation pattern of AS3_9 implies that this CNE can be divided into four sub-elements ( conservation graph in Fig 1B ) , prompting us to investigate the origins of the conserved sub-elements . Interestingly , we found that , in addition to the AmnSINE1 region , two other conserved sub-elements were derived from other TEs , namely X6b_DNA and MER117 , which are 74 . 4% and 72 . 8% identical to their full-length consensus sequences , respectively ( S8A and S8B Fig ) . X6b_DNA is a non-autonomous DNA transposon distributed in Theria ( placental mammals and marsupials ) , whereas MER117 is a hAT-type non-autonomous DNA transposon distributed only in placental mammals . Consistent with these distributions , we found that the orthologs of the X6b_DNA and MER117 elements in AS3_9 are only found in therian and placental mammals , respectively ( Fig 1B; S8 Fig ) . The other conserved region is not derived from a known TE or repetitive sequences ( gray bar in Fig 3 ) , as we confirmed that a RepeatMasker analysis and a blast search against the human genome returned no significant hit . Because each of the three TEs in AS3_9 was conserved as an independent sub-element of the locus ( S2 and S8 Figs ) , we expected that they make distinct contributions to overall enhancer function . To clarify each role , deletion constructs lacking various combinations of the TE regions were used for enhancer assays with transgenic mice ( Fig 3A–3I , S9A–S9I Fig ) . At E13 . 5 , enhancer activities were evaluated based on lacZ expression in the ventral region including the maxillary and mandibular processes , in the rostral region including the medial and lateral nasal processes , and in PS . Embryos harboring a construct with only the three TE regions ( Fig 3B and 3B' ) showed strong lacZ expression equivalent to that of AS3_9-lacZ mice ( Fig 1C , Fig 3A and 3A' ) . Conversely , the construct lacking the three TE-derived regions lacked enhancer activity in the frontonasal region ( Fig 3E and 3E' ) . These results indicated that the three TE regions were sufficient to recapitulate the full AS3_9 enhancer activity and that other regions , such as the non-TE conserved region ( gray bar in Fig 3 ) , probably do not contribute to enhancer function . Constructs lacking the MER117 region showed no or very weak enhancer activity in the medial and lateral nasal processes ( Fig 3C and 3C' ) , whereas only the constructs carrying MER117 yielded a lacZ signal in the apex of the nose ( compare Fig 3D' with 3E' , 3F' with 3I' and 3G' with 3H' ) . The MER117 region is , therefore , responsible for enhancer function in nasal processes , especially in the nose apex . The presence of X6b_DNA in the constructs always yielded lacZ expression in the ventral region , namely , the maxillary and mandibular processes ( Fig 3A' , 3B' , 3C' , 3F' and 3I' ) . Consistently , only the lack of X6b_DNA resulted in no lacZ expression in this region ( compare Fig 3A' with 3G' , 3D' with 3F' and 3E' with 3I' ) . Accordingly , the X6b_DNA region is responsible for the enhancer activity in the ventral region . The AmnSINE1 region alone did not yield lacZ expression ( Fig 3H and 3H' ) ; however , this region increased the range and intensity of the enhancer activity of X6b_DNA and MER117 . For example , X6b_DNA alone supported enhancer activity mainly in the maxillary process and weak activity in the mandibular process ( Fig 3I and 3I' ) ; addition of AmnSINE1 yielded strong lacZ expression in the mandibular process as well as limited parts of the nasal prominence ( Fig 3C and 3C' ) . Likewise , when AmnSINE1 was present , the enhancer signal of MER117 at the nose apex ( Fig 3D' ) extended somewhat further toward the upper region of the medial nasal processes and part of the lateral nasal processes ( Fig 3G and 3G' ) . Notably , enhancer activity in PS was observed only when all three TE regions were included in the same construct ( Fig 3A'' and 3B'' ) . These results suggested that each TE plays a distinct role in wnt5a enhancer function . To elucidate the molecular mechanism of the AS3_9 enhancer , we utilized the yeast one-hybrid system to search for transcription factors that bind the AS3_9 sequence . Twelve candidate genes were identified ( S1 Table , S10A–S10C Fig ) , of which three ( Msx1 , Msx2 , Gtf2ird1 ) are known to be involved in mammalian craniofacial development [31 , 32] . The most noteworthy finding was msx1 because it has been demonstrated as one of the genes responsible for cleft palate in humans [33] and mice [31 , 34 , 35] . We found an Msx1-binding motif ( TAATTG ) [36] within the X6b_DNA-derived sequence of AS3_9 . Mutation of this site abrogated Msx1 binding ( S10D–S10F Fig ) . Furthermore , we conducted enhancer analysis using the AS3_9 sequence in which an identical mutation was introduced in the Msx1-binding site . Intriguingly , the transgenic embryos showed limited enhancer activity in the medial nasal process and maxillary process as well as loss of activity in PS ( Fig 3J' , S9J Fig ) , similar to the X6b_DNA-deleted constructs ( Fig 3D' and 3G' ) . These results indicated that Msx1-binding is essential for the full enhancer function of AS3_9 and suggested that the msx1 and wnt5a signaling pathways may interact closely during the secondary palate development .
Our analysis of the orthologs within the AS3_9 locus revealed that the AmnSINE1 , X6b_DNA , and MER117 elements are present only among Mammalia , Theria , and Eutheria , respectively , suggesting that they were integrated in this order during evolution ( Fig 1A , S2 and S8 Figs ) . Fig 4 shows the evolutionary scenario for the establishment of the AS3_9 enhancer . Although the AmnSINE1-derived region alone lacks enhancer function , this region is evolutionarily conserved even between humans and platypus ( S2 Fig ) . Therefore , the AmnSINE1-derived region might have had another/unknown function in a common ancestor of mammals before 186 million years ago ( Mya ) [37] , and it might have acquired a new additional role as the AS3_9 enhancer during evolution . After divergence of monotremes , integration of X6b_DNA and subsequent acquisition of the Msx1-binding site resulted in co-option/exaptation 170–186 Mya [37] . The AmnSINE1 and X6b_DNA region might serve as the wnt5a enhancer in the developing maxillary and mandibular processes . After divergence of marsupials ( 170 Mya ) [37] , MER117 was integrated , and finally AS3_9 was established as the current complete enhancer with extended activity to the medial and lateral nasal processes as well as PS . This is the first demonstration , to our knowledge , of stepwise evolution via co-option/exaptation of a developmental enhancer . The full activity of the AS3_9 enhancer in the whole frontonasal region and PS was not observed with any one of AmnSINE1 , X6b_DNA , and MER117 alone and was only attained with the combination of all three sub-elements ( Fig 3A'' and 3B'' ) . Therefore , the three TEs act cooperatively and synergistically as one complex distal enhancer element during palatogenesis . The division of roles among the TEs ( Fig 3 ) implies that they undergo different epigenetic modifications in the different tissues . To address this possibility , we investigated the ChIP-seq data of the ENCODE project available in the UCSC database ( S11A Fig ) and the Roadmap Epigenomics project data ( S11B Fig; http://www . roadmapepigenomics . org/ ) . The UCSC genome browser shows that the chromatin states of AmnSINE1 and X6b_DNA regions of AS3_9 are open in various fibroblast cell lines ( black bar in S11A Fig ) . This strong signal for open chromatin in the X6b_DNA region is also observed in the Roadmap Epigenomics data ( see DNase column in S11B Fig ) . For histone modifications , the AmnSINE1 and MER117 regions show weak inactive/heterochromatin states ( e . g . , H3K9me3/H3K27me3/H3K36me3 ) in many other cells ( S11B Fig ) . The ChIP-seq data for transcription factors show the binding of CTCF to the AmnSINE1 + X6b_DNA region and the bindings of NR2C2 and SRF to the MER117 ( S11A Fig ) . Although involvement of NR2C2 or SRF in palatogenesis has not been reported , it is possible that these proteins are involved in the secondary palate formation in mammals . Unfortunately , these epigenetic states has not been tested in the frontonasal region or PS during the corresponding developmental stages of mice . Future examination for these epigenetic signals of AS3_9 may lead us to further understanding of the molecular mechanism for the formation of the secondary palate in mammals . It is generally considered that robust gene expression is ensured by the presence of a backup cis-regulatory system such as primary and secondary enhancers [38 , 39] . Actually , many studies have demonstrated that deletion of one enhancer can perhaps be compensated by another enhancer with little effect on phenotype [40–42] . Therefore , complete palatogenesis in the AS3_9-ko mice was probably due to the remaining of weak wnt5a expression ( Fig 2 , S7 Fig ) . It is likely that AS3_9 serves as one of multiple cis-regulatory elements responsible for wnt5a expression during secondary palate development . This hypothesis can be rationalized from paleontological evidence that suggests that acquisition of the bony secondary palate dates back 200 Mya [43] . Therefore , before the divergence of monotremes ( 184 Mya ) , other enhancer ( s ) , the presence of which was suggested above , might have been responsible for the formation of the bony secondary palate of early mammals . Because wnt5a-deficient mice reduced expression of msx1 , bmp4 , and shh in the anterior palate , wnt5a is considered to act upstream of these genes [12 , 15] . Little is known , however , about the molecular mechanisms by which wnt5a expression is regulated . Msx1 is also associated with human non-syndromic cleft palate [33] , and msx1-deficient mice have cleft secondary palate [34] as well as reduced expression of bmp4 and shh [35] . In the anterior PS , msx1 up-regulates bmp4 and vice versa , and bmp4 controls the downstream shh signaling that triggers PS growth [12 , 35] . As we showed in the present study , it is remarkable that the Msx1-binding site in AS3_9 is necessary for its enhancer function ( Fig 3J and 3J' , S9J Fig ) , suggesting that wnt5a expression in the anterior palate is controlled by msx1 . Therefore , taking the previous study suggesting that msx1 is one of the downstream genes of wnt5a into consideration [15] , wnt5a and msx1 may have a synergistic effect on palatogenesis , as is the case with msx1 and bmp4 [35] . Therefore , palatogenesis is presumably controlled not by simple hierarchical signaling but rather by various interdependent cis-regulatory elements . The present study showed that in the distal enhancer of wnt5a three TEs take their part cooperatively in palatogenesis ( Fig 3 ) . Notably , each of the X6b_DNA and MER117 regions of AS3_9 possesses a distinct tissue-specific enhancer property by itself ( Fig 3D' and 3I' ) , indicating that different function can be evolved independently by multiple TEs located close to one another . In general , hundreds or more of TE types ( subfamilies ) constitute 20–50% of vertebrate genomes [44] . For example , it has been reported that several TEs contain motifs of functional sequences such as the CTCF-binding motif in rodent B2 SINE [45 , 46] or in the MER20 DNA transposon [47] , the Nfi-binding motifs in MER130 [48] , and the OCT4-binding site in LTR7 or MER74A [49 , 50] , clearly demonstrating that certain TEs have the potential to acquire a function during evolution . Therefore , by multiple TEs being inserted close to one another , it is possible that they subsequently acquired a new regulatory function during evolution . We expect that many such coordinated TE-derived enhancers are hidden in mammalian genomes . To find clues that support this hypothesis , we searched the human genome for AmnSINE1 copies located proximal to other TEs , all of which overlap CNEs ( S3 Table ) . Among the 626 conserved AmnSINE1 loci , 54 elements ( 8 . 6% ) accompany other TEs that have been evolutionarily conserved , including all the major TE classes such as SINEs , LINEs , LTR-retrotransposons , and DNA transposons . Thus , the possibility arises that , at least at some loci among these 54 CNEs , several TEs located proximal to one another cooperate to modulate cis-regulatory networks that have been involved in the evolution of morphological innovations . This perspective extends the potential of TEs as genetic sources of a broader diversity of cis-regulatory elements . Further functional analysis of these TE-derived cis-regulatory elements will enhance our understanding of their involvement in morphological innovation during evolution .
The mouse strains B6C3F1 , C57BL/6 , and ICR were purchased from Sankyo Laboratory Service Corporation ( Tokyo , Japan ) . Animals were kept in ventilated cages under a 12-h light/dark cycle at 24°C . This study was approved by the Ethics Committee of Tokyo Institute of Technology and Institutional Animal Care and Use Committee of RIKEN Kobe Branch . A 2 . 1-kb DNA fragment of the mouse AS3_9 locus was amplified by PCR using primers AS3_9-F and AS3_9-R ( S2 Table ) containing Hind III recognition sites . The product was cloned into the Hind III site of plasmid HSF51 harboring the mouse heat-shock protein 68 promoter followed by the bacterial lacZ reporter gene and the SV40 poly-A signal , yielding the AS3_9-HSF51 construct . The AS3_9 deletion constructs containing various combinations of TEs ( Fig 3A–3I ) were generated by overlap extension PCR as described [26] . Briefly , internal primers overlapping complementary sequences ( S1 Table ) were designed to carry out the deletion of each TE region . The first PCR was performed with one of the internal primers and either one of the vector primers ( HSF51-F or HSF51-R; S2 Table ) using AS3_9-HSF51 as template . The resulting PCR fragments were used as templates for the second PCR with the two vector primers , and the PCR products were cloned into HSF51 upstream of the heat-shock protein 68 promoter via Sal I and Hind III sites . Transgenic mice were produced as described [23 , 26] . Briefly , the constructs were linearized with Sal I and Xho I . After purification using the Gel Extraction kit from Qiagen , the DNA fragments were dialyzed against microinjection buffer ( 5 mM Tris-HCl , 0 . 1 mM EDTA ) at 4°C overnight . Pronuclear microinjection was performed using 6–10 ng/μl of the DNA solution into a B6C3F1 zygote , and microinjected zygotes were transferred to the oviduct of pseudopregnant ICR females . Transgenic mouse embryos were identified by PCR genotyping using primers LZgt-F02 and LZgt-R01 from yolk samples . The transgenic embryos were fixed for 1 h in phosphate-buffered saline ( PBS ) containing 1% formaldehyde , 0 . 1% glutaraldehyde , and 0 . 05% ( v/v ) NP-40 and then stained with PBS containing 500 μg/ml X-gal , 5 mM K3Fe ( CN ) 6 , 5 mM K4Fe ( CN ) 6 , 2 mM MgCl2 , 0 . 02% NP-40 , and 0 . 01% sodium deoxycholate for >3 h at 37°C . Consistency among lacZ expression patterns was confirmed by multiple microinjection experiments . For staining sections , the fixed embryos were permeated with 30% sucrose in PBS overnight at 4°C and embedded with O . C . T . compound ( Tissue-Tek , Sakura , Torrance , CA ) for sectioning . Coronal sections produced with a cryostat ( Leica CM 1850 ) were counterstained with kernechtrot ( Nuclear Fast Red ) . In addition to the analysis of transient transgenic embryos , we generated three stable lines derived from AS3_9-lacZ transgenic mice ( S3B Fig ) . The E9 . 5–15 . 5 embryos of each heterozygote transgenic mouse were harvested and stained with X-gal solution as described above . Mouse embryos for ISH for wnt5a and lacZ were prepared from ICR and stable transgenic lines , respectively . E11 . 5 embryos for whole-mount ISH were fixed overnight in 4% paraformaldehyde dissolved in diethylpyrocarbonate-treated PBS at 4°C . For section ISH ( E13 . 5–15 . 5 ) , embryos were permeated overnight with 30% sucrose dissolved in diethylpyrocarbonate-treated PBS after fixation . Then , the embryos were embedded with O . C . T . compound and frozen . Sections were prepared on a Leica sledge microtome at 14 μm and individually mounted on slides . Digoxigenin-labelled antisense RNA probes were synthesized from linearized plasmids with T3 and T7 polymerase ( Roche , Basel , Switzerland ) . Respective plasmids carrying subcloned coding regions of wnt5a and lacZ were prepared . ISH was performed as described [51] . ISH was also performed on E13 . 5 coronal cryosections as described [25] to examine the expression of the AS3_9-proximal genes . Briefly , eight genes ( wnt5a , erc2 , lrtm1 , caca2d3 , selk , actr8 , il17rb , and chdh ) surrounding AS3_9 were identified with the UCSC Genome Browser ( http://genome . ucsc . edu/ ) . The cDNAs were amplified by PCR from an E14 or E17 cDNA pool , cloned into pGEM T-easy vectors ( Promega , Madison , WI , USA ) , and used for probe syntheses . Probes prepared by in vitro transcription using the DIG RNA Labeling kit ( Roche ) were purified by lithium chloride precipitation and used for ISH as described [25] . For the ISH of the AS3_9-ko mice for the wnt5a probe , the plasmid containing the mouse wnt5a cDNA ( 1 . 4 kb ) was generated via PCR with the primers in S2 Table and sub-cloned into pBluescript ( KS- ) . The embryos were fixed , embedded in Paraplast , and serially sectioned ( 10 μm thickness ) . Sections were subjected to ISH as described [51 , 52] . The AS3_9 sequence ( chr14:29028538–29029337 of GRCm38/mm10 ) , including the three TEs , was targeted ( S6A Fig ) . Two arm fragments ( 7 . 1 kb of the long arm and 3 . 6 kb of the short arm ) were amplified using an LA PCR kit ( Takara , Japan ) with the following primers: AS3_9-LongArm-F2 and AS3_9-LongArm-R2 for the long arm , and AS3_9-ShortArm-F and AS3_9-ShortArm-R for the short arm ( S2 Table ) . The long-arm fragment was sequenced and cloned into the 5’ ( Not I and Sal I ) cloning site of the PGK-Neo-pA cassette in the Targeting vector ( DT-A-pA/loxP/PGK-Neo-pA/loxP; see http://www2 . clst . riken . jp/arg/cassette . html for details ) ; the short arm was also sequenced and cloned in the 3’ ( Xho I ) site in the same vector . Homologous recombination was conducted using TT2 embryonic stem cells [53] , and two recombinants were used to produce chimeric AS3_9-deficient mice ( #49 and #154 ) For Southern hybridization , DNA probes for the 5’ and 3’ regions of the targeted sequence as well as the Neo sequence were amplified with specific primers ( S2 Table ) and labelled with [α-32P]dCTP . Genomic DNA ( 10 μg each ) from F1 individuals was digested with Bln I and Bsp1407 I ( for the 5’ probe ) , EcoR V ( for the 3’ probe ) , or Sac I ( for the Neo probe ) . Signals of 12 . 0 , 6 . 2 , and 6 . 3 kb for the mutant alleles were detected by Southern blotting using the 5’ , 3’ , and Neo probes , respectively ( S6C–S6E Fig ) . PCR genotyping was performed with AS3_9-gtF1 ( KO ) and AS3_9-gtR for detection of the knockout allele and AS3_9-gtF1 ( WT ) and AS3_9-gtR for the wild-type allele ( S6B Fig; S2 Table ) . The Matchmaker Gold Yeast One-Hybrid Library Screening System kit ( Clontech , Palo Alto , CA , USA ) was used to identify proteins that could bind to the AS3_9 sequence . A SMART cDNA library was constructed using the kit from the frontonasal tissues of 21 mouse embryos at E14 . 5 . The bait DNA ( chr14:29028539–29029109 of GRCm38/mm10 ) consisted of AmnSINE1 and X6b_DNA regions because the sequence is responsible for the enhancer activity in maxillary processes , the origin of PS outgrowth . Yeast one-hybrid screening was conducted with two consecutive selection steps with 100 ng/ml ( first step ) and 500 ng/ml ( second step ) of the antibiotic aureobasidin A . Among 11 . 4 million clones of the library screened , 14 positive clones ( S1 Table ) were isolated consisting of 12 genes in total , which included craniofacial developmental genes ( Msx1 , Msx2 , and Gtf2ird1 ) . Mutation in the Msx1-binding site of the bait sequence was introduced ( TAATTGG -> gccgTGt ) using appropriate primers ( S2 Table ) , and the yeast one-hybrid assay was performed according to manufacturer's protocol with aureobasidin A ( 250 ng/ml ) . We performed an in silico screen of the human genome ( hg38 ) to identify AmnSINE1-derived CNEs proximal to other TE-derived CNEs . By comparing a list of conserved elements ( phastConsElements100way ) from the UCSC genome database and the latest TE annotation list by RepeatMasker ( with the repeat library 20140131; http://www . repeatmasker . org/species/hg . html ) , all TEs overlapping >30 bp with the conserved elements ( LOD score >100 ) were extracted . Among the 626 evolutionarily conserved AmnSINE1 sequences found , those proximal to ( <600 bp ) another TE-derived CNE were collected and listed in S3 Table . Details of the AS3_9-ko lines are available at http://www2 . clst . riken . jp/arg/mutant%20mice%20list . html ( Accession No . CDB0941K ) .
|
Acquisition of cis-regulatory elements is a major driving force of evolution , but its whole evolutionary significance is poorly understood . Here , we found an unprecedented case in that three TEs cooperatively and synergistically act as a distal enhancer for wnt5a expression in mammalian frontonasal region , being involved in the evolution of the secondary palate formation . We elucidated the stepwise evolutionary history of getting the enhancer activity via ( retro- ) transposition of TEs during mammalian evolution . Taking various types of TEs constituting nearly half of mammalian genomes into consideration , this study provides a new perspective for the potentiality that almost infinite combinations of proximally-located different TEs could have generated a huge diversity of developmental enhancers and had a great impact on mammalian morphological evolution .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"sequencing",
"techniques",
"medicine",
"and",
"health",
"sciences",
"vertebrates",
"animals",
"mammals",
"palate",
"developmental",
"biology",
"sequence",
"motif",
"analysis",
"mammalian",
"genomics",
"molecular",
"biology",
"techniques",
"embryos",
"digestive",
"system",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"embryology",
"artificial",
"gene",
"amplification",
"and",
"extension",
"molecular",
"biology",
"animal",
"genomics",
"evolutionary",
"genetics",
"mouth",
"anatomy",
"polymerase",
"chain",
"reaction",
"genetics",
"biology",
"and",
"life",
"sciences",
"genomics",
"evolutionary",
"biology",
"amniotes",
"organisms",
"evolutionary",
"developmental",
"biology"
] |
2016
|
Coordinately Co-opted Multiple Transposable Elements Constitute an Enhancer for wnt5a Expression in the Mammalian Secondary Palate
|
Bartonella spp . are globally distributed bacteria that cause endocarditis in humans and domestic animals . Recent work has suggested bats as zoonotic reservoirs of some human Bartonella infections; however , the ecological and spatiotemporal patterns of infection in bats remain largely unknown . Here we studied the genetic diversity , prevalence of infection across seasons and years , individual risk factors , and possible transmission routes of Bartonella in populations of common vampire bats ( Desmodus rotundus ) in Peru and Belize , for which high infection prevalence has previously been reported . Phylogenetic analysis of the gltA gene for a subset of PCR-positive blood samples revealed sequences that were related to Bartonella described from vampire bats from Mexico , other Neotropical bat species , and streblid bat flies . Sequences associated with vampire bats clustered significantly by country but commonly spanned Central and South America , implying limited spatial structure . Stable and nonzero Bartonella prevalence between years supported endemic transmission in all sites . The odds of Bartonella infection for individual bats was unrelated to the intensity of bat flies ectoparasitism , but nearly all infected bats were infested , which precluded conclusive assessment of support for vector-borne transmission . While metagenomic sequencing found no strong evidence of Bartonella DNA in pooled bat saliva and fecal samples , we detected PCR positivity in individual saliva and feces , suggesting the potential for bacterial transmission through both direct contact ( i . e . , biting ) and environmental ( i . e . , fecal ) exposures . Further investigating the relative contributions of direct contact , environmental , and vector-borne transmission for bat Bartonella is an important next step to predict infection dynamics within bats and the risks of human and livestock exposures .
Bats ( Order: Chiroptera ) serve as reservoir hosts for viruses of concern for human and animal health [1 , 2] including SARS coronavirus , rabies virus , filoviruses , and henipaviruses [3–6] . Bats can also harbor protozoa and bacteria of potential zoonotic relevance [7–9] . Bartonella spp . are of particular interest , as these Gram-negative bacteria cause bacteremia and endocarditis in both humans and livestock [10 , 11] and exhibit high genetic diversity in bats across multiple continents and species [12–17] . Moreover , phylogenetic analyses show bats are reservoirs of zoonotic Candidatus B . mayotimonensis [18–20] , a causative agent of human endocarditis [21] . Given the zoonotic potential of bat-associated Bartonella , understanding transmission within bats is critical for understanding how Bartonella persists in bat populations and for assessing spillover risks [22 , 23] . Ectoparasites are frequently invoked as a transmission route [12 , 19 , 24] , in part because vector-borne transmission occurs in other taxa [25 , 26] and because Bartonella has been identified in bat flies and ticks [27–29] . While some bat ticks can feed on humans [30] , the high host specificity of bat flies [31 , 32] could limit opportunities for cross-species transmission through ectoparasites [31–33] . Transmission through close contact ( e . g . , biting ) could occur given detection of Bartonella in dog and cat saliva [34 , 35] as well as human infection following scratches from dogs and cats [36] . Phylogenetic patterns of weak Bartonella host specificity in Neotropical bat communities could not only reflect transmission through close contacts between species in multi-species roosts , but could also stem from transmission through generalist vectors [15 , 24 , 37] . Bartonella might also be transmitted through exposure to feces between bats and to humans that enter roosts or to domestic animals exposed to bat feces [18 , 38] . In addition to the potential risks of cross-species transmission from bats to livestock and humans , the infection dynamics of Bartonella in bats are also uncertain . In rodents , Bartonella prevalence varies through time [39 , 40] , but such patterns have not been well studied in bats [41] . Individual heterogeneities in infection by age and sex could also inform exposure patterns . Finally , global analyses suggest geographic structure in bat Bartonella genotypes , with notable differences in genotypes from Latin American and those from Africa , Europe , and Asia [42] . However , as such patterns appear driven by bat families restricted to different continents , analyses within narrower geographic and taxonomic ranges could inform the scale of Bartonella transmission and the role that dispersal plays in the spatial dynamics of this infection [43] . Common vampire bats ( Desmodus rotundus ) have high prevalence of Bartonella throughout their large geographic range in Latin America [15 , 16 , 24 , 44] . Vampire bats are of particular concern because they subsist on blood , which could create opportunities for Bartonella transmission to humans and livestock either from bites during blood feeding or through vector sharing facilitated by close proximity [45–48] . Here , we capitalize on a two-year , spatially replicated study of vampire bats to examine the genetic diversity and infection prevalence of Bartonella , including its geographic structure across the vampire bat range as well as individual and temporal correlates of infection status . To explore possible transmission routes of this bacterial pathogen , we also test for associations between bat fly infestation and Bartonella infection status , which would support vector-borne transmission , and by screening bat saliva and fecal samples for evidence of Bartonella DNA , which would support transmission through bites or grooming and environmental exposure to bacteria shed in feces , respectively .
Samples were collected as described in Becker et al . [49] in 2015 and 2016 across seven sites in Peru ( Departments of Amazonas [AM] , Cajamarca [CA] , and Loreto [LR] ) and two sites in Belize ( Orange Walk [OW] District ) . We sampled sites one to two times annually , capturing one to 17 individuals per site and sampling interval ( S1 Table ) . To screen for Bartonella by PCR , up to 30 μL blood was stored on Whatman FTA cards at room temperature . To assess the presence of Bartonella in saliva and feces , we collected oral and rectal swabs from vampire bats in Peru . Swabs were preserved in 2 mL RNAlater ( Invitrogen ) at –80°C until laboratory analyses . For Peru sites sampled in 2016 , we also recorded the number of bat flies per vampire bat [32] . Field procedures were approved by the University of Georgia Animal Care and Use Committee ( A2014 04-016-Y3-A5 ) and the University of Glasgow School of Medical Veterinary and Life Sciences Research Ethics Committee ( Ref08a/15 ) ; all procedures were conducted in accordance with accepted guidelines for humane wildlife research as outlined by the American Society of Mammalogists [50] . Bat capture , sample collection , and exportation were authorized by the Belize Forest Department under permits CD/60/3/15 ( 21 ) and WL/1/1/16 ( 17 ) and by the Peruvian Government under permits RD-009-2015-SERFOR-DGGSPFFS , RD-264-2015-SERFOR-DGGSPFFS , and RD-142-2015-SERFOR-DGGSPFFS . Access to genetic resources from Peru was granted under permit RD-054-2016-SERFOR-DGGSPFFS . We analyzed samples that were previously screened for the presence of Bartonella by Becker et al . [49] using nested PCR to amplify a region of the citrate synthase gene ( gltA ) [51] . Among the Bartonella-positive samples , we randomly selected 5–10 positive samples per site for Sanger sequencing ( n = 51 ) . PCR products were purified with DNA Clean & Concentrator Kits ( Zymo Research ) and sequenced in both directions at the Georgia Genomics Facility . Resulting chromatograms were checked for quality and trimmed using Geneious ( Biomatters ) [52] . Post-trimmed forward and reverse sequences were assembled to create 348 base pair ( bp ) consensus sequences for each sample ( n = 35; the quality of 16 chromatograms was too low ) . Sequences were considered part of the same genotype if they had >96% identity in gltA , an established cut-off for Bartonella species identification [53] . Sequences with >99 . 7% similarity were considered the same genetic variant [54] . We used a Chi-squared test with the p value generated via a Monte Carlo procedure with 1000 simulations [55] to assess whether our defined Bartonella genotypes were associated with region ( i . e . , Belize , eastern Peruvian Amazon , western Peruvian Amazon ) . Two datasets were created for phylogenetic analyses . Dataset 1 was designed to assess the spatial structure of vampire bat–associated Bartonella across Latin America and therefore included our new sequences plus all previously reported gltA sequences from Desmodus rotundus . Dataset 2 was designed to capture the relatedness of the new sequences to all previously described Bartonella spp . regardless of isolation source , which comprised sequences generated in this study plus sequences obtained by conducting a BLAST search of each new sequence against GenBank , selecting the top 10 hits , and removing duplicates . For both datasets , consensus sequences were aligned using MAFFT . Phylogenetic analyses were carried out in MrBayes using the GTR+gamma model suggested by jModeltest2 [56] . For dataset 1 ( Desmodus-associated sequences ) , we fit a codon partitioned substitution model by linking rates in codon positions 1 and 2 separately from codon position 3 . For dataset 2 , we used a simpler non-partitioned model because the more complex codon-partitioned model failed to converge . Dataset 2 included one sequence from Brucella abortus ( Genbank Locus: MIJI01000003 . 1 ) as an outgroup [13] . Both datasets were run for 2 . 5 million generations with convergence checked and burn-in periods selected by assessing posterior traces in Tracer [57] . With dataset 1 , we analyzed spatial clustering of vampire bat Bartonella by country ( Belize , Guatemala , Mexico , Peru ) using Bayesian Tip Association Significance Testing ( BaTS ) [58] . We here selected 1 , 000 trees from the posterior distribution of the MrBayes run and compared the country-level clustering to a null distribution from 10 , 000 trees with swapped tip associations [58] . We analyzed 193 samples from Desmodus rotundus to test whether temporal variation ( season and year ) and individual risk factors ( e . g . , age , sex ) explain differences in Bartonella infection , using generalized mixed effects models ( GLMMs ) with binomial errors and a logit link fit with the lme4 package in R [59 , 60] . We fit a single GLMM with an interaction between site and year to first test if prevalence varied over years across sampling locations; we excluded one site from this analysis ( i . e . , LR6 ) owing to sampling in only 2015 . We included bat identification number ( ID ) as a random effect to account for multiple sampling of a small number of bats ( n = 6 ) . To assess seasonality in infection , we fit a separate GLMM with season ( spring , summer , fall ) as a predictor to data from two sites in Peru ( AM1 and CA1 ) sampled across seasons ( n = 63 ) . We also fit a generalized additive model ( GAM ) with restricted maximum likelihood , binomial response , and a cyclic cubic regression spline for Julian date using the mgcv package [61] . We randomly selected repeatedly sampled bats , as including bat ID as a random effect here overfit the GAM . To identify individual risk factors for Bartonella infection , we fit a single GLMM with bat age , forearm size , sex , and reproductive status; we also included interactions between sex and reproduction , sex and age , sex and forearm size , and reproduction and forearm size . We included categorical livestock biomass as a predictor in the GLMM to control for a previously observed negative association with Bartonella infection ( 121/173 positive bats ) [49] . We fit this GLMM to a reduced dataset free of missing values ( n = 189 ) , included bat ID nested within site as a random effect , and calculated marginal R2 ( R2m ) to assess model fit [62] . Finally , for a data subset ( n = 40 bats sampled in Peru in 2016 ) , we fit two separate GLMs with bat fly intensity and presence as predictors to test whether ectoparasites explained Bartonella infection status . We fit a separate GLM with quasi-Poisson errors to test for sex and age differences in bat fly intensity . To examine possible transmission of Bartonella through biting , grooming , blood sharing , or fecal–oral exposure , we used metagenomic data from a parallel study to screen vampire bat saliva and fecal samples for Bartonella DNA . Three saliva and three fecal pools were shotgun sequenced , each containing nucleic acid extractions from swabs collected from ten vampire bats from one to two colonies . Pooled samples contained individuals from the same colonies of bats tested for Bartonella in blood through PCR , though not necessarily the same individuals . As described previously [8] , total nucleic acid was extracted from swabs and pooled equally according to RNA concentration . Pooled samples were DNAse treated and ribosomal RNA depleted , then cDNA synthesis was performed . Libraries were prepared using a KAPA DNA Library Preparation Kit for Illumina ( KAPA Biosystems ) modified for low input samples , and were individually barcoded during the PCR reamplification step [10] . The libraries included in this study were combined in equimolar ratios with other metagenomic libraries for sequencing on an Illumina NextSeq500 at the University of Glasgow Centre for Virus Research . Reads were demultiplexed according to barcode and quality filtered using TrimGalore [63 , 64] with a quality threshold of 25 , minimum read length of 75 bp , and clipping the first 14 bp of the read . Low complexity reads were filtered out using the DUST method and PCR duplicates removed using PRINSEQ [65] . We screened cleaned reads for the Bartonella genotypes detected in this study using nucleotide BLAST [66] against a custom database composed of the PCR-generated Bartonella sequences from this study , retaining only the best alignment ( the high-scoring segment pair with the lowest e-value ) for a single query–subject pair . To investigate the presence of Bartonella species other than genotypes detected in blood samples from vampire bats , cleaned reads were de novo assembled into contigs using the assembly only function of SPAdes [67] . Individual reads and contigs were screened for sequences matching Bartonella using protein alignment in Diamond [68] , and close matches at the protein level were further characterized by nucleotide BLAST against the Genbank nt database . As the gltA gene is not highly transcribed , we also tested sequences for matches to Bartonella DNA-directed RNA polymerase subunit B ( rpoB ) . We selected two rpoB sequences ( Genbank accessions KY629892 and KY629911 ) from a study of vampire bat Bartonella [16] for which the same individuals exhibited 100% identity in the gltA gene to our blood sequences , and we used Bowtie2 to map quality filtered reads and contigs to those sequences [69] . Lastly , because nucleic acid pools were DNase treated for metagenomic sequencing , potentially reducing detection sensitivity , we used the same nested PCR protocol as used for blood-derived DNA [51] to test for the presence of gltA in DNA from individual saliva and fecal swab samples that made up metagenomic pools ( n = 58; 28 saliva and 30 feces ) . As with our blood samples , we randomly selected a subset of positive amplicons for Sanger sequencing .
Bartonella prevalence across the 193 vampire bats included in this study was 67% . Our phylogenetic analysis of 35 vampire bat Bartonella sequences showed 78 . 8–100% pairwise identity in gltA and revealed at least 11 paraphyletic genotypes ( S2 Table ) . BaTS analysis of all Desmodus-associated Bartonella showed significant phylogenetic clustering by country ( association index = 3 . 81 , parsimony score = 31 . 51 , p<0 . 001 ) , although most vampire bat Bartonella genotypes were still widely distributed ( Fig 1 ) . For the 11 genotypes delineated from our 35 sequences , we observed no association with the geographic study region ( χ2 = 23 . 3 , p = 0 . 27 ) . Genotypes 1 and 2 were detected across all regions , and genotypes 7–10 were detected within both Belize and Peru , highlighting the broad distribution of vampire bat Bartonella genotypes ( Fig 2 ) ; however , genotype 3 was unique to both regions of Peru , genotypes 4–6 were unique to the western Peruvian Amazon , and genotype 11 was only detected in Belize . We also assessed the phylogenetic position of our vampire bat Bartonella sequences among known Bartonella genotypes ( Fig 3 , S3 Table ) . Half of our Peruvian and Belizian sequences ( 18/35 ) were nearly identical ( >99 . 7% identity ) to Bartonella from common vampire bats ( Desmodus rotundus ) from Mexico ( e . g . , GenBank accession numbers KY629837 and MF467803 ) , again confirming the wide geographic distribution of these genotypes . Other sequences ( 9/35 ) fell within the same clade ( >96% pairwise identity ) as Bartonella from bat flies ( Strebla diaemi ) in Panama ( JX416251 ) , from Parnell's mustached bat ( Pteronotus parnellii ) in Mexico ( e . g . , KY629828 ) , from phytophagous bats in Peru ( e . g . , Carollia perspicillata; JQ071384 ) and Guatemala ( e . g . , Glossophaga soricina; HM597202 ) , or from Mexican vampire bats as noted above . Eight sequences were novel ( <96% identity to GenBank sequences ) but were most similar to Bartonella from phytophagous bats in Costa Rica ( e . g . , 90–93% to KJ816666 [Anoura geoffroyi] ) and from Mexican vampire bats ( e . g . , 93% to MF467776 ) . Other novel sequences were weakly related to B . bovis from livestock in Israel and Malaysia ( e . g . , 89–90% to KJ909844 and KR733183 ) , to B . chomelii from cattle in Spain ( e . g . , 89% to KM215693 ) , to B . capreoli from elk in the United States ( e . g . , 89% to HM167503 ) , and to B . schoenbuchensis from roe deer in Germany ( e . g . , 89% to AJ278186 ) ; indeed , posterior support for a bat–ruminant clade was low ( <50%; Fig 3 ) . Our BLAST procedure also identified weakly related Bartonella from rodents ( e . g . , 90% to Rattus norvegicus from the United States [KC763951] and 92% to Apodemus agrarius from China [KX549996] ) and from carnivores ( e . g . , 89% to Procyon lotor from the United States [CP019786] ) . However , these livestock , rodent , and carnivore sequences formed separate phylogenetic clades from bat- and bat fly–derived Bartonella sequences ( Fig 3 ) . Despite the geographic proximity of our field sites to Brazil , our BLAST procedure found no Bartonella seqeunces similar to those recently described in Brazilian bat or rodent species [70–72] . An additional phylogenetic tree that includes these recently identified Bartonella is provided in S1 Fig . Bartonella was detected by PCR in all nine sites in each year , with prevalence ranging from 30–100% ( Fig 4 ) . Prevalence did not differ by year across all sites ( χ2 = 3 . 13 , p = 0 . 54 ) nor within individual sites ( site*year; χ2 = 2 . 82 , p = 0 . 90 ) . The seasonality GLMM for the western Peruvian Amazon ( n = 63 ) showed no difference in odds of infection between spring , summer , and fall ( χ2 = 1 . 99 , p = 0 . 37; S2 Fig ) . The GAM also showed no significant seasonal variation ( χ2 = 0 , p = 0 . 68; S2 Fig ) . Recaptures were rare ( n = 6 ) but showed changes in infection from negative to positive ( n = 2 , 68–424 days ) and from positive to negative ( n = 2 , 15–369 days; S3 Fig ) . After controlling for site-level livestock biomass , vampire bat sex and forearm size were the strongest predictors of infection ( Fig 5 ) ; no interactions were significant ( all χ2≤1 . 18 , p≥0 . 28 ) and were dropped from the final GLMM ( R2m = 0 . 28 ) . The odds of Bartonella infection were highest for vampire bats with larger forearms ( OR = 1 . 2 , p<0 . 001 ) and for males ( OR = 5 . 41 , p<0 . 01 ) , were marginally higher for non-reproductive individuals ( OR = 2 . 36 , p = 0 . 10 ) , and did not differ between subadult and adult bats ( OR = 1 . 58 , p = 0 . 38 ) ; our sample did not contain juveniles . Individual bat fly intensities were highly variable ( 0–28 , median = 7 . 5 ) and showed overdispersion ( ϕ = 5 . 08 in an intercept-only quasi-Poisson GLM ) . The bat fly GLMs showed that neither ectoparasite intensity ( OR = 0 . 98 , p = 0 . 81 ) nor ectoparasite presence ( χ2 = 1 . 13 , p = 0 . 29 ) were associated with Bartonella infection status . We note that the majority of infected bats in this sample were infested with at least one bat fly ( 31/36 ) , limiting conclusive assessment of the ectoparasite–infection relationship . Our multivariable quasi-Poisson GLM showed that ectoparasite load did not vary by bat sex ( χ2 = 0 . 86 , p = 0 . 35 ) or bat age ( χ2 = 0 . 09 , p = 0 . 77 ) . There were no matches in any of the screened saliva and fecal metagenomic pools to the Bartonella gltA sequences detected in the blood or to previously published Bartonella rpoB sequences . The saliva pool from Amazonas had no matching Bartonella-like reads or contigs ( S4 Table ) , while one read each from the Loreto and Cajamarca saliva pool was assigned as Bartonella by nucleotide BLAST; however , these reported species assignments should be interpreted cautiously as they are based on one read and percent identity was low . Pooled fecal samples from all departments of Peru contained Bartonella-like reads and contigs . Bartonella ancashensis , B . australis , and B . bacilliformis were all identified at both the read and contig level in fecal samples . However , because percent identity was relatively low , species assignments should again be interpreted cautiously . Subsequent BLAST hits following the top hit also frequently ( though not always ) matched to Bartonella , suggesting the presence of poorly characterized Bartonella species present or that these may be matches to other bacteria . In contrast , nested PCR of individual swabs detected gltA in 21 . 4% of saliva samples ( 6/28 ) and 30% of fecal samples ( 9/30 ) . For swab samples that were also assessed by PCR in blood ( n = 15 for saliva , n = 28 for feces ) , both corresponding positive saliva samples were positive in blood; most positive fecal samples were also PCR positive in blood , although one fecal-positive sample was PCR negative in blood ( S4 Fig ) . For our random subset of sequenced positive saliva ( n = 4 ) and fecal ( n = 5 ) samples , phylogenetic analyses suggested that all sequences shared a minimum of 97% identity to one or more of our 35 blood-derived Bartonella sequences ( S5 Table , Fig 6 ) . In many cases , saliva and fecal sequences were the same genotype as blood sequences derived from the same geographic region ( e . g . , the saliva sequence from D234 shared >96% identity to the blood sequence from D98 , both from AM1 ) . For the one case in which we sequenced positive samples from the same individual bat ( i . e . , D203 ) , both the blood sequence and fecal sequence shared 100% identity ( S5 Table , Fig 6 ) . For the few sequences at the lower range of our similarity spectrum , BLAST still demonstrated that the closest relatives were all derived from vampire bats ( i . e . , 8368 from CA1 was identical to MF467797 from Mexico ) .
Despite an increasing focus on Bartonella genetic diversity and prevalence in bat communities , individual risk factors and transmission routes of this pathogen in bats remain largely unknown . For example , a survey of vampire bats within Guatemala found neither geographic , dietary , demographic , or viral coinfection correlates of Bartonella infection status [44] . Using a larger sample across a more diverse range of study sites and timepoints , we here show that Bartonella is genetically diverse , geographically widespread and endemic within vampire bat populations , and that individual-level odds of infection are highest for large , male , and non-reproductive bats . Furthermore , we use several approaches to suggest vector-borne transmission to be likely in addition to possible direct contact and environmental sources of Bartonella exposure in bats . The Bartonella genotypes we identified were paraphyletic and closely related to those from other vampire bat populations , other Neotropical bat species , or bat flies . Although BLAST also identified Bartonella spp . sequenced from rodents , carnivores , and livestock within our hit selection criteria , these consistently formed separate phylogenetic clades that did not contain bat- or bat fly–derived Bartonella ( Fig 3 ) . These phylogenetic patterns indicate that Bartonella has commonly shifted between bat host species in the Americas but do not support frequent transmission between bats and other host groups . Our BaTS analysis also showed that vampire bat Bartonella sequences clustered by country more than expected by chance . However , given that several Bartonella genotypes were present in vampire bats from both Central and South America , we suspect this clustering mostly resulted from variation in locally abundant genotypes rather than true barriers to the spread of Bartonella . Because vampire bats are largely sedentary and non-migratory [45] , dispersal of these Bartonella genotypes across large distances is unlikely to be attributable to bat movement alone . Bartonella genotypes may also have infected vampire bats over long evolutionary timescales , and thus the biogeography of the pathogen may have followed that of its host . Alternatively , Bartonella dispersal by other arthropod vectors ( e . g . , ticks ) or other bat species that share Bartonella genotypes with vampire bats may be conceivable and could be resolved by further field surveys combined with population genetic analyses of alternative bat host species , arthropod vector species , and Bartonella genotypes . Few studies have examined temporal patterns of bat Bartonella , emphasizing the general need for more longitudinal studies to understand how pathogens persist in bat populations [1 , 6] . Here , Bartonella was detected at relatively high prevalence across both study years within each sampling site , and neither year nor its interaction with site were predictive within our analyses . Similarly , no temporal patterns in Bartonella were observed for a limited sample of Myotis mystacinus , Pipistrellus spp . , Myotis daubentonii , and Nyctalus noctula in the United Kingdom [41] . Such findings contrast with highly seasonal Bartonella infections in rodents , which show high prevalence in summer and fall due to seasonality in birth and ectoparasite intensity [39 , 40] . The lack of seasonality in our western Peruvian Amazon sample in particular could simply be due to low statistical power; alternatively , no seasonality in infection could also be explained by the non-seasonal or less-pronounced birth pulses observed for vampire bats ( but see [73] ) . While high Bartonella prevalence in bats has been proposed to stem from persistent infection [15] , this seems unlikely , as we observed possible clearance of infection in some recaptured bats . While this could also reflect bacteria DNA loads too low to be detected by PCR , infection risk did not increase with age , as would also be expected if bats could not clear infection [74] . However , we do note that our sample only contained adult ( n = 162 ) and subadult ( n = 28 ) bats , limiting more robust tests of age-dependent infection . Alternatively , Bartonella infections could be chronic and vary in infection intensity over time or could become latent ( i . e . , be undetectable in erythrocytes but persist in endothelial cells ) , particularly as infection does not appear to confer long-term immunity [36] . Such explanations could be confronted in future work with larger sample sizes of recaptured bats , multiple assessments of infection status over time , and quantitative PCR . Bat forearm size , sex , and reproductive status were important predictors of Bartonella infection status , with odds of infection being higher in larger , male , and non-reproductive bats . While subadult status itself was not an important predictor of Bartonella infection , these findings could suggest higher risk in young male bats that are relatively large for their age . Our previous work has shown stronger innate immune defense ( i . e . , bacterial killing ability ) in reproductive ( mostly male ) vampire bats , also suggesting greater susceptibility of non-reproductive hosts [49] . Similarly , subadults across a Mexican bat community also had higher odds of Bartonella infection [16] , and young male vampire bats play key roles in the long-distance dispersal of rabies virus [43] and display higher rates of rabies exposure , possibly owing to more direct contacts during the first year of life [75] . Larger forearm size could also relate to direct contact if larger bats are more dominant and aggressive , as found in other phyllostomid bat species [76] . Although vector-borne transmission is generally assumed for Bartonella in other hosts [12 , 19 , 25 , 77] , including some Neotropical bats [16 , 78] , infection status in vampire bats was not associated with bat fly intensity . Further supporting this observation , male bats had higher odds of Bartonella infection but did not differ in their bat fly intensities compared to females . Weak correspondence between bat fly intensity at the time of sampling and Bartonella infection thus may cast doubt on bat flies as a primary transmission route . Time lags could provide one reason for this discrepancy , given that new Bartonella infections may take days or weeks to develop and become detectable and over which time ectoparasite load may have changed due to the mobile nature of bat flies [26 , 36 , 79] . On the other hand , it is possible that vector presence ( rather than abundance ) is a more important driver of transmission . Unfortunately , nearly all bats in this study had ectoparasites , so comparisons of Bartonella presence in bats with and without bat flies had little statistical power ( 31/36 Bartonella-positive bats were infested with at least one bat fly ) . Given that ectoparasitism predicted Bartonella infection more generally across a Mexian bat community [16] , larger sample sizes with greater variation in bat fly intensity could provide better inference . However , our phylogenetic analysis does provide a tentative line of evidence supporting vector-borne transmission , as several of Bartonella genotypes fell within the same clade as Bartonella from streblid bat flies [29] . A recent survey of Mexican bats and their sympatric bat flies suggested that corresponding hosts and their bat flies had varied Bartonella genotypes , although one vampire bat did show complete sequence homology with the Bartonella from its paired bat flies [37] . As genetic similarity between Bartonella in bat flies and hosts has been interpreted as evidence of vector-borne transmission in other bat species [29 , 54] , further assessments of Bartonella genotypes between vampire bats and their various ectoparasites ( bat flies but also ticks ) would shed additional light on possible routes of vector-borne transmission . Lastly , we analyzed bat saliva and feces using metagenomics and PCR to explore alternative transmission routes , namely through close contact and fecal exposure . Metagenomics detected no Bartonella DNA matching to gltA or rpoB in either saliva or fecal pools . This absence could be explained in that the short sequences ( 345–425 bp ) used as targets , and the large size of bacterial genomes together make the likelihood of detecting a specific gene low . However , the Bartonella-like reads and contigs recovered from saliva and feces were short fragments ( 51–258 bp ) and showed low homology to known Bartonella from GenBank ( S4 Table ) . Notably , we used a similar approach to search for other bacteria ( i . e . , hemoplasmas ) and found clear evidence of their presence [8] . While this could suggest true absence of Bartonella from bat saliva and feces , our PCR found Bartonella in a subset of individual saliva and fecal samples . This discrepancy between methods could stem from treating saliva and fecal pools with DNase before metagenomic sequencing . Furthermore , phylogenetic analyses confirmed that these sequences were closely related to those identified in blood , which argues against these PCR positives only representing bacteria derived from environmental contamination or from feeding on prey . PCR results further showed strong correspondence between blood and saliva , suggesting that Bartonella infection may be systemic in vampire bats . While fecal and blood PCR results also mostly matched , we found one case where a bat was negative in blood but positive in feces . As consumption of ectoparasites during grooming has been observed in other bat families ( e . g . , Pteropodidae [80] ) , this discrepancy could suggest the incidental ingestion of ectoparasites during grooming and that this does not lead to systemic infection more generally indicated by the concordance between blood , saliva , and fecal positives and their close genetic similarity . Similar prevalence of Bartonella in saliva and feces suggests that direct contact and environmental exposure could serve as complemenry transmission routes to arthropod vectors . The presence of Bartonella in saliva samples contrasts with previous work showing an absence of Bartonella in vampire bat saliva [44 , 81] , providing evidence for possible direct transmission . Bartonella in fecal samples could also suggest environmental transmission between bats [18] . Both saliva-borne and fecal–oral transmission of vampire bat Bartonella could further pose potential risks to humans or livestock , either through bites during feeding or by environmental exposure of humans that enter roosts or to domestic animals exposed to bat feces [18 , 38 , 48 , 82] . For the former pathway , however , a recent survey of Bartonella in Mexican ruminants did not identify being bitten by vampire bats as a risk factor for infection [81] , and our phylogenetic results provide relatively more support for the possibility of vector-borne transmission . Vector-borne transmission of vampire bat Bartonella might reduce their potential to infect humans or livestock , given the high host specificity of most bat flies [31 , 32] . However , ectoparasite transfer between individuals could still occur during pupal deposition and close contact [83] , facilitating Bartonella transmission within vampire bat colonies and to other bat species . While our analyses of ectoparasitism only considered bat flies , we have observed heavy tick burdens of vampire bats in other field sites ( e . g . , Belize ) . Bartonella has been detected in ticks infesting other bats [28] , and these ectoparasites could also be more likely to facilitate cross-species transmission [30] . Metagenomics also potentially identified Bartonella ancashensis and B . bacilliformis in vampire bat fecal samples , and these species cause notable infectious disease in humans likely through phlebotomine sand flies in Andean regions of Peru [84 , 85] . Controlled infection trials and more extensive phylogenetic analyses of Bartonella in vampire bats , their various ectoparasites , and sympatric prey are therefore needed to examine the contributions of different transmission routes for bacterial spread within vampire bats and to recipient prey and to confirm whether saliva and feces represent viable transmission routes . Given the high rates of bat bites and proximity to wildlife , humans , and domestic animals that define vampire bat ecology , such efforts to verify the possibility and frequency of oral and environmental exposures would elucidate Bartonella transmission dynamics in this common host species and the risks of cross-species transmission .
|
Bartonella are globally distributed bacteria that can cause endocarditis in humans and domestic animals . Bats have been implicated as a likely reservoir host for these bacteria , but little is known about how prevalence varies over time , routes of transmission , and the genetic diversity of Bartonella in bats . We present results from a two-year , spatially replicated study of common vampire bats , which have previously shown high infection prevalence of Bartonella and could pose risks of cross-species transmission due to their diet of mammal blood . We found that vampire bat Bartonella is genetically diverse , geographically widespread and endemic , and that individual-level infection risk is highest for large , male , non-reproductive bats . Phylogenetic analysis supported vector-borne transmission , and we found support for potential transmission through direct contact and fecal exposures through PCR . Confirming whether arthropod vectors are the main route of transmission for bat Bartonella is needed for understanding infection dynamics in bats and for predicting risks of cross-species transmission to humans and livestock .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"biogeography",
"medicine",
"and",
"health",
"sciences",
"ecology",
"and",
"environmental",
"sciences",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"population",
"genetics",
"vector-borne",
"diseases",
"microbiology",
"vertebrates",
"saliva",
"animals",
"mammals",
"parasitic",
"diseases",
"ectoparasitic",
"infections",
"molecular",
"biology",
"techniques",
"population",
"biology",
"bacteria",
"bacterial",
"pathogens",
"research",
"and",
"analysis",
"methods",
"infectious",
"diseases",
"geography",
"artificial",
"gene",
"amplification",
"and",
"extension",
"medical",
"microbiology",
"microbial",
"pathogens",
"phylogeography",
"molecular",
"biology",
"bartonella",
"eukaryota",
"blood",
"anatomy",
"polymerase",
"chain",
"reaction",
"physiology",
"bats",
"earth",
"sciences",
"genetics",
"biology",
"and",
"life",
"sciences",
"evolutionary",
"biology",
"amniotes",
"organisms"
] |
2018
|
Genetic diversity, infection prevalence, and possible transmission routes of Bartonella spp. in vampire bats
|
Depletion of synaptic neurotransmitter vesicles induces a form of short term depression in synapses throughout the nervous system . This plasticity affects how synapses filter presynaptic spike trains . The filtering properties of short term depression are often studied using a deterministic synapse model that predicts the mean synaptic response to a presynaptic spike train , but ignores variability introduced by the probabilistic nature of vesicle release and stochasticity in synaptic recovery time . We show that this additional variability has important consequences for the synaptic filtering of presynaptic information . In particular , a synapse model with stochastic vesicle dynamics suppresses information encoded at lower frequencies more than information encoded at higher frequencies , while a model that ignores this stochasticity transfers information encoded at any frequency equally well . This distinction between the two models persists even when large numbers of synaptic contacts are considered . Our study provides strong evidence that the stochastic nature neurotransmitter vesicle dynamics must be considered when analyzing the information flow across a synapse .
Synapses act as information gates in neuronal networks . Presynaptic action potentials are communicated to postsynaptic neurons by causing synaptic neurotransmitter vesicles to release their contents , which then bind to receptors on a postsynaptic neuron's membrane , evoking a transient change in membrane conductance . After a vesicle is released , it typically takes several hundred milliseconds for it to be replaced at a synaptic contact ( see Fig . 1 for a schematic of synaptic release and recovery ) . This refractoriness induces a form of short term synaptic depression that alters the filtering properties of synapses [1] . An accurate description of synaptic vesicle dynamics and their impact of on information transfer is necessary for a thorough understanding of coding in neuronal networks . A widely used model of synaptic depression treats vesicle release and recovery as deterministic processes [2]–[6] . While this deterministic model accurately describes the trial-averaged synaptic response to a presynaptic spike train presented repeatedly to a cell [7]–[11] , it fails to capture the variability introduced at each trial by the probabilistic nature of vesicle release and recovery [12] . Regardless , the model has been used in studies for which neural variability and information transfer are central themes [13]–[18] . The aim of our paper is to determine the impact ( if any ) of stochastic vesicle dynamics on the filtering properties of depressing synapses . Past studies have begun to address this aim by considering how variability from stochastic vesicle release and recovery affects the amount of information transmitted through a synapse as well as the firing rate of a postsynaptic cell [12] , [19] , [20] , but a thorough investigation of the impact of stochastic vesicle dynamics on synaptic filtering has not been performed . We derive a compact description of the filters imposed by short term synaptic depression when stochastic vesicle dynamics are taken into account and when they are ignored . We find that variability introduced by stochastic vesicle dynamics plays a fundamental role in shaping the way in which depressing synapses filter presynaptic information . In particular , a model that ignores this variability transmits presynaptic information encoded at any frequency with the same fidelity [16] , [17] . In contrast , a model that captures this variability reduces overall information transmission , and transmits quickly varying signals with higher fidelity than slowly varying signals . Differences between the two models persist over a broad range of physiologically motivated parameter values , even when a large number of synaptic contacts is considered and even at the population level . Our results suggest important implications for how signals encoded at different timescales are propagated through the nervous system and show that synaptic variability must be taken into account to accurately address such questions .
To gain an intuition for the signal processing properties of depressing synapses , we first study the case of a single Poisson presynaptic spike train , , with constant rate . Since a homogeneous Poisson process has equal power at every frequency , this approach allows us to investigate synaptic filtering at all frequencies simultaneously . Later , we will consider the response to an inhomogeneous Poisson process whose rate encodes a signal . The magnitude of the response of the conductance , , at frequency to fluctuations in the input , , is quantified by the cross-spectrum , , between these quantities ( see Methods ) . For both the deterministic and stochastic models of vesicle dynamics , the cross-spectrum is given by ( see Eq . ( 25 ) in Methods ) ( 2 ) where denotes the Fourier transform and is a kernel that captures the filtering properties of synaptic depression ( see Eq . ( 20 ) in Methods and Fig . 3A ) . The fact that is identical for the stochastic and deterministic models can be understood intuitively by noting that stochasticity in vesicle dynamics is uncorrelated from and therefore does not contribute to the covariability of and . It should be noted that , while Eq . ( 2 ) is exact for the deterministic model , it is an approximation for the stochastic model ( see Methods ) , which is validated by simulations ( Fig . 3B ) . The shape of can be understood by its components in Eq . ( 2 ) . The low-pass filter , , which captures postsynaptic channel dynamics , suppresses power at frequencies higher than ( see Fig . 3A and [29] ) . The high-pass filter , which captures the deterministic dynamics of short term depression , suppresses power at frequencies lower than ( see Fig . 3A , Methods and [17] ) . Their product , which determines through Eq . ( 2 ) , is then band-pass with most of its power at frequencies between and ( Fig . 3B ) . Thus , only fluctuations in the presynaptic input within this frequency band are reflected faithfully by fluctuations in the postsynaptic conductance . The low-frequency limit of is nearly zero for the parameter values chosen in Table 1 ( Fig . 3B ) . This can be explained by noting that the zero-frequency cross-spectrum is related to the gain by [30]For large , the mean conductance saturates and the gain decays to zero like ( see Eq . ( 1 ) and Fig . 2E ) . Thus , which decays to zero for large ( Fig . 4Ai ) . More specifically , when vesicles become depleted , which occurs when release is faster than recovery , i . e . , . Note , though , that is larger for higher frequencies , meaning that faster fluctuations in cause larger transient fluctuations in when compared to changes in the steady state mean conductance , , caused by static changes in [3] , [10] , [27] , [28] . The trial-to-trial and temporal variability of the conductance at frequency is quantified by its power spectrum , , which is given by ( see Eq . ( 25 ) in Methods ) ( 3 ) Here is a constant that represents variability introduced by the interaction of Poisson input with deterministic vesicle dynamics , captures variability introduced by stochastic recovery , and captures variability introduced by probabilistic vesicle release . For the deterministic model , , but and are positive for the stochastic model ( see Methods and Fig . 3C ) . As a result , the stochastic model predicts a larger power spectrum than the deterministic model ( Fig . 3D ) . The decay of at high frequencies is due to the low-pass nature of the synaptic conductance kernel , ( see Fig . 3A and [29] ) . The power spectrum predicted by the two models differs most significantly at low frequencies , where it is nearly zero for the deterministic model but much larger for the stochastic model ( Fig . 3D ) . This can be understood by noting that [30]where is the number of vesicles released in a window of length . For the parameter values in Table 1 , so that vesicles are mostly depleted and therefore the number of vesicles released in a large time window is determined largely by the number of recovery events during that window ( Fig . 2A–D ) . For the stochastic model , recovery events at each contact occur as a Poisson process with rate . Since there are contacts and a Poisson process has power equal to its rate , when is large . This intuition is confirmed by noting that for the stochastic model . In contrast , for the deterministic model , recovery is deterministic and therefore the amount of neurotransmitter taken up , and hence released , over a large time window has a small variance . This is confirmed by noting that for the deterministic model and therefore approaches zero for large . For the synaptic parameters in Table 1 , the power spectra produced by the stochastic and deterministic models disagree for larger than a few Hz ( Fig . 4Aii ) . The fidelity with which fluctuations in the postsynaptic conductance , , reflect fluctuations of the input , , at frequency is quantified by their coherencewhere is the power spectrum of the Poisson input . Since is identical for the two models , but is larger for the stochastic model ( Fig . 3B , D ) , it follows that is smaller for the stochastic model ( Fig . 5 ) . We now investigate the differences between the coherences produced by the two models in more depth . Since for the deterministic model , the cross-spectrum , , and power spectrum , , are proportional to one another ( see Eqs . ( 2 ) and ( 3 ) ) so that dividing them gives a flat coherence ( i . e . , a coherence that does not depend on , Fig . 5 and [16] , [17] ) , Here and in subsequent expressions , a ( ) superscript indicates identities for the deterministic ( stochastic ) model . Synaptic variability in the stochastic model increases the power spectrum , giving a frequency-dependent coherencewhich is high-pass ( Fig . 5 ) . Thus , stochastic vesicle dynamics introduce high-pass frequency dependence into the fidelity of a synaptic filter . In addition to introducing frequency dependence , stochastic vesicle dynamics also decrease the coherence substantially , especially at lower frequencies where the coherence is nearly zero for the stochastic model ( Fig . 5 ) . The fact that coherence is small at low frequencies for the stochastic model can be understood intuitively through the following relation [30] , where is the Pearson correlation coefficient between the number of presynaptic spikes , , and the number of vesicles released , , in a window of length . When , synapses are mostly depleted in the steady state . As a result , the number of vesicles released during a long time interval is determined primarily by the number of recovery events in that time window and hence mostly independent of the number of presynaptic spikes ( Fig . 2A–C and [31] ) . Therefore , for the stochastic model , the number of vesicles released over a long time window is uncorrelated from the number of presynaptic spikes and , as a result , is small . These intuitions are confirmed by appealing to the asymptotic expressions derived for the cross-spectrum and power spectrum above . For the stochastic model , and when . Since for Poisson input , it is then clear thatfor the stochastic model when . For the deterministic model , however , , , and so that approaches a positive constant for sufficiently larger than . For the parameter values in Table 1 , the coherences for the stochastic and deterministic models disagree substantially when is more than a few Hz ( Fig . 4Aiii ) . The disagreement between the stochastic and deterministic models is most dramatic when since the postsynaptic response is determined primarily by vesicle recovery dynamics in this regime , as discussed above . In the figures considered so far , we have used , motivated by measurements of pyramidal–to–pyramidal synapses in rodent neocortex [2] , [19] . However , both shorter and longer time constants have also been reported in cortex [5] , [7] , [8] , [32] , [33] . When other parameters are set to the values from Table 1 , the two models disagree substantially when ( see Fig . 4Bi–iii ) . A proposed justification for using a deterministic model of vesicle dynamics is that stochasticity introduced at each contact averages out when a presynaptic cell makes several contacts [17] . The number , , of contacts a presynaptic cell makes with a single postsynaptic cell varies greatly across cell subtypes and brain regions . Rodent and cat pyramidal cells in the hippocampus and neocortex typically make –12 contacts onto other pyramidal cells or onto interneurons . Interneurons in the same regions make –17 contacts onto pyramidal cells . On the other hand , the Calyx of Held synapse can make more than contacts onto a single postsynaptic target in the rodent auditory brainstem and Purkinje cells can receive over contacts from single presynaptic cells in the rodent cerebellum ( see [34] for values of measured in various animals and synapses ) . When other parameters are set to the values from Table 1 , the stochastic and deterministic models disagree substantially for ( see Fig . 4Ci–iii ) . In summary , over a broad range of synaptic parameters , stochastic vesicle dynamics both attenuate and impart a high-pass nature to the coherence between a pre-synaptic spike train and the post-synaptic conductance response . We next explore the implications of these effects on the transfer of rate-coded information . Time-varying stimuli are often encoded in fluctuations of the firing rate of neuronal populations [35] . To address the question of how information about a rate-coded signal is filtered by vesicle dynamics , we use a model from [16] and [17] in which a time-varying signal is encoded in the firing rate of a presynaptic spike train to yield a doubly stochastic Poisson process , ( see Methods ) . In this model , the instantaneous presynaptic rate conditioned on a signal , , is given by and , without conditioning on , is given by . The power spectrum of the presynaptic spike train is given by ( 4 ) where is the power spectrum of . Eq . ( 4 ) can be interpreted as follows: represents the power of Poisson noise and represents the power of the signal . Unless is identically zero , inherits non-Poisson statistics from , which violates the Poisson assumptions used to derive the spectral properties given above . In the Methods , we derive a linear approximation ( valid when ) to the synaptic filter induced by the deterministic and stochastic models of vesicle dynamics and use it to obtain approximations to the cross-spectrum , , between the signal and conductance as well as the power spectrum , , of the conductance for this model ( see Eqs . ( 27 ) and ( 28 ) in the Methods ) . These approximations allow an investigation of the information transfer of the signal across the synapse in various frequency bands . We model as a Gaussian process with Gaussian-shaped power spectrum ( Fig . 6A , B ) , ( 5 ) where is the bandwidth , the central frequency , and the peak power of the signal . We use a narrow-band signal ( small ) to more clearly illustrate the dependence of synaptic fidelity on signal frequency . Since is Gaussian , there is a positive probability that so that the instantaneous firing rate of the presynaptic cells becomes negative . However , when , this occurs rarely and can be disregarded by considering negative rates as zero [17] . The coherence , , between the signal and the conductance quantifies the fidelity with which the signal , , is represented in the postsynaptic conductance , . For the deterministic model of vesicle dynamics , the coherence is given by ( from Eqs . ( 27 ) ) so that changing merely shifts , but does not change its amplitude ( Fig . 6C , D dashed red line ) . Thus , a signal coded within any frequency band is transmitted with the same fidelity , consistent with the conclusions reached above using the Poisson model and also consistent with previous studies [16] , [17] . For the stochastic model , however , Since is high pass ( Fig . 3A ) and is mostly flat ( Fig . 3B ) , is larger when concentrates its power in higher frequencies . For example , the amplitude of the coherence is larger when than when for the stochastic model , but independent of for the deterministic model ( Fig . 6C , D ) . The rate of linear information transferred from the signal to the conductance is given by [36] , [37]In particular , represents the total information per unit time that a linear decoder can obtain about the signal , , by observing the conductance , , and also represents a lower bound on the Shannon information [36] , [37] . The stochastic model predicts a dramatically lower linear information rate than the deterministic model ( Fig . 7A ) . Since , for the deterministic model , the amplitude of is independent of the central signal frequency , , the linear information rate is also independent of the central frequency ( Fig . 7A ) . The stochastic model , however , transmits quickly varying signals with more fidelity than slowly varying signals ( Fig . 7A ) . Hence , stochastic vesicle dynamics introduce frequency dependence into the transfer of linear information across a synapse . In summary , our results show that the high pass nature of synaptic depression combined with low frequency synaptic noise limits the transfer of low frequency information through a synapse , while higher frequency information is transmitted more reliably . We next investigate these conclusions in a population setting . So far , we have studied the conductance induced by a single presynaptic spike train that makes several contacts onto a postsynaptic cell . However , information about a stimulus is often encoded by populations of several presynaptic cells . We now consider a population model in which a collection , , of presynaptic spike trains all encode the same signal , , as described for the single-cell model above . These inputs induce individual synaptic conductances , , in a single postsynaptic cell . Define the total presynaptic input , , and the conductance induced by this input , . For simplicity , we assume that all synapses have the same synaptic parameters , , , and . The signal , , introduces variability that is shared between the presynaptic spike trains . Such shared variability is commonly referred to as signal correlation since it is informative of the signal . Populations of presynaptic neurons that code for the same stimulus also share non-informative variability , known as noise correlation [38] , [39] . As a simple model of presynaptic noise correlation , we assume that each pair of spike trains , and with , share a proportion of their spike times . The pairwise cross-spectra are then given bywhere represents the contribution of noise correlations and represents the contribution of signal correlations . As we have done for the single input model above , we gain an intuition for the population-level filter imposed by short term depression by first considering purely Poisson spike trains , which is achieved by setting so that . Even though the cross-spectrum , , is identical for the stochastic and deterministic models , the power spectrum , , is larger for the stochastic model due to noise introduced by synaptic variability ( see Fig . 8A , B and Eq . ( 29 ) in Methods ) . Therefore the coherence , , between the total presynaptic signal and the total conductance is smaller for the stochastic model . Moreover , the deterministic model predicts a flat coherence , while the stochastic model predicts a high-pass coherence ( Fig . 8C ) . These conclusions are identical to those reached for a single input above , but the disparity between the two models is reduced at the population level ( compare Figs . 3 and 5 with Fig . 8 ) . Notice also that the power spectrum , , is peaked within the beta frequency band even though the inputs are Poisson and therefore have a flat power spectrum . This effect could exaggerate beta frequencies in recorded data . We return to this topic in the Discussion . A potential justification for using a deterministic model of vesicle dynamics is that , since stochastic release and recovery events are uncorrelated across all synapses , the extra variability introduced by synaptic noise averages out at the population level . So far , we have compared the two models for a population size of . For the parameter values in Table 1 , the low frequency cross-spectrum is identical for the two models , but the coherence and power spectrum disagree considerably until ( Fig . 4Di–iii ) . The value of at which the models begin to agree depends on the pairwise correlation , , between the presynaptic inputs . Notably , in the absence of correlations ( and ) , the population-level coherence is identical to the individual coherences , , so that the coherence predicted by the stochastic and deterministic models disagree by the same amount for any value of ( Fig . 4Diii , lightest lines ) . As increases , the two models agree at smaller population sizes ( Fig . 4Diii , darker lines ) . Hence , presynaptic correlations must be present and must be large if the deterministic model is to be used in place of the stochastic model for large populations . We now study the transfer of rate coded information at the population level by allowing . In particular , we are interested in how information about a rate-coded signal , , is transferred to the population conductance , . As above , we use a signal with Gaussian shaped power spectrum given by Eq . ( 5 ) . A linear approximation to the cross-spectrum , , for this model is calculated in the Methods ( see Eq . ( 29 ) ) , which allows us to calculate the coherence , , between the signal and the postsynaptic response and the linear information rate , , which depends on the central frequency at which the signal is coded in a qualitatively similar manner as for a single presynaptic spike train ( compare Figs . 7A to 7B , C ) . In particular , low frequency information transfer is reduced for the stochastic model of synaptic depression . Moreover , the stochastic model transfers information in a frequency dependent manner and the deterministic model transfers information at all frequencies equally ( Fig . 7 ) . The disparity between the models is substantial when , but reduced considerably when ( compare panels B and C in Fig . 7 ) . We remind the reader that represents the number of presynaptic neurons that encode the shared signal , , which could be much smaller than the total number of presynaptic inputs a cell receives . This suggests that , due to the stochastic nature of vesicle release and recovery , large presynaptic populations must be used to encode slowly varying signals .
There is an extensive experimental and theoretical literature addressing how synapses that exhibit short term depression transmit different patterns of presynaptic spikes [3] , [26] , [27] , [40] , [41] . One recurring observation in these studies is that the steady state mean conductance ( equivalently , the mean rate of vesicle release ) saturates with the presynaptic firing rate , which causes the gain , , to approach zero for large presynaptic rates ( Fig . 2E ) . However , the gain only captures the sensitivity of the steady-state mean , , to static changes in . Previous studies show that temporal changes in are reflected more reliably in the transient mean of than static changes of are reflected in the steady-state mean of [3] , [10] , [27] , [28] . This observation can be understood through our analysis by noting that higher frequency components of are larger than the low-frequency components ( Fig . 3B ) . Note that the decay of at very high frequencies is due to the low-pass properties of the post-synaptic conductance kernel , , ( Fig . 3A and [29] ) and not to synaptic depression . The filtering effects of depression are captured by the kernel , which is high-pass ( Fig . 3A ) . A second shortcoming of the gain as a descriptive quantity is that it does not capture the trial-to-trial variability in the conductance , which is a vital component of information transfer . We quantify this trial-to-trial variability as a function of frequency using the power spectrum , . We show that the frequency-independence of information transfer through a deterministic synapse model depends on the precise shape of [16] , [17] , and the high-pass frequency-dependence of information transfer through a stochastic synapse model likewise depends on the shape of . Furthermore , we show that stochastic vesicle dynamics cause an overall decrease in information transfer by increasing . Thus , trial-to-trial variability in must be considered to obtain an accurate description of information transfer through a synapse . While other studies of synaptic depression have investigated the transfer of rate-coded signals at various frequencies , we are not aware of a study that derives an explicit approximation to the filter induced by a depressing synapse . Such an approximation is derived in the Methods , givingwhere and are the Fourier transforms of the presynaptic spike train and postsynaptic conductance respectively ( see Methods for definitions of other terms ) . This expression can be used to predict the spectral properties of the postsynaptic response to a presynaptic input with a given power spectrum . A generalization of this expression that can be used in the case of a population of correlated presynaptic spike trains is given by Eq . ( 26 ) . For the parameters in Table 1 , the power spectrum is peaked within the beta frequency band ( ) for both the stochastic and deterministic models ( Fig . 8B ) . We emphasize that the presynaptic spike trains in this case are Poisson processes with flat power spectra and cross-spectra . Thus , the peaked power spectrum of the conductance is due completely to synaptic filtering: Frequencies below are suppressed by synaptic depression and frequencies above are suppressed by post-synaptic channel dynamics . The conductance power spectrum is peaked between these two frequencies . This effect could potentially cause an exaggeration of beta or other frequencies in recordings such as local field potentials that reflect large pools of synaptic currents . Parameters can be chosen within a physiologically realistic range to produce a more exaggerated peak than that shown in Fig . 8B or to produce a peak within another frequency band ( not shown ) . Further work is needed to determine the role that synaptic filtering plays in generating or exaggerating rhythms within beta or other frequency bands in functioning neural circuits . We used a simplified model of neurotransmitter release and recovery . In particular , we assumed that each contact contains only one release site . However , individual contacts can have multiple release sites and recent results show that multiple vesicles can be released by a single contact in response to a single presynaptic action potential [22] , [23] . Such situations can be modeled in our framework by interpreting as the total number of release sites at all contacts . However , this interpretation is only valid if the release of vesicles is statistically independent between release sites that share a contact . If the probability of release at one site depends on release at another site – for instance if a contact has several release sites but can only release one vesicle per presynaptic spike [12] , [42] – then our model would need to be adjusted to account for this dependency . To the authors' knowledge , the precise structure of such dependencies are a subject of current research and not presently understood . In the depleted state ( ) , a contact with several release sites will rarely have more than one vesicle available for release at any point in time and our single-vesicle model should provide an accurate approximation regardless of dependencies between release sites , as long as the recovery time constant is properly adjusted [12] . We modeled stochasticity introduced by probabilistic vesicle release and random recovery times , but did not model stochasticity introduced by randomness in the amount of neurotransmitter contained in each vesicle [43] , [44] . In addition we did not model variability at the postsynaptic site ( e . g . , randomness in the number of bound receptors , the number of open channels , or the availability of messenger molecules ) , which could introduce variability in the amplitude of the postsynaptic conductance elicited by each vesicle released . Assuming statistical independence of these sources of variability between release events , they can be captured by multiplying each response amplitude , , by a random number . This would simply scale the power spectrum of the conductance linearly and would not alter our central conclusions . The cross-spectrum between presynaptic input and postsynaptic conductance decays to zero at high frequencies , but the coherence between the two does not ( Figs . 3A and 5 ) . This is due to the fact that the power spectrum also decays at high frequencies and cancels perfectly with the cross-spectrum . However , any additional high frequency noise would destroy this balance . For example , if one were to instead compute the coherence between the presynaptic input and the current across the postsynaptic membrane , high frequency channel noise [45] could increase the power spectrum without increasing the cross-spectrum and therefore cause the coherence to decay at high frequencies . Thus , information transfer from presynaptic input to postsynaptic current is effectively bandpass . Similar observations were discussed in [17] for the deterministic model of vesicle dynamics with additive noise . We used a linear approximation to predict the spectral properties of the postsynaptic conductance induced by non-Poisson presynaptic spike trains . However , the approximation is only assured to be accurate when inputs are approximately Poisson , i . e . , have a nearly flat power spectrum . This restriction is implicit in our assumption that ( see Eq . ( 4 ) and the surrounding discussion ) . Presynaptic spike trains that exhibit highly non-Poisson properties , such as bursts or a high degree of regularity , can interact with synaptic depression in a fundamentally different manner than Poisson spike trains [12] , [46] . Further work is needed to extend our results to highly non-Poisson presynaptic spiking statistics . We focused on short term depression caused by the depletion of synaptic neurotransmitter vesicles . However , other sources of short term depression as well as several forms of short term facilitation affect the filtering properties of synapses [1] , [40] . Our mathematical methods could be extended to take these additional forms of plasticity into account . To quantify information transfer through a synapse , we used an information metric that only captures the amount of information available to a linear decoder observing the conductance . The Shannon information measures the maximum amount of information available to any decoder [47] . Interestingly , for our choice of , the deterministic model of vesicle dynamics transmits Shannon information perfectly because every presynaptic spike elicits a postsynaptic response ( Fig . 2D ) and hence each spike time can be resolved by detecting jumps in [17] , [19] . In contrast , the stochastic model of vesicle dynamics exhibits failures due both to probabilistic release and to vesicle depletion ( Fig . 2C , E ) . Due to the presence of synaptic failure , the stochastic model reduces Shannon information since some presynaptic spikes have no effect on the postsynaptic conductance . A few studies have investigated the reduction of Shannon information through synapses with synaptic failure [20] , [46] , [48] but focus on the impact of probabilistic release and ignore stochasticity in vesicle recovery dynamics . In contrast , we studied the reduction of linear information induced by both probabilistic release and stochastic recovery . The qualitative differences we observed between stochastic and deterministic models depend on the stochasticity of vesicle recovery since it introduces low frequency variability into the conductance ( Fig . 3C , D ) . To our knowledge , only one study [19] has investigated information transmission in a model with both probabilistic release and stochastic recovery . Using simulations , they found that stochastic vesicle dynamics reduce Shannon information by orders of magnitude , consistent with our results for linear information . These previous studies of information transmission do not quantify the dependence of information transfer on the frequency band in which presynaptic information is encoded . Furthermore , care must be taken when drawing conclusions about neural coding from studies of Shannon information . Shannon information quantifies the maximal information that can be extracted by a decoder , but it is not always clear whether a neural decoder can achieve optimal or even near-optimal decoding .
Consider a single presynaptic neuron that fires action potentials at times and define the presynaptic spike train as a point process , where is the Dirac delta function . The number of presynaptic spikes in is then given by . Define to be the number of functional contacts that the presynaptic neuron makes onto a postsynaptic cell [48] and , for simplicity , assume that each contact can have at most one vesicle available for release at any point in time . Let be the total number of vesicles available for release at time . Let be the number of vesicles released by the th presynaptic spike , with . The total number of vesicles released up to time is given by and the effective synaptic input is a marked point process defined by ( 6 ) We first consider a model of synaptic vesicle dynamics that treats vesicle release and recovery stochastically [12] , [19] , [24] , [25] . At each presynaptic spike time , , each contact at which a vesicle is available releases this vesicle independently with probability . After a synaptic contact releases its vesicle , vesicle recovery occurs as a Poisson process with rate . That is , the waiting time from vesicle release until recovery at a single contact is exponentially distributed with mean and independent from the state of other contacts , so that the probability of a recovery event during the interval is . This model can be described by the equation ( 7 ) where is the increment of an inhomogeneous Poisson process with instantaneous rate that depends on through ( here , denotes conditional expectation ) and is given by Eq . ( 6 ) where each is a binomial random variable with mean . Since each trial with a fixed input , , yields a different , random realization of the response , , we hereafter refer to this model as the “stochastic model” of vesicle dynamics . A popular simplification of the stochastic model replaces the random increments , and , in Eq . ( 7 ) with their expected values conditioned on and [2] , [3] , [5] , [6] . Since and , this gives ( 8 ) This model treats as a continuous variable where a proportion of the available vesicles are released at each input and recovery occurs exponentially with time constant . We hereafter refer to the model described by Eq . ( 8 ) as the “deterministic model” of vesicle dynamics since the response , , is determined completely by the presynaptic input , . Stochasticity in this model is only introduced by randomness in . When is a homogeneous Poisson process , the deterministic model is analytically tractable: the first two moments of and can be derived exactly , as we show below . We also show that the first moments agree for two models . The second moments for the stochastic model are difficult to derive analytically , but we derive a more tractable diffusion approximation below . Furthermore , when is not a homogeneous Poisson processes , closed form approximations can be obtained for both the deterministic and stochastic models . Assume that is a homogeneous Poisson process with rate . Then the increment , , is independent from the current value of so that , by taking expectations in Eq . ( 8 ) , for the deterministic model . Similarly , . Combining these gives ( 9 ) Eq . ( 9 ) is also obtained by taking expectations in Eq . ( 7 ) , which implies that the deterministic model and the stochastic model yield the same means when is a homogeneous Poisson process . The following equation for can be obtained using Eq . ( 9 ) and the fact that , ( 10 ) The stationary mean of is given by the unique steady state solution to Eq . ( 9 ) [4] , ( 11 ) Furthermore , after a perturbation of or starting from an initial condition , decays exponentially back to with time constantThe stationary mean number of vesicles released by each presynaptic spike is given by and the stationary mean of the postsynaptic signal is , which represents the steady state rate of vesicle release . Furthermore , approaches its steady state exponentially with the same time constant , , as . The calculations of first moments above depend on the fact that and are independent for any . This can only be assumed to hold when Eq . ( 8 ) is interpreted in the It sense ( so that is updated directly after a spike ) and is a homogeneous Poisson process . If is not a homogeneous Poisson process , then the equations for the first moments are not valid and the first moments may not agree for the two models . Second moments for the stochastic model are difficult to derive analytically , so we obtain approximations by considering a diffusion approximation ( 12 ) where is a standard Wiener process that models stochasticity in vesicle recovery . Stochasticity in vesicle release is captured by the stationary process , , with moments given by , , and for . We assume that , , and are mutually independent . These equations should be interpreted in the It sense , so that the increments and are independent from the history of the noise terms , , for any time [49] . Since , it is clear that the diffusion approximation defined by Eq . ( 12 ) has first moments that satisfy Eq . ( 9 ) . The noise coefficients , and , quantify the degree of randomness introduced by stochastic release and recovery respectively . To find appropriate values for these coefficients , we compute the infinitesimal variance of and conditioned on the drift terms that appear in their respective equations in Eq . ( 12 ) [50] . Since vesicle recovery events are Poissonian , the variance of its increment is equal to its rate , giving the conditional varianceNote that the term that appears on the right hand side of Eq . ( 12 ) does not contribute to this conditional variance since . Conditioned on and the occurrence of a presynaptic spike , the number of vesicles released has a binomial distribution with mean and therefore has conditional variance given byOptimally , we would set and , but doing so would give rise to nonlinear multiplicative noise in Eq . ( 12 ) , which is difficult to treat mathematically . Instead , we obtain an approximation by replacing with its stationary mean , , to obtain ( 13 ) All calculations for the stochastic model are carried out using the diffusion approximation from Eq . ( 12 ) with the noise coefficients from Eq . ( 13 ) , and therefore expressions obtained are approximations to the full stochastic model described above . However , in all figures , simulations are performed using the full stochastic model from Eq . ( 7 ) ( light blue lines ) and show excellent agreement with the closed form approximations ( dark blue lines ) . Note that the deterministic model can be recovered by taking in Eq . ( 12 ) . Thus , we can proceed in our analysis by considering Eq . ( 12 ) without instantiating or to obtain results that apply to both the deterministic and stochastic models . We quantify temporal and trial-to-trial variability between two stationary processes , and , using the cross-covariance function , and its Fourier transform , the cross-spectrum , The cross-covariance ( cross-spectrum ) between a process and itself is called an auto-covariance ( power spectrum ) . To quantify the variability of the postsynaptic response , we now derive the auto-covariance , , and the power spectrum , , for the synapse model in Eq . ( 12 ) . From Eqs . ( 9 ) and ( 10 ) it is apparent that , for , the expectations and decay exponentially to their steady state , given any initial distribution , , imposed on and . From this fact , it is apparent that should inherit this exponential shape and therefore that should have an exponential shape with time constant . We now make this argument more precise using a regression theorem from [49] . Define the bivariate Markov process , Then Eqs . ( 9 ) and ( 10 ) show thatfor whereIn Sec . 3 . 7 . 4 of [49] , it is shown that this impliesfor Solving this linear differential equation givesfor . Thus , due to stationarity , for and where is a constant . By symmetry , we have . Note also that , since is a marked point process , there is a Dirac delta function that contributes to at [51] . Finally , we may conclude that the auto-covariance of has the form ( 14 ) for some constants and . To calculate the coefficients and in Eq . ( 14 ) , we must first calculate a few infinitesimal moments using stochastic calculus techniques [52] . In our calculations , we ignore terms of order and higher , but must include terms of the form order and because their expectation is of the order [50] . The second moment of conditioned on is given by ( 15 ) ( 16 ) where ( 15 ) follows from the fact that and are independent from each other and from , that , and that ; and ( 16 ) follows from the fact that . The calculation of the conditional mixed moment , , is similar and gives To calculate the stationary second moment , , we modify a strategy from Sec . 4 . 4 . 7c of [49] to derive a linear differential equation for the time dependent second moment and find its steady state . First note thatThe first term in this sum is given bywhere we used the fact that and are independent ( see above ) and the last line follows from the equation for derived above . Now calculatewhere we have eliminated terms of order and used the fact that is independent from all other terms; and the last line follows from the equation for above . Combining these expressions gives a differential equation for the time course of the second moment of , where is given by the solution of Eq . ( 9 ) above . The stable fixed point of this linear differential equation is the stationary second moment of , ( 17 ) where is the stationary mean of , given in Eq . ( 11 ) . The delta function in has area given by ( 18 ) where we used Eq . ( 16 ) above and where is given by Eq . ( 17 ) . To calculate the one-sided limit , , first calculatewhere we have used the fact that and are independent of all of the other terms when . Each of the terms in the sum above can be calculated by conditioning on a spike at time and on the value of , where is expectation over the variable . Similarly , where is expectation over . Combining the expressions above givesFinally , since from above , we have ( 19 ) where and are the stationary first and second moments of , given in Eqs . ( 11 ) and ( 17 ) . The auto-covariance of is then given by Eq . ( 14 ) with and given by Eqs . ( 18 ) and ( 19 ) . The power spectrum is obtained from the auto-covariance through a Fourier transform , where ( 20 ) is a deterministic linear kernel , is the noise intensity introduced by the interaction between the stochastic input and deterministic vesicle dynamics , is the noise introduced by stochasticity in vesicle recovery , andis the noise introduced by stochasticity in vesicle release . Note that for the deterministic model since . To measure the covariability between the presynaptic spike trains and the postsynaptic response , we now derive the cross-covariance between the input , , and the response . By a similar argument to the one made above for , we may conclude that is the sum of a delta function and an exponential , except that the exponential is one-sided since for . For , we can find the peak of the exponential by first conditioning on a spike at time , then conditioning on a spike at time , since . Thus , The area of the delta function in is given bysince . Thus , we havewhere is the Heaviside step function . Taking the Fourier transform gives the cross-spectrumwhere is defined in Eq . ( 20 ) above . The statistics of the postsynaptic response to a population , , of uncorrelated presynaptic spike trains can be easily calculated from the statistics of individual responses , which are calculated above . However , neurons that contact a shared postsynaptic cell often exhibit correlations between their spiking activity [39] , [53] . To determine the postsynaptic response to a population of correlated presynaptic spike trains , we must first calculate the pairwise cross-spectra of the conductances induced by these inputs . Assume that each spike train , , in the presynaptic population is a Poisson process with rate . Introduce correlations by assuming that each pair , and , of spike trains share a proportion of their spike times so that [54] . We use subscripts to denote quantities associated with each spike train and double subscripts as necessary . For simplicity , assume that the synaptic parameters , , and are identical for all synapses . The asymmetric case can be treated identically , but the expressions obtained are more cumbersome . The power spectrum , , and the cross-spectrum , , are given above ( where they are written as and ) . Below , we derive expressions for and for . First , following the same arguments used above to derive the moments of and in the case of a single presynaptic spike train , we obtain the bivariate momentsSimilarly , and , equivalently , We now derive a differential equation for to get the stationary second moment . First note that so that ( 21 ) By symmetry , the first and second terms in Eq . ( 21 ) are the same and they can be derived from Eq . ( 12 ) asThe last term in Eq . ( 21 ) is given byCombining these giveswhich has a fixed point at ( 22 ) We now calculate the cross-covariance between and . By a similar argument to that used to derive Eq . ( 14 ) above , the cross-covariance between and has the form ( 23 ) where we have used the symmetry of and , inherited from the symmetry in parameters , to conclude that . The area of the delta function is given bywhere is given in Eq . ( 22 ) . To find , we first calculateso thatwhich gives through Eqs . ( 22 ) and ( 23 ) . Finally , we will derive and . Once again , by linearity , each of these is the sum of a delta function and an exponential . The area of the delta function is given byWe also haveThus , and thereforeBy symmetry , Finally , the cross-spectra can now be found through a Fourier transform to obtainwhere ( 24 ) So far we have described the statistics of the processes , , which quantify the release of vesicles released over time . The postsynaptic conductance induced by vesicle release is then defined as where denotes convolution and represents the time course of conductance induced by the release of a single vesicle ( with for ) . The statistics of can easily be derived from those of using standard signal processing identities [29] to give ( 25 ) for and the steady state variance of is given by . So far , we have discussed statistics of the conductance induced by a population of homogeneous Poisson presynaptic spike trains , but spike trains measured in vivo do not always exhibit homogeneous Poisson statistics [55] . For example , time-varying stimuli can induce fluctuations in the firing rate of presynaptic neurons . As a simple model of rate-coded signals , we assume that a shared , time-varying signal , , is encoded in the firing rates of a presynaptic population , . In this model , each presynaptic spike train is a doubly stochastic Poisson process [51] . The instantaneous firing rate of each presynaptic neuron , conditioned on , is given by . Without loss of generality , we assume that the signal has zero bias , , so that the unconditioned firing rates are . Signal correlations are introduced in this model by the shared signal , . We include noise correlations , i . e . , correlations that are not due to shared signal [38] , [39] , by assuming each pair of presynaptic spike trains share a proportion of their spike times . To compute the auto- and cross-covariance functions we first note that , for , where is distribution of in the steady state ( ) . In addition , has a Dirac delta function at with mass equal to the rate of synchronous spikes , . Thus , for . The auto-covariance ( ) can be obtained by taking . The cross-covariance function between and is be computed similarly to obtain . Taking Fourier transforms gives the spectra , where is the power spectrum of the signal . Exact expressions for the statistics of the postsynaptic conductance are difficult to obtain for this inhomogeneous Poisson model because is correlated with and with , which invalidates the methods used in the derivations for the homogeneous Poisson model above . However , when , the firing rate inhomogeneities are weak compared to the background firing rate and temporal correlations are weak as a result ( analogously , ) . In this case , a linear approximation to the synaptic response can be obtained . To obtain this approximation , we find a linear filter that maps presynaptic spike trains to conductances and that is consistent with Eqs . ( 25 ) when inputs are Poisson . The following filter satisfies this requirement ( 26 ) Here , is standard Gaussian white noise , is unbiased stationary noise with power spectrum that accounts for stochasticity in vesicle recovery , and similarly for , which accounts for stochastic vesicle release . The noise terms and are zero for the deterministic model . All noise terms here are independent except that and are correlated with cross-spectrumwhere is given by Eq . ( 24 ) . The spectra predicted by Eq . ( 26 ) can be easily calculated using the fact that for stationary processes , and , where and denotes complex conjugation [56] . Thus , where we used the independence of the noise sources to eliminate several terms . Other spectra can be derived in a similar manner to obtain the following generalizations of Eqs . ( 25 ) ( 27 ) for . These expressions agree with Eqs . ( 25 ) when inputs are Poisson , i . e . , when , because and in this case . When and , these expressions give a linear approximation which is verified using simulations in several figures below . The fidelity with which the signal , , is represented in the conductances , , depends on the cross-spectrum which can be calculated in analogous manner to above to obtain ( 28 ) We are especially interested in the population spectra , , and , where is the total presynaptic input and is the total conductance induced by . These are given by using the bilinearity of covariances to obtain ( 29 ) A similar inhomogeneous Poisson input model was used in [17] to investigate the transfer of rate-coded signals for the deterministic model of synaptic depression . Their model is analogous to our deterministic model with ( since their response amplitudes are normalized ) and ( since they consider the postsynaptic response , , before convolution with a conductance kernel ) . Under these substitutions , our expression for agrees with their expression for exactly ( where we use an “” superscript to indicate expressions from [17] ) . However , our expression for for the deterministic model only agrees with their expression for when ( i . e . , when the input is a homogeneous Poisson process ) . Our expression has an additional term that accounts for power introduced by the signal . In particular , for the deterministic model when and . Theoretical results are obtained for arbitrary parameter values , but for all figures we use the parameters from Table 1 , which are chosen to represent values from experimental studies . The values used for and have been deemed “typical” for pyramidal-to-pyramidal synapses in the rodent neocortex [2] , [19] and the value of is typical for several cortical areas [34] . The form of is chosen to model AMPA dynamics and its units are rescaled so that . This rescaling simplifies the exposition in the Results .
|
Neurons communicate through electro-chemical connections called synapses . Action potentials in a presynaptic neuron cause neurotransmitter vesicles to release their contents which then bind to nearby receptors on a postsynaptic neuron's membrane , transiently altering its conductance . After it is released , the replacement of a neurotransmitter vesicle takes time and the depletion of vesicles can prevent subsequent action potentials from eliciting a postsynaptic response , an effect that represents a form of short term synaptic depression . When a vesicle is available for release , an action potential elicits its release probabilistically and depleted vesicles are replenished randomly in time , making the transmission of presynaptic signals inherently unreliable . We analyze a mathematical model of vesicle release and recovery to understand how signals encoded in sequences of presynaptic action potentials are reflected in the fluctuations of a postsynaptic neuron's conductance . We find that slow modulations in the rate of presynaptic action potentials are more difficult for a postsynaptic neuron to detect than faster modulations . This phenomenon is only observed when randomness in vesicle release and replacement is taken into account . Thus , by including stochasticity in the workings of synaptic dynamics we give new qualitative understanding to how information is transferred in the nervous system .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"synapses",
"computational",
"neuroscience",
"single",
"neuron",
"function",
"biology",
"neuroscience",
"neurophysiology",
"coding",
"mechanisms"
] |
2012
|
Short Term Synaptic Depression Imposes a Frequency Dependent Filter on Synaptic Information Transfer
|
Chagas disease , caused by the protozoan Trypanosoma cruzi , is the leading cause of heart failure in Latin America . The clinical treatment of Chagas disease is limited to two 60 year-old drugs , nifurtimox and benznidazole , that have variable efficacy against different strains of the parasite and may lead to severe side effects . CYP51 is an enzyme in the sterol biosynthesis pathway that has been exploited for the development of therapeutics for fungal and parasitic infections . In a target-based drug discovery program guided by x-ray crystallography , we identified the 4-aminopyridyl-based series of CYP51 inhibitors as being efficacious versus T . cruzi in vitro; two of the most potent leads , 9 and 12 , have now been evaluated for toxicity and efficacy in mice . Both acute and chronic animal models infected with wild type or transgenic T . cruzi strains were evaluated . There was no evidence of toxicity in the 28-day dosing study of uninfected animals , as judged by the monitoring of multiple serum and histological parameters . In two acute models of Chagas disease , 9 and 12 drastically reduced parasitemia , increased survival of mice , and prevented liver and heart injury . None of the compounds produced long term sterile cure . In the less severe acute model using the transgenic CL-Brenner strain of T . cruzi , parasitemia relapsed upon drug withdrawal . In the chronic model , parasitemia fell to a background level and , as evidenced by the bioluminescence detection of T . cruzi expressing the red-shifted luciferase marker , mice remained negative for 4 weeks after drug withdrawal . Two immunosuppression cycles with cyclophosphamide were required to re-activate the parasites . Although no sterile cure was achieved , the suppression of parasitemia in acutely infected mice resulted in drastically reduced inflammation in the heart . The positive outcomes achieved in the absence of sterile cure suggest that the target product profile in anti-Chagasic drug discovery should be revised in favor of safe re-administration of the medication during the lifespan of a Chagas disease patient . A medication that reduces parasite burden may halt or slow progression of cardiomyopathy and therefore improve both life expectancy and quality of life .
Chagas disease afflicts about 7 million people in South and Central America [1] , where it is the leading cause of heart failure . More than 10 , 000 deaths are estimated to occur annually due to this disease . Despite joint efforts in Latin America to eradicate the transmission of Trypanosoma cruzi through screening of blood banks and control of triatomine vectors , Chagas disease presents a risk to 70 million people living in endemic countries [1 , 2] . International travel , infected blood transfusions , co-infection with HIV , mother to fetus transmission , and northward migration of the “kissing bug” insect vector [3] , all help to drive up the number of cases and push the incidence outside its historic range . Chagas disease is now seen in Europe , North America and Asia and seems set to become an urgent public health issues in countries far beyond its focal source in South America [4 , 5] . An annual economic burden due to Chagas disease , calculated by simulation models as overall cost , reaches 7 . 19 billion US dollars , largely from the loss of productivity and premature mortality caused by cardiomyopathy [6 , 7] . Human infections by T . cruzi result in a significant mortality rate in children in the acute phase , or may lead to cardiomyopathy in chronically infected adults [8 , 9] . About 40% of infected individuals develop chronic manifestations of the disease: ten percent of patients develop gastrointestinal symptoms ( e . g . , mega colon and mega esophagus ) ; and 30% of patients develop cardiac disease characterized by cardiomyopathy , arrhythmias and interstitial fibrosis accompanied by cardiac inflammation[9] . The clinical treatment of Chagas disease is limited to two drugs: nifurtimox and benznidazole , developed about 60 years ago . Nifurtimox is now discontinued in several countries [10 , 11] , while benznidazole has been recently FDA-approved only for use in children of 2 to 12 years old [12 , 13] . Both benznidazole and nifurtimox are about 80% effective in the acute stage of Chagas disease [14] . Limitations of current therapy include variable efficacy against T . cruzi of different genetic backgrounds and elevated toxicity with severe side effects , including widespread dermatitis , digestive intolerance , polyneuritis and bone marrow depression , leading to poor patient compliance [10 , 11] . Both drugs are used in cases of new infections , congenital infections , reactivation and/or re-aggravation associated with immunosuppression and as a preventive measure against laboratory accidents [8] . The efficacy of benznidazole against the more prevalent chronic stage of Chagas dissease was investigated in the BENEFIT clinical trial [15 , 16]—the first randomized , placebo controlled , clinical study on the effects of benznidazole on the clinical progression of chronic Chagas disease patients with compromised cardiac function . Drug treatment led to a marked reduction of the circulating parasite load in patients from Brazil ( strain TcII ) and Argentina and Bolivia ( strains TcV and TcVI ) , but not in patients from Colombia or El Salvador ( strain TcI ) . In all cases , benznidazole failed to reduce cardiac function deterioration when evaluated at the 5–7 year follow-up . [15] . These results suggest that benznidazole has limited clinical utility in patients with moderate to advanced cardiac compromise ( class I or II heart failure , New York Heart Association terminology ) . However , an important qualification is that previous observational , not randomized , studies [17] suggest that the drug is effective in patients in the asymptomatic ( indeterminate ) stage or those with incipient cardiac compromise . A combination of benznidazole with posaconazole in the treatment of asymptomatic patients ( the STOP-Chagas clinical trial ) also showed no advantage over benznidazole monotherapy , as judged by the PCR test alone . Clinical disease as evidenced by decreased cardiac function or other cardiomyopathy signs were not assessed in this study [18] . In either case , there is an urgent need for safer and more efficacious drugs and drug combinations to meet the etiological challenges of this complex disease . As an alternative to the use of benznidazole in patients with chronic Chagas disease [19] , significant efforts have been made to repurpose antifungal azole drugs targeting sterol biosynthesis . Among validated sterol biosynthetic targets , CYP51 is one of the most extensively exploited for the development of new therapeutics for fungal and parasitic infections [20 , 21] . The CYP51 inhibitors posaconazole ( Noxafil , Merck ) and ravuconazole ( E1224 , Eisai , Tokyo ) , which have undergone extensive pharmacological and toxicological optimization in antifungal programs , have demonstrated efficacy and curative activity in animal models of Chagas disease [22] , and alleviated chronic Chagas disease in a patient with systemic lupus erythematosus [23 , 24] . Both drugs have been tested in controlled clinical trials for Chagas disease [18 , 25 , 26] . The perceived inferiority of both drugs to the current standard-of-care drug , benznidazole , [25 , 27] was due to their failure to produce sterile cure ( PCR negative ) , and triggered discussions in the Chagas research community about the validity of CYP51 as a target [28 , 29] . Two concerns have been expressed: ( i ) differential activity of CYP51 inhibitors against different strains of T . cruzi or between the replicative ( amastigote ) and non-replicative ( trypomastigote ) stages of the parasite and ( ii ) the slow-acting mechanism of CYP51 versus fast-acting benznidazole [28–30] . A third factor that may have affected the outcomes of the clinical trials is that the repurposed antifungal drugs , including posaconazole and ravuconazole , were not optimized to target T . cruzi CYP51 . In parallel with the clinical trials , a number of laboratories pursued novel chemical scaffolds specifically targeting T . cruzi CYP51 ( reviewed in [21] ) . Using a target-based structure-aided drug discovery approach , a 4-aminopyridyl-based scaffold was identified as efficacious and further developed into a series of lead compounds active against T . cruzi both in vitro and in vivo [31–37] ( Fig 1 ) . Two optimized leads of the series , 9 [37] and 12 [35] ( Fig 1 , compound numbers correspond to those in the cited references ) , have now been evaluated for both toxicity and parasitological cure in the acute and chronic animal models of T . cruzi infection . Although a sterile cure was not achieved , 9 and 12 were proven safe for long term administration in mice and suppressed parasitemia in both the acute and chronic phases . In the acute model , these lead compounds improved survival , protected mice from hepatic injury and drastically reduced cardiac inflammation . In the chronic phase , these lead compounds prevented spontaneous T . cruzi relapse for up to 4 weeks post-treatment .
The No Observed Adverse Effect Level ( NOAEL ) , of 9 and 12 was evaluated according to the Organization for Economic Cooperation and Development ( OECD ) guidelines . Escalating doses of 12 were administered orally to male and female Swiss mice every hour; adverse effects were observed only at concentrations higher than 300 mg/kg for male and 250 mg/kg for female mice . Cumulative in vivo effects were analyzed using uninfected BALB/c male and female mice treated with 9 or 12 at 25 mg/kg orally for up to 28 days , b . i . d . No adverse clinical signs ( such as ruffled fur , hunched posture , reduced mobility , or tremor ) or alteration in general health were observed in any of the mice . Blood was collected after the end of treatment and serum was evaluated in a chemistry panel that included liver enzymes and markers of renal function . No alteration of blood levels for alanine aminotransferase ( ALT ) , aspartate aminotransferase ( AST ) , bilirubin ( BIL ) , albumin ( ALB ) , blood urea nitrogen ( BUN ) or creatinine ( CRE ) was detected after the course of treatment ( Fig 2 ) . Histological analysis of brain , heart , liver , kidney , GI tract and lungs did not show any alteration of tissue morphology . The weights of the animals remained steady throughout the treatment . Since no toxicity was detected in uninfected mice , we performed a 28-day oral treatment at 25 mg/kg of compounds in Swiss male mice infected with T . cruzi Y strain ( 104 inoculum ) , an established model of acute infection recommended for drug screening and development by the Fiocruz Program for Research and Technological Development on Chagas Disease ( PIDC/Fiocruz ) and the Drugs for Neglected Diseases Initiative ( DNDi ) [38] . Treatment with 9 or 12 at 25 mg/kg significantly reduced parasitemia , reaching the minimum limit of detection by the Pizzi-Brener method at 9 days post infection ( dpi ) , with inhibition levels of 99 . 9% for 9 and 99 . 3% for 12 . Parasitemia remained undetectable in the treated mice , while untreated and vehicle-treated mice showed high parasitemia and all of these mice died by 18 dpi ( Fig 3A ) . While parasites were not detected , only 20% of the infected mice treated with 9 and 80% of the infected mice treated with 12 survived the entire 30 day study ( Fig 3B ) . 100% of benznidazole-treated mice survived . All mice were euthanized at the end of treatment . Compared to the negative controls , treatment with the test compounds did provide partial protection and delayed the death of the mice . Death of the treated animals in this model of acute disease with Y strain parasites may be due to yet unknown T . cruzi Y strain-specific factor ( s ) that could interfere with the compound effect in the infected mice . That effect with Y strain parasites was not seen in uninfected animals or animals infected with T . cruzi CL-luc strain . Upon study completion , mice were euthanized and serum was assessed for hepatic enzymes and renal function markers ( Fig 3C–3F ) . Data from the blood chemistry analysis showed that T . cruzi Y infection induced elevated serum levels of alanine aminotransferase ( ALT ) and aspartate aminotransferase ( AST ) , indicating infection-induced liver injury . The animals treated with 9 or 12 had reduced levels of these enzymes when compared to untreated controls . Urea and creatinine ( CRE ) levels , markers of renal function , were also analyzed; treated animals show normal levels similar to that of untreated controls , suggesting no renal toxicity was caused by 9 or 12 . Given that neither 9 nor 12 produced sterile cure in the T . cruzi Y-infected animals , we next evaluated their effect in a less severe acute model using the transgenic CL-Brenner T . cruzi strain expressing “red-shifted” luciferase ( CL-luc , a gift from Dr . John Kelly , UK ) . This strain carries a stable bioluminescent marker , which allows one to detect live parasites in tissues of a live mouse with a sensitivity limit exceeding that of the RT-PCR method for up to a year after infection [39 , 40] . Since mice gender influences the level of infection [41 , 42] , we used both male and female mice for the 28 day b . i . d . treatment beginning 14 dpi . Males not treated with experimental or reference compounds showed levels of parasitemia higher than females ( Figs 4 , 5 and 6 ) . As judged by the bioluminescence ( Fig 4 ) , the level of CL-luc T . cruzi infection markedly decreased at 18 dpi , which corresponds to 4 days of treatment with the experimental or reference compounds , benznidazole or posaconazole . By 32 dpi ( 18 days of treatment ) , the parasite bioluminescence in all treated mice was reduced to the background level , with the photon count lower than that of the uninfected mice injected with luciferin . Both groups of negative control mice , including a group of infected but untreated mice , as well as a group of infected and vehicle-treated mice , had bioluminescence levels two orders of magnitude higher than the uninfected control mice . This trend was maintained up to 49 dpi ( 7 days after the end of treatment ) . However , by 53 dpi ( 11 days after the end of treatment ) animals from the benznidazole ( male and female ) , 12 ( female ) , and 9 ( male and female ) groups all showed resurgence of parasites ( Figs 4 , 5 and 6 ) . Posaconazole treated mice showed resurgence of parasites at 67 dpi . By 81 dpi , the majority of the animals in the 9- and 12-treated groups were T . cruzi-positive as indicated by bioluminescence . All animals in the male group treated with 9 showed parasitemia as high as untreated controls at 81 days post infection . Both compounds reduced the parasite load in female mice , where bioluminescence levels were 10× lower than untreated controls in 4 out of 5 mice for both compounds at 81 dpi , with one 9-treated female negative ( Table 1 ) . Posaconazole reduced T . cruzi bioluminescence to undetectable levels in 1 out of 5 males and 2 out of 4 females , with the remaining mice showing reactivation of parasites . Posaconazole markedly reduced parasite load compared to the untreated controls . Finally , benznidazole was not able to suppress parasites in all the mice , with one female showing circulating parasites after treatment withdrawal ( Figs 4 , 5 and 6 , Table 1 ) . Ex vivo bioluminescence imaging of internal organs was performed at 81 dpi . As previously described for this model [39 , 43] , parasites were detected consistently in the gastro-intestinal ( GI ) tract in all T . cruzi-positive mice ( Figs 7 and 8 ) . Moreover , bioluminescence above the background level was also observed randomly in heart , skeletal muscles , liver and mesenteric fat in T . cruzi-positive mice throughout the groups , suggesting a dynamic infection process ( Fig 7 ) . Ex vivo quantification of the parasite burden in the organs analyzed ( heart , liver , kidneys , lungs , spleen , gastro-intestinal tract , skeletal muscle and mesenteric fat ) was consistent with the whole-mouse imaging . One female mouse treated with benznidazole showed traces of infection in the GI tract , while posaconazole-treated animals revealed T . cruzi in the GI tract and lung . For the experimental inhibitors , no parasites were detected in two females treated with 12 and one female treated with 9 . All other animals in the 9- or 12-treated groups showed T . cruzi bioluminescence in GI tract , liver , lung and mesenteric fat ( Figs 7 and 8 ) . Tissue architecture and inflammation in the heart was evaluated through conventional histology and H&E staining from both lethal acute and bioluminescent models , with the levels of inflammation being quantified using FIJI software [44] . Uninfected mice had normal cardiac tissue , as expected ( Fig 9D ) . T . cruzi infection in untreated mice resulted in marked inflammation in heart tissue with inflammatory infiltrates and interstitial fibrosis in both acute models ( Fig 9A and 9F ) . BALB/c mice infected with CL-luc showed mild cardiac inflammation when compared to Swiss mice infected with T . cruzi Y strain; the latter showed levels of inflammation at least 2× higher than the former and included the presence of amastigote nests , which were not observed in BALB/c mice infected with CL-luc ( Fig 9A , 9F , 9E and 9J ) . In both models , mice treated with benznidazole ( Fig 9B and 9G ) , 9 ( Fig 9C and 9H ) , or 12 ( Fig 9I ) had normal heart tissue , with a significant reduction of inflammatory cells ( Fig 9E and 9J ) compared to vehicle treated controls , and no signs of interstitial fibrosis or amastigote nests , suggesting that the reduction of parasite load induced by treatment with the CYP51 inhibitors improved cardiac pathology , even without sterile cure . Most patients in need of treatment are in the chronic phase of Chagas disease . In this later stage , parasite load is low enough to require sensitive techniques for parasite detection [45] . To recapitulate these conditions , we evaluated performance of 9 in a chronic mouse model . Because males are more susceptible to infection than females , BALB/c males were infected with T . cruzi CL-luc strain allowing highly sensitive bioluminescence detection . Following the treatment scheme reported [46] , the compounds were administered at 25 mg/kg for 28 days starting at 126 dpi , when chronic infection was established and the parasite signal was consistently detected . Similar to the acute model described above , T . cruzi bioluminescence levels dropped soon after the start of treatment . After 28 days , all treated groups , including 9 , showed only background luminescence ( Fig 10A–10C ) . Mice were then followed for 4 weeks after compound administration had ceased . For up to 27 days post-treatment , groups treated with posaconazole or 9 had bioluminescence levels slightly above the background defined by the uninfected mice and 10–100× lower than the untreated controls ( Fig 10D and 10E ) . Since the parasite load was below the detection level 4 weeks post-treatment , the animals were immunossupressed with cyclophosphamide . After 2 rounds of immunossupression , parasites relapsed as evidenced by the bioluminescence levels similar to those of the untreated controls .
A similarity in sterol biosynthesis pathways between T . cruzi and fungi is that both produce ergosterol and ergosterol-like sterols as membrane building blocks [47] . This similarity encouraged the application of antifungal drugs for the treatment of Chagas disease . However , in human clinical trials for Chagas disease , both posaconazole and ravuconazole failed to demonstrate superiority to the current standard-of-care drug , benznidazole , using PCR as a marker of continued or reactivated T . cruzi infection [25 , 27] . The failure of posaconazole and ravuconazole to attain sterile cure in humans raised concerns about the CYP51 target . Differential activity of CYP51 inhibitors against the replicative ( amastigote ) and non-replicative ( trypomastigote ) stages of T . cruzi , a slow-acting mechanism of action , and the stochastic nature of T . cruzi infection with the non-replicating or rarely-replicating cryptic amastigotes ‘hidden’ inside the tissues [39] , were listed as potential drawbacks of the CYP51 target [28 , 48] . On the other hand , it has been argued that both the dose and the duration of anti-fungal agents used in the clinical trials to treat human T . cruzi infection have been suboptimal . Urbina et al . noted that the plasma exposure in patients for the dose used in clinical trials corresponds to 10–20% of the curative dose in mice [29 , 49] . The post-clinical trial tendency to balance risks in the Chagas drug discovery portfolio , and to identify drug candidates aimed at other molecular targets is logical . At the same time , it is critical to not reject promising targets based on clinical studies with drugs not properly optimized , dosed , or clinically evaluated . An important , but overlooked factor , that may have affected the performance of the anti-fungal drugs in Chagasic patients , is the loose drug-target fit demonstrated by posaconazole in co-crystal structures with Trypanosome CYP51 . The electron density of the bound drug is poorly defined and its pendant phenyl-2-hydroxy-pentantriazolone group adopts alternative conformations to make multiple interactions outside of the active site [50 , 51] . Several novel CYP51 inhibitors developed in this collaboration [33–37] and elsewhere [51–53] demonstrated drug-target fits superior to posaconazole . The most potent 4-aminopyridyl-based inhibitors of the 4-aminopyridyl-based series bind entirely in the CYP51 target interior , making tight interactions with hydrophobic residues constituting the CYP51 active site [35 , 37] . Improved drug-target interactions may be responsible , at least in part , for the superior potency of the experimental inhibitors in the acute and chronic mouse models of infection [54] . Neither of the two lead compounds of the 4-aminopyridyl-based series evaluated in these studies attained a sterile cure . However , both leads were proven safe for long term administration in mice and , efficiently suppressed parasitemia in both acute and chronic models of infection . In the acute phase , compounds improved survival in highly stringent acute mouse models , protected mice from hepatic injury , and drastically reduced acute cardiac inflammation . In a model of chronic Chagas disease , 9 prevented spontaneous T . cruzi relapse for up to 4 weeks post-treatment . Similar results—supression of parasitemia , no spontaneous relapse after treatment withdrawal and parasite reactivation after immunossupression—were also achieved by other research groups using CYP51 inhibitors based on different molecular scaffolds [41 , 46 , 55] . Collectively , 9 is more efficacious in the treatment of the chronic phase of the disease with low parasite load . Although 9 did not eradicate cryptic reservoirs of parasites in vivo after 28 days of treatment , it successfully kept parasites under control and prevented the inflammation responsible , in part , for cardiac tissue damage . This outcome is not unique for inhibitors that target the ergosterol synthesis pathway . Treatment of chronically infected mice with N , N-dimethylsphingosine , an inhibitor of sphingosine kinase , also failed to produce a sterile cure , but reduced parasite load leading to a marked decrease in inflammation and fibrosis . Furthermore , there was a reduction of inflammatory mediators and an improvement of heart function measured as exercise capacity [56] . In addition , mice infected with resistant strains of T . cruzi showed decreased tissue parasitemia , reduced myocarditis and less electrocardiographical alterations after treatment with benznidazole , even though the drug failed to completely eliminate parasites in this model [57] . Several mechanisms have been proposed to explain the pathogenesis of Chagas’ cardiomyopathy , including parasite-dependent inflammation , autoimmunity , autonomic neuronal degeneration and damage of microvasculature [58 , 59] . Although more than one mechanism may be involved in Chagas disease pathogenesis , a consensus is that tissue damage is related to parasite persistence [58–60] . At the same time , 60–70% of infected individuals are asymptomatic . A balance between host and parasite in asymptomatic cases may be maintained by expression of the anti-inflammatory cytokine IL-10 , while cardiomyopathy is associated with inflammation triggered by IFN-gamma and TNF-alpha [61] . Reducing the parasite burden diminishes inflammation even without complete elimination of the parasite [60] . In this regard , several non-randomized clinical trials have shown that etiological treatment of chronic patients with benznidazole resulted in slower progression to advanced stages of cardiomyopathy evaluated by electrocardiography and echocardiography [17 , 62] . Sterile cure is a highly desirable treatment outcome , however , it may not always be achieved , and drug discovery efforts are often hampered by deficiencies in understanding the nuances of disease pathogenesis . There is currently no sterile cure of HIV infection . The desirable outcome of the antiretroviral treatment is a long term plasma HIV-RNA count below 50 copies/ml [63] . The WHO recommends antiretrovirals in people of all ages , including pregnant women as soon as the diagnosis is made; once treatment is begun , it is recommended to continue throughout the entire life span without interruptions [63] . Benefits of treatment include a decreased risk of progression to AIDS and a decreased risk of death [64] . Highly active antiviral therapy options are available as drug ‘cocktails’ consisting of at least three medications belonging to at least two different classes of antiviral agents [65] . As of 2017 , 19 . 5 million people are accessing antiretroviral therapy and more than half of all people living with HIV are on treatment [66] . By analogy with HIV/AIDS , the treatment option of non-toxic medications should be developed for Chagas patients to slow down progression to cardiomyopathy and to improve life expectancy and quality of life . With the scarce arsenal of anti-T . cruzi agents , the drug discovery community cannot afford to be prejudiced against CYP51 , or any other target , if the inhibitors have acceptable safety profiles and achieve a marked reduction in parasite load , even in the absence of sterile cure . Regardless of the molecular target affected by the drug , development of an efficacious and safe treatment for Chagas disease would be a breakthrough for society , medicine and science .
Research performed at UC San Diego was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adheres to the principles stated in the Guide for the Care and Use of Laboratory Animals , National Research Council , 2011 . The facility where this research was conducted is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International . Animal research was conducted under approved protocol S14187 from the Institutional Animal Care and Use Committee , University of California , San Diego . Research performed at Oswaldo Cruz Foundation—FIOCRUZ , Rio de Janeiro , Brazil , was approved by the Committee for Ethics in the Use of Animals of FIOCRUZ , under protocol number LW-37/13 and is in compliance with Brazilian Federal Law number 11794/08 , Federal Brazilian Decree number 6899/09 and Brazilian Normative Resolution number 1 ( July 9th , 2010 ) of the National Council for the Control of Animal Experimentation . Euthanasia was accomplished by CO2 inhalation or by sodium pentobarbital overdose ( 60 mg/kg ) , followed by cervical dislocation . These methods of euthanasia have been selected because they cause minimal pain and distress to animals , are relatively quick , and do not adversely impact interpretation of the results of studies . All methods are in accord with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association . Compounds 9 and 12 were synthesized by following the procedures previously reported [35 , 37] . In vivo experiments were performed at the University of California San Diego ( UCSD ) , La Jolla , California , USA and Oswaldo Cruz Foundation—FIOCRUZ , Rio de Janeiro , Brazil . At FIOCRUZ , Swiss Webster male and female mice weighting 18–20 g were obtained from CEMIB ( Centro Multidisciplinar para Investigação Biológica ) , UNICAMP ( Campinas , SP , Brazil ) . At UCSD , male and female 6 weeks old BALB/c mice , in the same weight range , were purchased from Jackson Laboratories ( Farmington , CT , USA ) . Mice were housed in a maximum number of 5 animals per cage and kept in a conventional room at 20 to 24°C under a 12 h/12 h light/dark cycle . The animals were provided with sterilized water and chow ad libitum . Acute toxicity was evaluated by administration of escalating doses of the compounds to male and female Swiss Webster mice ( n = 2/group ) orally by gavage , at 100 ul/hour 50 mg/kg dose formulated in 20% solutol ( also known as Kolliphor HS15 ) ( Sigma #42966 ) . The general health of the animals was closely monitored for up to 48 h and the last dose before the onset of toxic symptoms were observed was defined as NOAEL according to the OECD guidelines . Cumulative toxicity after prolonged treatment was evaluated using BALB/c females ( n = 5/group ) treated orally by gavage with the experimental CYP51 inhibitors , 9 ( 25 mg/kg ) or 12 ( 25 mg/kg ) , dissolved in 10% solutol , at 100 ul/dose , b . i . d , for 28 consecutive days . Mice were weighed once a week and their general health was assessed daily . After treatment , mice were euthanized , blood was collected for analysis of several biochemical markers of general health . Brain , heart , liver , kidney , gastro-intestinal ( GI ) tract and lungs were collected , briefly rinsed in PBS and fixed in buffered formalin solution including 10% formaldehyde , 33 mM NaH2PO4 , 45 mM Na2HPO4 for histological evaluation . The T . cruzi Y parasites were obtained from the bloodstream of infected Swiss Webster mice at the peak of parasitemia , as previously described [67] . Transgenic T . cruzi CL Brener parasites expressing a red-shifted luciferase that emits light in the tissue-penetrating orange-red region of the spectrum ( a gift from Dr . John Kelly , London School of Hygiene and Tropical Medicine , London , United Kingdom ) , were obtained as described previously [39] . Epimastigote forms were maintained at 28°C in LIT media supplemented with 10% FBS and 100 μg/ml of antibiotic G418 to keep selective pressure in favor of the luciferase marker [68] . Epimastigotes were induced to differentiate to trypomastigotes through metacyclogenesis as previously described [69] . Metacyclic trypomastigotes were used to infect C2C12 myoblasts monolayers . After 5–7 days , trypomastigotes released in supernatant were collected by centrifugation for 15 min at 3300 rpm , re-suspended in DMEM and used to infect mice . Swiss Webster male mice weighting 18–20 g were infected intraperitoneally with 104 bloodstream trypomastigote form of T . cruzi Y parasites . For bioluminescence imaging , six week old BALB/c male and female mice were infected by intraperitoneal injection with 103 T . cruzi CL-luc trypomastigotes derived from cell culture supernatant . All drugs were solubilized in 10% solutol and administered orally , b . i . d , at previously optimized doses: 25 mg/kg for 9 and 12 , 50 mg/kg for benznidazole , and 20 mg/kg for posaconazole . The treatment of Swiss mice acutely infected with T . cruzi Y strain was started with parasitemia onset at 5 days post-infection ( dpi ) . The treatment of BALB/c mice infected with CL-luc parasites started at 14 dpi ( acute phase ) , when parasitemia reached a peak as detected by bioluminescence , or at 126 dpi ( chronic phase ) , when a chronic state of infection was established [46] . In all models , only parasite-positive mice ( 5 mice/group ) were used in the treatment course lasting for 28 days . To assess if sterile cure was achieved , immunossupression was performed in the chronic model of infection 4 weeks after the end of treatment , with two doses of cyclophosphamide ( 200 mg/kg ) by intraperitoneal ( i . p . ) injection at 3-day intervals . In Swiss mice acutely infected with T . cruzi Y strain , parasites in the blood of each animal were quantified by using the Pizzi-Brener method . The total number of parasites are counted in 50 fields under 400X magnification of freshly prepared blood samples ( 5 μl drops ) obtained from the tail veins of mice , collected 3 times a week , starting at 5 dpi and continued until the end of treatment [70] . Mortality was monitored daily and % survival was calculated using GraphPad prism software . BALB/c mice infected with parasites carrying a bioluminescent marker were imaged at 13 dpi , before treatment was initiated , and then once a week , both during the 28-day treatment period and 39 days post-treatment , as previously described [35] . Briefly , mice were injected i . p . with 150 mg/kg D-luciferin potassium salt in PBS ( Gold Biotechnology , St . Louis , MO ) , and 5 minutes later , anesthetized by isofluorane inhalation ( 3–5% ) and imaged using IVIS Lumina in vivo imaging system ( PerkinElmer , Waltham , MA ) with 180s exposure time . Data acquisition and analysis were performed with the LivingImage V4 . 1 software ( PerkinElmer , Waltham , MA ) . Uninfected controls were imaged in parallel to establish a negative threshold . To evaluate sites of parasite persistence in BALB/c mice infected with T . cruzi expressing luciferase , we performed ex vivo imaging of selected internal organs according to the protocol adapted from Lewis et al . , 2014 [39] . Briefly , the animals were injected i . p . with 150 mg/kg of D-luciferin , euthanized in a CO2 chamber and perfused with 10 ml of D-luciferin . Then , heart , liver , kidneys , lungs , spleen , mesenteric fat , skeletal muscle ( excised from left thigh ) and the whole gastro-intestinal ( GI ) tract were removed , placed in a petri dish with PBS containing 300 μg/ml of D-luciferin , and imaged using the IVIS Lumina system . Brain , heart , liver , kidney , GI tract and lungs from uninfected mice for toxicity analysis , and heart and GI tract from infected animals were removed and fixed as described above . Samples were processed for routine histologic examination in the Histology Core of Moore Cancer Center ( UCSD ) , embedded in paraffin , sectioned and stained with hematoxylin and eosin . The slides were scanned using Nanozoomer Slide Scanner ( Hamamatsu Photonics , NJ , USA ) and images were obtained through NDP viewer software ( Hamamatsu Photonics , NJ , USA ) . To quantify levels of inflammation , 5 random images of mouse hearts ( 10× magnification ) were obtained from each animal , 5 animals/group . At this magnification , 5 fields comprise the majority of the area of the heart section . Image processing was performed using Fiji software [44] , where cell nuclei was segmented through the Particle Analyzer plugin and the fraction of total area of the image occupied by all cell nuclei was then measured . Even though cardiomyocytes and cardiac fibroblasts cell nuclei are being counted together with inflammatory cells , uninfected heart sections were used as controls and provide a baseline number . Terminal blood collection was performed via cardiac cavity exsanguination in uninfected and T . cruzi-infected mice . Blood was collected in serum separator tubes ( Microtainer , BD Biosciences ) , allowed to clot for 0 . 5–2 . 0 h and then centrifuged for 90 s at 10000 g . Serum was removed and analyzed at the Central Animal Facilities of the Oswaldo Cruz Foundation ( Rio de Janeiro , Brazil , CECAL/Fiocruz platform ) using Vitros 250 ( Ortho Clinical-Johnson & Johnson ) , or at the UC Davis Comparative Pathology Laboratory ( Davis , CA , USA ) , where samples were analyzed using Roche Cobas Integra 400 Plus clinical chemistry analyzer . In both facilities , tests were performed for electrolytes and enzyme metabolites indicative of liver , kidney and cardiac functions , including alanine aminotransferase ( ALT ) , aspartate aminotransferase ( AST ) , total bilirubin ( BIL ) , albumin ( ALB ) , alkaline phosphatase ( ALP ) , blood urea nitrogen ( BUN ) , creatinine ( CRE ) , urea , calcium , phosphorus , glucose , total protein . Student’s t-test was used for evaluation of differences in experimental data between groups . Values were considered statistically significant when p≤ 0 . 05 . Statistics were analyzed by GraphPad Prism Software .
|
Chagas disease is a parasitic disease caused by the Trypanosoma cruzi . The infection may result in gastrointestinal manifestations and cardiomyopathy . Benznidazole , the current treatment , has limited efficacy and often leads to serious side effects . Aiming to develop new treatments , our group has identified new inhibitors that block the synthesis of parasitic lipids , resulting in parasite death . In this work , we evaluated the safety and efficacy of two of these compounds , 9 and 12 , in mouse models of T . cruzi infection . Both compounds were well-tolerated by animals throughout the 28-day administration . In acutely infected mice , the compounds drastically reduced bloodstream parasites and increased survival . When treatment was initiated during the chronic phase , parasitemia dropped to background levels and remained undetectable for 4 weeks after drug withdrawal; parasites were re-activated by chemically-induced immunosuppression . Thus , the experimental compounds tested in these studies had an acceptable safety profile , achieved a marked reduction in parasite load and prevented heart injury due to inflammation , even in the absence of sterile cure . We conclude that the development of non-toxic medications capable of slowing the progression of cardiomyopathy is a valuable treatment option for Chagas disease patients because it could enhance the quality of life .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"cardiovascular",
"anatomy",
"immunology",
"tropical",
"diseases",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"electromagnetic",
"radiation",
"parasitology",
"animal",
"models",
"parasitemia",
"model",
"organisms",
"protozoans",
"heart",
"signs",
"and",
"symptoms",
"experimental",
"organism",
"systems",
"neglected",
"tropical",
"diseases",
"quantitative",
"parasitology",
"research",
"and",
"analysis",
"methods",
"inflammation",
"mouse",
"models",
"protozoan",
"infections",
"luminescence",
"immune",
"response",
"physics",
"trypanosoma",
"cruzi",
"trypanosoma",
"eukaryota",
"chagas",
"disease",
"bioluminescence",
"anatomy",
"diagnostic",
"medicine",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"organisms"
] |
2017
|
4-aminopyridyl-based lead compounds targeting CYP51 prevent spontaneous parasite relapse in a chronic model and improve cardiac pathology in an acute model of Trypanosoma cruzi infection
|
An extensive proteostatic network comprised of molecular chaperones and protein clearance mechanisms functions collectively to preserve the integrity and resiliency of the proteome . The efficacy of this network deteriorates during aging , coinciding with many clinical manifestations , including protein aggregation diseases of the nervous system . A decline in proteostasis can be delayed through the activation of cytoprotective transcriptional responses , which are sensitive to environmental stress and internal metabolic and physiological cues . The homeodomain-interacting protein kinase ( hipk ) family members are conserved transcriptional co-factors that have been implicated in both genotoxic and metabolic stress responses from yeast to mammals . We demonstrate that constitutive expression of the sole Caenorhabditis elegans Hipk homolog , hpk-1 , is sufficient to delay aging , preserve proteostasis , and promote stress resistance , while loss of hpk-1 is deleterious to these phenotypes . We show that HPK-1 preserves proteostasis and extends longevity through distinct but complementary genetic pathways defined by the heat shock transcription factor ( HSF-1 ) , and the target of rapamycin complex 1 ( TORC1 ) . We demonstrate that HPK-1 antagonizes sumoylation of HSF-1 , a post-translational modification associated with reduced transcriptional activity in mammals . We show that inhibition of sumoylation by RNAi enhances HSF-1-dependent transcriptional induction of chaperones in response to heat shock . We find that hpk-1 is required for HSF-1 to induce molecular chaperones after thermal stress and enhances hormetic extension of longevity . We also show that HPK-1 is required in conjunction with HSF-1 for maintenance of proteostasis in the absence of thermal stress , protecting against the formation of polyglutamine ( Q35::YFP ) protein aggregates and associated locomotory toxicity . These functions of HPK-1/HSF-1 undergo rapid down-regulation once animals reach reproductive maturity . We show that HPK-1 fortifies proteostasis and extends longevity by an additional independent mechanism: induction of autophagy . HPK-1 is necessary for induction of autophagosome formation and autophagy gene expression in response to dietary restriction ( DR ) or inactivation of TORC1 . The autophagy-stimulating transcription factors pha-4/FoxA and mxl-2/Mlx , but not hlh-30/TFEB or the nuclear hormone receptor nhr-62 , are necessary for extended longevity resulting from HPK-1 overexpression . HPK-1 expression is itself induced by transcriptional mechanisms after nutritional stress , and post-transcriptional mechanisms in response to thermal stress . Collectively our results position HPK-1 at a central regulatory node upstream of the greater proteostatic network , acting at the transcriptional level by promoting protein folding via chaperone expression , and protein turnover via expression of autophagy genes . HPK-1 therefore provides a promising intervention point for pharmacological agents targeting the protein homeostasis system as a means of preserving robust longevity .
Aging is sensitive to both internal and environmental stimuli , and is tuned by multiple emergent genetic circuits . External signals include the nutritive value and quantity of the food supply; internal signals originate from discrete cellular sources , such as mitochondria or ribosomes , and discrete tissue sources , such as the reproductive system[1–7] . An overarching theme currently emerging in the aging field is one of homeostasis- homeostasis at the level of genome maintenance and gene expression[8–10] , and homeostasis at the level of proteome folding and stability[11 , 12] . The gradual loss of homeostasis , from precision of gene expression to protein folding and degradation is a common hallmark of aging organisms . Therefore , longevity is often extendable by manipulations that increase overall stress resistance , such as thermal shock or hypoxia , by a phenomenon known as hormesis . It is generally believed that hormesis extends longevity by bolstering organismal and cellular stress response pathways , which subsequently offsets aging-related decline in these pathways[13] . A major aim in aging research is to improve quality-of-life with advancing age ( often referred to as healthspan ) by reinforcement of those maintenance pathways that ensure the integrity of biological processes[6] . Indeed , a number of aging-related diseases , such as Alzheimer’s and Parkinson’s dementias , are believed to arise from the decline in the systems that maintain proteome stability and plasticity , by injury to or defects in the cellular processes that promote accurate protein folding and elimination of misfolded and damaged proteins[14] . Maintaining protein homeostasis ( proteostasis ) is the collective process that preserves a robust and functional proteome; an equation balanced by rates of protein synthesis , protein folding , and protein turnover . Protein synthesis places stress on the proteome by increasing the total concentration of cellular protein . Protein concentrations within the cell can approach saturation levels achieved within crystals[15] . Thus , a major challenge in maintaining proteostasis is an issue of solubility , which is managed by molecular chaperones . Chaperones play an integral role both in assisting in the correct maturation of nascent polypeptides , and in the elimination of proteins through chaperone-mediated degradative pathways . Cells eliminate misfolded , damaged or unneeded polypeptides by ubiquitin-mediated proteosomal degradation as well as by macroautophagy ( hereafter referred to as autophagy ) at the lysosome . Yet , chaperones have a limited buffering capacity to maintain proper folding under different forms of cellular stress . Thus potent stress response mechanisms act to resolve both acute and chronic stress to the proteome , through refolding , degradation , and sequestration . Protein homeostatic mechanisms are regulated at the transcriptional and post-transcriptional levels . For instance , the heat shock transcription factor HSF-1 activates transcription of the chaperone genetic network in response to a wide range of stresses , the most well-known being acute thermal stress[16] . A myriad of transcription factors in C . elegans have been shown to promote autophagy at the level of gene expression and autophagosome formation in response to various environmental stressors[17]; including FOXA ( PHA-4 ) [18] , TFEB ( HLH-30 ) [19 , 20] , Mondo/Mlx ( MML-1/MXL-2 ) [21] , the HNF4-related nuclear hormone receptor ( NHR-62 ) [22] , and several transcription factors necessary for ER and mitochondrial unfolded protein responses[23–25] . There is a growing body of evidence that demonstrates that the loss of autophagy and the decline of proteostasis are conserved hallmarks of normal aging[12 , 26 , 27] . Consistently , during aging there is a general deterioration in the ability of cells to activate these transcriptional responses to proteotoxic stress . However , there is evidence in C . elegans suggesting that normal cell non-autonomous signals impair proteostasis . For example , under normal conditions , thermosensory neurons inhibit the ability of distal tissues to resolve proteotoxic stress mediated by polyglutamine expression[28] , and the onset of reproduction triggers both a rapid decline in protein quality control in the soma[29] and chromatin silencing at stress response genes limits the somatic heat shock response[30] . In mammals , the autophagy system antagonizes the progression of multiple neurodegenerative disorders[31] . Thus , identifying signals that either positively or negatively impact the inducibility of proteostatic mechanisms , as well as how they are regulated and coordinated will be essential for the treatment of age-associated proteotoxic disease and to maximize healthy aging . In this study , we describe the C . elegans homolog of the HIPK homeodomain-interacting protein kinase , or HPK-1 , as an essential co-factor of multiple transcriptional responses that collectively preserve proteostasis . The Hipk gene family encodes a set of conserved kinases that act as transcriptional co-factors important for the regulation of cell growth , development , differentiation and apoptosis[32 , 33] . Hipks are activated by metabolic and genotoxic stressors from yeast to mammals[34–37] . In a previous study we found that hpk-1 is necessary for wild-type lifespan and the extended longevity of insulin signaling mutants[38] . Here we show that HPK-1 reinforces proteostasis in C . elegans in response to both thermal stress and dietary restriction by transcriptional activation of chaperone and autophagy pathways , respectively . We show that HPK-1 is necessary for chaperone induction by HSF-1 in response to heat stress and opposes sumoylation of HSF-1 , an inhibitory post-translational modification in mammals[39–41] . Consistently , we show that RNAi of the SUMO moiety encoded by smo-1 enhances HSF-1 dependent chaperone induction after heat stress . Thus , we propose that HPK-1 preserves HSF-1 activity in C . elegans by inhibiting sumoylation . As a separate output of HPK-1 , we showed that HPK-1 ( but not HSF-1 ) is necessary for induction of autophagosome formation and autophagy gene expression in response to dietary deprivation and reduced TORC1 signaling . HPK-1 overexpression both extended longevity and conferred protection against polyQ protein aggregate formation and toxicity . The ability of hpk-1 to increase lifespan and prevent protein aggregation relies on PHA-4 ( FoxA ) and MXL-2 ( Mlx ) , Thus , HPK-1 functions as a regulatory hub for multiple transcription factors with proteostasis-preserving activities .
We initially identified the hpk-1 gene in an RNAi screen aimed at identifying genes necessary for the extension of longevity in daf-2 ( e1370 ) mutant animals . The hpk-1 gene represented an attractive target for subsequent investigation because ( 1 ) members of the HIPK gene class are known to have broad physiological roles in transcription factor regulation in response to nutrient availability and other environmental cues , collectively suggesting a central role for HIPKs in stress response pathways across eukaryotes , and ( 2 ) initial characterization in our laboratory revealed that hpk-1 promotes the global maintenance of proteostasis by protecting animals from the formation of age-associated polyglutamine protein aggregates , and one of the phenotypic hallmarks of aging organisms is a gradual and progressive decline in proteostasis . In order to verify our previous observations that hpk-1 RNAi produces a progeric phenotype , we obtained an hpk-1 ( pk1393 ) deletion mutant strain that lacks most of the kinase domain and tested whether hpk-1 was essential for normal lifespan . Loss of hpk-1 shortened mean lifespan approximately 30% from 18–21 to 12–14 days ( Fig 1A , p<0 . 0001 , S1 Table ) , in line with a previous study[42] . To verify that the shortened lifespan displayed by hpk-1 ( pk1393 ) animals was the result of the hpk-1 deletion and not an independent mutation , we created transgenic animals expressing an HPK-1::GFP translational fusion under control of its own promoter . Inheritance of the Phpk-1::HPK-1::GFP transgene rescued the progeric phenotype of hpk-1 ( pk1393 ) , consistent with previous reports[42] . In one trial we found that hpk-1 overexpression with the endogenous promoter slightly increased lifespan ( S1 Table ) , contrary to a previous study[42] . Non-transgenic hpk-1 ( pk1393 ) siblings remained short-lived ( Fig 1A , S1 Table ) . One concern when studying mutants or gene inactivations that shorten lifespan is that such genes are essential for viability and that their disruption produces a non-specific and overall “sickly” phenotype . However , overexpression of such genes would not be predicted to extend longevity unless they exert broad regulatory control over essential processes or are themselves “rate-limiting” for lifespan ( like the heat shock transcription factor hsf-1 and the daf-16/FOXO transcription factor ) . To determine if hpk-1 is such a gene , we tested whether constitutive overexpression of hpk-1 could increase lifespan by placing it under the control of a strong ubiquitously-expressed promoter ( Psur-5 ) . Overexpression of hpk-1 ( Psur-5::HPK-1::CFP ) increased mean lifespan between 7–16% ( Fig 1B , p<0 . 0001 and S1 Table ) . We next tested whether hpk-1 plays a cytoprotective role in maintaining protein homeostasis . Age-associated decline in protein homeostasis can be measured in C . elegans through the visualization of in vivo polyglutamine aggregate formation in muscle cells harboring the Punc-54::Q35::YFP transgene , or later in life as aggregate formation overwhelms the chaperone network and locomotory paralysis ensues[43] . Loss of hpk-1 conferred either by RNAi or the pk1393 deletion resulted in the premature accumulation of fluorescently-labeled polyglutamine Q35::YFP aggregates ( Fig 2A–2C , S3 Table p <0 . 0001 for both comparisons ) . In the representative trial displayed in Fig 2 , on day 2 of adulthood , wild-type animals displayed 18 . 0+/-2 . 7 aggregates while the hpk-1 ( pk1393 ) null mutant and hpk-1 RNAi-treated Q35::YFP animals averaged 28+/-5 . 3 and 26 . 0+/-5 . 1 aggregates , respectively ( Fig 2D , S3 Table ) . Similarly , by day 8 of adulthood , 77–78% of hpk-1 ( RNAi ) and hpk-1 ( pk1393 ) animals were paralyzed while 50% of control Q35::YFP animals were paralyzed ( Fig 2E , S3 Table ) . We next tested whether overexpression of hpk-1 was capable of conferring a cytoprotective phenotype in protein aggregation assays . Overexpression of hpk-1 reduced both Q35::YFP foci accumulation ( Fig 2D , S3 Table ) and protected aging animals from Q35::YFP-associated paralysis ( Fig 2E , S3 Table ) . Further experiments with additional transgenic lines were largely consistent with these results ( S3 Table ) . Thus , hpk-1 is vital for preserving protein solubility and protecting against aggregate toxicity in adult animals as they age . In order to visualize the spatiotemporal pattern of hpk-1 expression , we analyzed the expression of a Phpk-1::hpk-1::GFP transgene . hpk-1 is expressed broadly during embryogenesis , but becomes more restricted in expression during larval development ( S1A Fig ) . L3-stage larvae display robust expression of the GFP fusion in many head and motor neurons , and lower levels of expression in the intestine and the seam cells of the hypodermis . By late L4 stage , GFP expression is largely restricted to neurons , and is maintained in nerve cells of the head and nerve cord during adulthood , congruent with a previous study[44] . Localization of HPK-1::GFP protein is most concentrated in the nucleus often within distinct sub-nuclear sites ( S1B Fig ) , consistent findings in mammals[45] and the predicted function of HPK-1 as a transcriptional regulator . Identifying spatiotemporal requirements in longevity control is necessary for understanding how age-associated decline in individual tissues contributes to the larger gestalt of overall animal viability . Thus , we sought to discover where anatomically and when chronologically HPK-1 was essential for a normal lifespan . We first used stage-specific RNAi feeding to test whether the requirement of hpk-1 for normal longevity was restricted to a particular life stage . To assess whether larval-specific activities of hpk-1 are critical for normal adult lifespan , animals were raised on hpk-1 RNAi during development , and were transferred to dcr-1 RNAi at the late L4 stage in order to terminate continued silencing of hpk-1 by RNAi . Animals raised on RNAi bacteria targeting hpk-1 during larval development exhibited a shortened lifespan similar to lifelong inactivation of hpk-1 ( S2A and S2B Fig ) , while adult-restricted inactivation of hpk-1 displayed a weaker progeric phenotype ( S1C Fig ) . Interestingly , these temporal requirements are essentially identical to those previously described for hsf-1[46] . Given the broad developmental expression of hpk-1 , we next sought to define the tissues where hpk-1 acts to promote normal longevity . Tissue-restricted RNAi of hpk-1 in the intestine or the hypodermis both caused a significant progeric phenotype ( S2D and S2E Fig ) , consistent with intestinal and hypodermal expression being limited to larval developmental stages . In contrast , inactivation of hpk-1 in muscle cells had little effect on lifespan ( S2F Fig ) , consistent with the absence of HPK-1 expression in these cells as determined with our fluorescent reporter ( S1A Fig , and S1 File ) . We next tested whether neuronal hpk-1 function was necessary for normal lifespan using an enhanced neuronal RNAi ( RNAi ( en ) ) strain , as RNAi efficiency in neurons is low in wild-type animals . Neuronal inactivation of hpk-1 showed reduced lifespan to an extent comparable to inactivation of hpk-1 by systemic RNAi ( S2G Fig ) and hpk-1 null mutant animals ( Fig 1A ) . As a positive control to confirm RNAi ( en ) activity , daf-2 ( RNAi ) significantly increased lifespan in the RNAi ( en ) strain ( S2H Fig ) while a control strain lacking the dsRNA channel sid-1 and enhanced neuronal RNAi did not ( S2I Fig ) , consistent with previous reports[47 , 48] . Thus hpk-1 is required across all of the tissues in which we have observed its expression during the larval stages of development to ensure wild-type lifespan . Because HPK-1::GFP expression is restricted to neurons in adult animals ( S1A Fig ) , we interpret the longevity-extending activity of HPK-1 observed in the intestine and hypodermal seam cells ( S2D and S2E Fig ) to arise largely from larval-stage functions of HPK-1 in those tissues , although hpk-1 does have modest longevity-extending effects in adulthood as well ( S2C Fig ) . Based on a pilot screen to identify putative genetic interactions between known longevity genes and hsf-1 loss of function , we investigated the extent to which hpk-1 exerts its effects on overall longevity and proteostasis via hsf-1 . We measured the extent to which hpk-1 exerts its effects on overall longevity and proteostasis via the HSF-1 pathway by examining whether their loss-of-function phenotypes were additive . We observed that hsf-1 inactivation in hpk-1 ( pk1393 ) null mutant animals did not result in a meaningful additional decrease in lifespan ( Fig 3A ) . Further , inactivation of hsf-1 by RNAi was sufficient to suppress the increased lifespan of the long-lived hpk-1 overexpression line ( Fig 3B ) . Conversely , hpk-1 was necessary for the extended longevity observed in hsf-1-overexpressing animals ( Fig 3C ) . The reciprocal requirement we observed between hpk-1 and hsf-1 for animal longevity is evidence of a genetic interaction between these factors . Next we tested whether hpk-1 and hsf-1 function together or in separable pathways to protect animals from Q35::YFP foci formation by comparing animals lacking both hpk-1 and hsf-1 to animals deficient in only one of these genes . As expected , loss of hpk-1 or hsf-1 alone resulted in the premature accumulation of protein aggregates and onset of paralysis ( Fig 3D and 3E ) . Inactivation of hsf-1 by RNAi in the absence of hpk-1 failed to produce a statistically detectable increase in the accumulation of foci or onset of paralysis over time ( Fig 3D and 3E ) . Additional experiments corroborated these results: hpk-1 RNAi-treatment alone had as great or a greater negative impact on proteostasis when compared to hsf-1 RNAi treatment ( S3 Table ) . These results are consistent with the notion that hpk-1 and hsf-1 function to maintain protein homeostasis and delay the progression of aging through a shared mechanism . That homeodomain interacting protein kinases function as direct regulators of transcription factor activity suggested that the interaction between HPK-1 and HSF-1 may be direct . To begin to explore this possibility , we examined whether HPK-1 co-localizes with HSF-1 at the subcellular level by comparing localization of a Phsf-1::hsf-1::GFP transgene to a translational fusion between hpk-1 and the fluorescent tdtomato protein ( Phpk-1::hpk-1::tdtomato ) using confocal microscopy ( Fig 4 ) . Though not perfectly overlapping , HPK-1 and HSF-1 localization were often coincident with each other . We next sought to determine if hpk-1 shares functionality with HSF-1 with respect to the heat shock stress response by examining its role in regulating the transcriptional induction of chaperone gene expression after exposure to heat . We initially tested whether hpk-1 is important for preserving thermotolerance in response to heat shock . We observed that survival at 35°C was significantly reduced in hpk-1 ( pk1393 ) mutant animals , and that thermotolerance could be restored to wild type levels by the rescuing hpk-1 transgene ( Phpk-1::hpk-1::GFP ) ( Fig 5A ) . Furthermore , we observed that the long-lived hpk-1 overexpression line ( Psur-5::hpk-1::CFP ) increased thermotolerance survival from ~50% in wild type animals to ~75% , and that this increase in thermotolerance was fully dependent on hsf-1 ( Fig 5B ) . HSF-1 transcriptional activity is regulated in mammals through a complex array of post-translational modifications including phosphorylation , acetylation , and sumoylation ( reviewed in[41 , 49] ) . We sought to determine whether loss of hpk-1 altered either expression levels and/or post-translational modifications to the HSF-1 protein . Unmodified HSF-1 displayed the predicted mobility of a ~75 kD protein[50] . We additionally observed in wild-type animals two higher molecular weight isoforms between ~90 and 95 kD ( Figs 6A , S3 ) . Loss of hpk-1 resulted in an increase in the ratio of higher molecular weight isoforms of HSF-1 to the unmodified 75 kD species ( Figs 6B , S3B ) and an increase in overall levels of HSF-1 protein relative to the β-actin control ( Figs 6A and 6C , S3C ) . Sumoylation typically results in an electrophoretic mobility upshift of ~15 kD[51] . Mammalian HSF-1 has been shown to be sumoylated[40] . Consequently , we hypothesized that the two higher MW isoforms of HSF-1 might represent a sumoylated product ( S isoform ) ( Figs 6A and S3 , lower ~90 kD MW band ) , and SUMO plus phosphorylation ( S+P isoform ) ( Figs 6A and S3 , upper ~95 kD MW band ) . Consistent with this hypothesis , lambda protein phosphatase treatment of hpk-1 null extracts resulted in the loss of the highest MW isoform of HSF-1 but not the ~90 kD isoform ( Figs 6A , S3A ) , suggesting that the 90 kD isoform is not a result of phosphorylation events . This result confirms the hypothesis that HSF-1 phosphorylation is a modification of the 90 kD isoform . However , the 95 kD band is still observed in an hpk-1 mutant , indicating that HPK-1 is not the kinase responsible for this phosphorylation event . We determined that the higher molecular weight isoforms likely represent sumoylated forms of HSF-1 by two approaches . First , hpk-1 null mutant animals were grown on smo-1 RNAi , which reduces expression of the C . elegans SUMO gene that produces the SUMO moiety . smo-1 RNAi of hpk-1 mutant animals resulted in a decrease in the ratio of the pair of high MW bands to unmodified HSF-1 at 75 kD . Second , HSF-1 was immunoprecipitated and treated with SUMO protease , which resulted in a relative increase in the 75 kD ( unmodified ) HSF-1 isoform and a relative decrease in the 90–95 kD ( sumoylated ) bands of HSF-1 ( Fig 6D ) . While the loss of the sumoylated species was incomplete , each result is consistent with our prediction that the higher molecular weight isoforms of HSF-1 are the result of sumoylation . Thus , HPK-1 acts directly or indirectly to oppose HSF-1 sumoylation , either by blocking sumoylation or promoting de-sumoylation . We then tested our hypothesis that sumoylation is an inhibitory modification on HSF-1 by analyzing induction of the HSF-1 transcriptional program in response to heat shock in animals grown on smo-1 RNAi ( Fig 7 ) . In response to thermal stress , smo-1 RNAi-treated animals harboring the Phsp-16 . 2::GFP transgene displayed enhanced induction of GFP relative to empty vector control animals ( Fig 7A–7D ) . It is worth noting that in the absence of heat shock , the Phsp-16 . 2::GFP is not induced , indicating that loss of sumoylation is not sufficient of itself to activate the HSF-1 transcriptional response . In a complementary experiment , we analyzed protein induction of GFP expressed from the hsp-16 . 2 promoter as well as the endogenous HSP-16 . 2 protein in response to heat shock ( Fig 7E ) . Prior to heat shock , animals were raised on RNAi targeting GFP , hsf-1 , smo-1 or the empty vector control . In response to heat shock , both GFP and HSP-16 . 2 were induced . GFP but not HSP-16 . 2 induction was blocked by GFP ( RNAi ) . Both GFP and HSP-16 . 2 induction were significantly reduced by hsf-1 ( RNAi ) , confirming that induction of these proteins was dependent on the presence of HSF-1 . Finally , smo-1 ( RNAi ) increased the level of GFP and HSP-16 . 2 protein when compared to the EV control , supporting our hypothesis that HSF-1 sumoylation is a modification that is likely to inhibit the HSF-1 transcriptional program in response to heat shock . In general , the sustained hardship of chronic exposure to an environmental stressor , or application of a stressor to a level above a physiologically tolerable threshold typically compromises organism viability ( and consequently limits organismal lifespan ) . Hormesis is the initially counterintuitive but now well-validated and generalizable phenomenon produced by acute exposure to an environmental stressor . Brief intervals of exposure to stressors ( e . g . heat , ROS , etc . ) can confer a “hormetic effect” characterized by a significant extension of longevity accompanied by sustained resistance to physiologic stress . For instance , a transient 30°C pulse during development and/or early adulthood increases lifespan in C . elegans , while sustained growth at 25°C hastens the onset of animal morbidity[52] . It is generally postulated that the mechanism of extended longevity following heat shock is derived from upregulation of chaperone systems and turnover of misfolded proteins , which consequently promotes long term proteostatic robustness/resiliency once heat stress is removed[53] . HSF-1 is necessary for the hormetic effect of transient heat exposure on lifespan[54] . We tested whether hpk-1 was required for heat-induced hormetic extension of longevity . While defining experimental conditions for heat-induced hormesis , we found that growth at 25°C for the first 3 days of adulthood is a form of mild heat stress that to our knowledge has not been previously documented in C . elegans ( S4A–S4F Fig ) . We subsequently observed that transient growth at 25°C during this interval in early adulthood ( when heat-resistance mechanisms at 20°C normally undergo a dramatic downregulation[29 , 30 , 55] ) was sufficient to increase lifespan only in the presence of hsf-1 and hpk-1 ( S4G and S4H Fig , p<0 . 001 ) . Thus hpk-1 , like hsf-1 , is necessary for hormetic extension of longevity in response to heat stress , a result consistent with a positive regulatory function of HPK-1 over HSF-1 . hipk family members canonically function as positive regulators of transcriptional co-activators . If the interaction between HPK-1 and HSF-1 is direct , hpk-1 could be promoting HSF-1 activity at various regulatory points in the chain of events beginning with newly translated HPK-1 and ending with induction of gene transcription by HSF-1 . For example , post-translational modification of HSF-1 by HPK-1 could affect its stability , subcellular localization , DNA binding or transactivation activity at the level of recruitment of RNApol II , additional transcription factors , or chromatin modifying factors . We undertook several experiments to identify mechanisms of regulatory control . Immediately following thermal stress , acute cytoplasmic misfolding challenges the chaperone network to free HSF-1 from Hsp90 cytoplasmic sequestration , allowing HSF-1 to undergo translocation to the nucleus , where it becomes concentrated in stress granule-like subnuclear speckles that colocalize with markers of active transcription[56] . Re-allocation and compartmentalization of HSF-1 to nuclear speckles occurs within minutes of initiating thermal stress ( [57] and this study ) . When we tested whether hpk-1 was necessary for HSF-1 re-localization or compartmentalization in response to heat shock , we found that these early events of HSF-1 activation are hpk-1 independent , as hpk-1 ( RNAi ) had no effect on either readout of HSF-1 activation ( S5 Fig ) . Therefore hpk-1 is likely to play a role in subsequent events during activation and/or establishment of the HSF-1 transcriptional response . We tested whether hpk-1 was necessary for the transactivation of HSF-1 by determining whether the induction of chaperone target genes in response to thermal stress was compromised in the absence of hpk-1 . We first analyzed whether hpk-1 was required for induction of the Phsp-16 . 2::GFP reporter . hsp-16 encodes a small chaperone that is induced by heat shock in a manner requiring hsf-1[58] . We found that hpk-1 was also necessary for hsp-16-2::GFP induction in response to transient heat shock ( Fig 8A–8D ) as previously shown[42] . In addition , we found that hsf-1-dependent transcriptional induction of the endogenous chaperones encoded by hsp-16 . 2 and hsp-70 also required the presence of hpk-1 for heat shock inducibility ( Fig 8E and 8F ) , which unexpectedly differs from a previous report[42] . HPK-1 regulation of chaperone gene expression is dependent on heat stress , as loss of hpk-1 did not significantly alter basal expression levels of hsp-16 . 2 and hsp-70 ( Fig 8G ) . In contrast , hsf-1 inactivation has been reported to reduce endogenous levels of hsp-16 . 2 and hsp-70 mRNA by ~40%[50] , suggesting some basal HSF-1 activity remains in the absence of hpk-1 . We also considered the possibility that HSF-1 functions upstream rather than downstream of HPK-1 , or as part of a feedback loop with HSF-1 . We examined HPK-1 expression in response to multiple conditions of stress including heat shock , and tested whether HPK-1 expression is regulated by HSF-1 . Under basal conditions , transgenic animals that express a GFP translational fusion that includes the hpk-1 open reading frame ( Phpk-1::hpk-1::GFP ) displayed a broad pattern of developmental expression in the intestine , hypodermal seam cells , and neurons . The expression pattern of this transgene became restricted to neurons as animals transitioned to adulthood ( Figs 9A , S1A ) . We next tested whether the pattern of hpk-1 expression is regulated by thermal stress . We observed robust induction of hpk-1 expression in transgenic animals expressing the translational reporter after heat shock ( Fig 9B ) . Induction of HPK-1 was greatest in hypodermal seam cells , neurons , and to a much lesser extent within intestinal cells ( Figs 9B , S5 ) , and this induction did not require hsf-1 ( Fig 9C ) . We asked whether HPK-1 induction was transcriptional or post-transcriptional , as early events in many stress response pathways including the heat shock response do not require active transcription . Consistently , we could not discern an increase in fluorescence in transgenic animals expressing a transcriptional fusion of the hpk-1 promoter to GFP ( Phpk-1::GFP ) ( S7A and S7B Fig ) . However , we did notice that hypodermal seam cells appeared much larger and swollen compared to unstressed controls ( S7A and S7B Fig , S1 File ) . As extrachromosomal transgenic reporter lines lack both normal gene copy number and the context of endogenous chromatin , we further examined hpk-1 mRNA levels in wild-type animals by qRT-PCR but found no significant difference in mRNA expression as a function of heat shock ( S7C Fig ) . Importantly , endogenous hsp-16 . 2 mRNA was induced in the heat stressed sample ( S7D Fig ) . Because hsf-1 is a global activator of chaperones and other heat shock response genes , we asked whether the thermal inducibility of HPK-1 translation requires hsf-1 . Consistent with a mode of post-transcriptional regulation of hpk-1 in response to heat shock , hsf-1 ( RNAi ) had no effect on the induction of the translational Phpk-1::hpk-1::GFP reporter after heat shock ( Fig 9C ) and induction was not blocked by pre-treatment with the RNA polymerase inhibitor α-amanitin ( Fig 9D ) . In contrast , α-amanitin pre-treatment completely blocked the induction of the known transcriptional Phsp-16 . 2::GFP reporter in response to heat shock ( S8 Fig ) . Induction of HPK-1 protein is specific to thermal stress , as oxidative damage ( by tert-butyl hydroperoxide ) ( Fig 9E ) and DNA damage ( by UV ) ( Fig 9F ) failed to alter Phpk-1::HPK-1::GFP levels or its pattern of expression . Thus , HPK-1 is specifically induced by thermal stress and this induction is post-transcriptional . In addition to heat shock , DNA damage , and oxidative stress , another type of stress that contributes to longevity is metabolic or nutritional stress . Many longevity transcription factors linked to environmental stress responses are also responsive to changes in nutritional status . For example , HIPK2 induction under conditions of glucose deprivation has been described in mammalian cell culture[59] . This suggested that HPK-1 expression might be inhibited by the insulin-like receptor ( daf-2 ) and/or the TOR signaling complex , both of which control key nutritional responses in C . elegans and extend longevity when disrupted . We tested whether RNAi inactivation of daf-2 , daf-15 ( corresponding to raptor , a TORC1 subunit ) , or rict-1 ( corresponding to rictor , a TORC2 subunit ) altered HPK-1 expression . We observed no change in expression of the Phpk-1::HPK-1::GFP reporter in response to daf-2 or rict-1 inactivation , but we did observe a significant increase in HPK-1::GFP protein expression in head neurons in daf-15 ( RNAi ) animals ( Figs 10A–10D , S8A ) . HPK-1 induction was distinct mechanistically from heat shock induction of HPK-1 in two ways: 1 ) hpk-1 mRNA levels were increased by inactivation of daf-15 ( Fig 10E ) but not heat shock ( S7C Fig ) , and 2 ) HPK-1 induction by daf-15 RNAi treatment was restricted to neurons whereas heat shock induced a broader expression pattern ( S9B–S9D Fig ) . daf-15 ( RNAi ) and let-363 ( RNAi ) ( TOR kinase ) have been shown to increase lifespan in wild-type animals[60 , 61] . We subsequently asked whether hpk-1 was necessary for the enhanced longevity of either daf-15 ( RNAi ) or let-363 ( RNAi ) -treated animals . We observed suppression of the extended longevity phenotype of daf-15 ( RNAi ) or let-363 ( RNAi ) -treated animals to the lifespan observed in hpk-1 mutant animals alone ( Fig 10F and 10G respectively ) , suggesting that hpk-1 may be an inhibitory target of TORC1 that is critical for changes in longevity mediated by altered TOR signaling . Under nutrient rich conditions , TOR promotes cellular growth by activating protein translation ( e . g . transcription of translation components ) while inhibiting protein turnover ( e . g . transcription of chaperones[62 , 63] and autophagy genes[64] , and the initiation of autophagy[65] ) . TOR inhibition , or genetic activation of any of these targets of TOR inhibition , results in extension of longevity[18 , 60] . We tested whether regulation of any of these cellular processes was dependent on hpk-1 . Autophagy is induced in response to fasting across many species , and can be visualized in C . elegans using the LGG-1::GFP reporter , in which GFP is C-terminally fused to the autophagosome component LGG-1 ( i . e . LC3/Atg8 in mammals and yeast , respectively ) . Stimulation of autophagy is observed in epidermal seam cells during fasting as the formation of discrete LGG-1::GFP puncta[18] . We tested whether hpk-1 was necessary for autophagosome formation following six hours of bacterial deprivation ( BD ) relative to replete , or ad libitum ( AL ) conditions . Consistent with published reports , LGG-1::GFP foci were rarely observed under ad libitum conditions ( Fig 11A–11C ) , but were readily observable in seam cells upon bacterial deprivation ( Fig 11D ) . We observed a near total loss of LGG-1::GFP foci formation in response to BD in hpk-1 ( RNAi ) -treated animals ( Fig 11E ) . Interestingly , LGG-1::GFP foci formation following bacterial deprivation did not require hsf-1 ( Fig 11F and 11G ) . This result reveals that autophagosome formation in hypodermal seam cells constitutes a biological function for HPK-1 that is separable from its role in regulating HSF-1 activity and HSF-1-dependent proteostatic outputs . Because HIPKs are transcription factor regulators , we tested whether any of the genes negatively regulated by TOR require hpk-1 for their expression . We observed that hpk-1 is necessary for the induction of two autophagy genes ( atg-18 and bec-1 ) previously reported to be induced in response to daf-15 inactivation and that are essential for autophagosome production ( Fig 12A ) [18 , 20 , 64 , 66] . In contrast , loss of hpk-1 had no effect on daf-15 ( RNAi ) -induced down regulation of ifg-1 and iftb-1 ( eIF-4G and eIF2beta , respectively ) , two translation initiation genes whose transcription is induced by active TORC1 ( Fig 12B ) . Similarly , loss of hpk-1 had no effect on the modest induction of molecular chaperones that occurs with TORC1 inactivation ( Fig 12C ) , an induction known to be hsf-1 dependent[63] . Therefore , a previously described interaction between TORC1 and HSF-1 leading to chaperone transcriptional induction is hpk-1-independent . Conversely , HPK-1 functions specifically in the autophagy axis of TORC1 signaling while hsf-1 does not . Therefore , HPK-1 and HSF-1 must each have cellular functions that are distinct from each other in addition to their shared control of heat shock responses . Our result that HPK-1 promotes autophagy through a mechanism independent of HSF-1 suggests that HPK-1 regulates at least one additional transcription factor that is necessary for autophagy induction in response to DR or TORC1 inactivation . PHA-4/FoxA , HLH-30/TFEB , NHR-62/HNF4-related nuclear hormone receptor are transcription factors known stimulate autophagosome assembly[18 , 22 , 64 , 66–68] . MXL-2/Mlx ( part of the MML-1/ ( MondoA/ChREBP ) complex ) has been implicated in autophagy regulation because it is the binding partner of MML-1 , which promotes autophagy gene expression upon inactivation of TOR[21] . As we are primarily concerned with those functions connecting TOR and HPK-1 activity to lifespan , we examined whether the increased lifespan arising from HPK-1 overexpression ( Fig 1B ) was suppressed by inactivation of any of the above transcription factors . Of these , we found that inactivation of pha-4 or mxl-2 is epistatic to hpk-1 overexpression , suppressing the increased lifespan conferred by hpk-1 overexpression ( Fig 13A and 13B ) . In contrast , animals overexpressing hpk-1 were still long-lived after inactivation of hlh-30 or nhr-62 ( Fig 13C and 13D ) . Thus , hpk-1 overexpression extends longevity in a manner dependent on pha-4 and mxl-2 , but not nhr-62 or hlh-30 . Autophagy has been shown to ameliorate aggregate formation in response to polyglutamine tracts in C . elegans , as well as other systems[69–71] . We therefore tested whether these autophagy-inducing transcription factors mitigate polyQ aggregate formation and/or toxicity in muscle cells , as has been reported in two other studies[55 , 72] , and whether their ability to do so requires hpk-1 . Inactivating pha-4 or mxl-2 in otherwise wild-type Q35::YFP animals resulted in both accelerated protein aggregate formation and early onset of paralysis ( Fig 14A–14D ) , consistent with previous findings . Curiously , inactivation of hlh-30 also accelerated Q35 aggregate formation without enhancing its associated locomotory toxicity . Next , we wished to determine if these transcription factors act in parallel or as part of an hpk-1 regulatory network that promotes protective proteostatic responses to oppose the aggregation-associated phenotypes of polyglutamine repeats . Loss of either pha-4 or mxl-2 did not worsen the accelerated polyglutamine phenotypes of an hpk-1 ( pk1393 ) mutant , suggesting that pha-4 and mxl-2 protect against aggregate formation in conjunction with hpk-1 , possibly as direct transcription factor targets . In contrast , the hlh-30 results were equivocal: while hlh-30 RNAi did not accelerate the formation of Q35::YFP aggregates in the absence of hpk-1 , hlh-30 RNAi surprisingly resulted in partial restoration of the premature aggregate-associated motility defects of hpk-1 ( pk1393 ) . The apparent contradiction that loss of hlh-30 ameliorates the proteotoxicity of an hpk-1 ( pk1393 ) mutant will require additional experimentation to determine the regulatory relationships between hlh-30 , hpk-1 , mxl-2 , and pha-4 . Nevertheless , our results- which tentatively place hpk-1 in a regulatory network with mxl-2/pha-4 but not hlh-30- are consistent with previous reports that have found hlh-30 and pha-4 have distinct pro-longevity functions[19] and pha-4 and mxl-2 have similar pro-longevity functions[72] . It is tempting to speculate that loss of hlh-30 may be compensated by upregulation of MXL-2 or PHA-4 , which partially rescues the proteotoxicity that results from loss of hpk-1 . Bacterial deprivation is known to activate autophagy and the nuclear translocation of DAF-16/FoxO and HLH-30/TFEB[19 , 73] . Since hpk-1 is required for autophagosome formation after bacterial deprivation , we tested whether hpk-1 was also required for DAF-16 or HLH-30 activation . Loss of hpk-1 did not impair cytoplasm-to-nuclear translocation of HLH-30::GFP or DAF-16::GFP after bacterial deprivation ( S10 and S11 Figs ) . This is consistent with the notion that hpk-1 and daf-16/hlh-30 have separable functions in nutrient sensing . Collectively , our results support a model in which PHA-4 and MXL-2 represent a nutritionally-responsive arm of HPK-1 regulation in addition to its role in HSF-1 activation in response to thermal stress and in preserving the integrity of the proteome under unstressed conditions .
In this study we describe HPK-1 as a transcriptional regulator of proteostasis and longevity in C . elegans . HPK-1 is the lone representative of the homeodomain-interacting kinase gene family in C . elegans , a relatively understudied family of kinases . hpk-1 is most closely related to yak1 in Saccharomyces cerevisiae and to the HIPKs and DYRK kinases in mammals . In mammals a total of four HIPK orthologues respond to a number of external cues including the DNA damage response , hypoxia response , reactive oxygen species ( ROS ) , glucose availability , and viral infection[32 , 45 , 74 , 75] . In general HIPK family members regulate the activity of transcription factors , chromatin modifiers , signaling molecules and scaffolding proteins in response to cellular stress . For example , genotoxic damage induces mammalian Hipk2 , and HIPK2 potentiates p53 pro-apoptotic activity through direct phosphorylation[34] . We originally identified hpk-1 in an RNAi screen for genes acting in the daf-2/insulin signaling pathway ( i . e . genes necessary for the increased lifespan of daf-2 mutants ) . A subset of the gene inactivations that shorten daf-2 ( e1370 ) lifespan also confer no additional lifespan shortening effect in a daf-2;daf-16 genetic background[38] , implying function specifically within the ILS pathway . hpk-1 ( RNAi ) met these phenotypic criteria , but we ultimately concluded that hpk-1 was unlikely to be component of the canonical DAF-2 ( insulin/IGF1R ) -AGE-1 ( PI3K ) -AKT-DAF-16 ( FoxO ) signaling pathway , because loss of hpk-1 only modestly suppressed daf-2 lifespan , and to a degree that was not proportionally greater than the reduction of lifespan observed in response to hpk-1 inactivation in wild-type animals[38] . Additionally , induction of the DAF-16 target gene sod-3 under conditions of decreased ILS does not require hpk-1[38] . Our conclusion has since been supported by a separate study reporting that only eight of 259 DAF-16/ILS regulated genes showed decreased expression in animals lacking hpk-1[42] . Lastly , in this manuscript we show that decreased ILS does not alter hpk-1 expression ( Fig 10B ) . Collectively these findings are more consistent with a model where HPK-1 functions not within but parallel to the canonical ILS pathway . We show that overexpression of hpk-1 extends natural longevity , which suggests that HPK-1 exerts a regulatory function on longevity pathways , rather than simply being required for some essential physiological function . We show that a translational HPK-1::GFP reporter is expressed broadly during C . elegans embryogenesis and larval development ( in intestine , hypodermal seam cells and neurons ) , but its expression pattern becomes restricted to neurons in adults . These data are consistent with reported patterns of HIPK expression in mammals and C . elegans [76 , 77] . Because stress response pathways are intimately tied to longevity , we tested whether hpk-1 expression was induced by oxidative stress , DNA damage or heat shock . We found that heat shock induces HPK-1 in the same tissues where it is normally expressed during development: hypodermis and neurons , and to a lesser extent , the intestine . We showed that heat shock induction was at the protein level , as it occurred in the presence of α-amanitin . A role for HPK-1 in heat shock responses is likely to be evolutionarily conserved , as yak1 is also induced by heat stress and is required for normal thermal stress survival in S . cerevisiae [78 , 79] . In contrast , we observed no induction of HPK-1 in response to oxidative or DNA damage . We investigated whether hpk-1 functions as part of the well-known HSF-1 heat shock response pathway . We found that hpk-1 was required for HSF-1 mediated thermotolerance , the hormetic extension of longevity , and transcriptional induction of two key HSF-1 target genes , the chaperones encoded by hsp-16 . 2 and hsp-70 . hsf-1 and hpk-1 are mutually required for the increased lifespan arising from overexpression of the other . We observed that an HPK-1::GFP fusion protein was localized to subcellular structures coincident with HSF-1 localization , an observation consistent with a model in which HPK-1 and HSF-1 reside within a regulatory complex . We believe that HPK-1 is likely to be a positive regulator of HSF-1 activity because of its homology to kinases that activate transcription factors by phosphorylation . The homology of HPK-1 to the homeodomain-interacting kinase family , together with our result that HPK-1 induction was not blocked by global inhibition of transcription meant that we were unsurprised to discover that the heat shock inducibility of HPK-1 was also HSF-1-independent , A separate study has recently reported that C . elegans HPK-1 is induced by thermal stress . However the authors concluded that hpk-1 was not part of the HSF-1 heat shock response because they saw no dependence on hpk-1 for induction of endogenous molecular chaperones after heat shock[42] . We believe we can reconcile this discrepancy as a consequence of the timing at which chaperone induction was measured . We heat shocked animals and examined endogenous chaperone induction during late larval development , while Berber et al . tested for chaperone induction after the onset of reproduction . However , the onset of reproductive maturity in C . elegans is characterized by an extreme downshift in the ability of animals to respond to heat shock . Within 4 to 8 hours of the onset of reproduction , repressive chromatin marks are laid down at stress loci; these repressive chromatin modifications severely curtail the ability of animals to respond to heat shock[30] . This seems a likely explanation for why Berber et al . observed only low-level induction of hsp-70 and hsp-16 in response to heat shock , and why they were unable to detect a dependency on hpk-1 . In contrast , we conducted our experiments prior to the timing of chromatin repression at heat shock loci . Resultantly , we observed a far larger transcriptional induction for hsp-70 and hsp-16 . 2 compared to Berber et al . ( ~100 and 4500-fold , respectively , versus ~6–8 fold ) and nearly all of this induction required the presence of hpk-1 . This suggests that hpk-1 is essential for the activation of the heat shock response prior to chromatin silencing at stress loci , but after the heat shock response is compromised through chromatin remodeling at stress loci , hpk-1 is no longer essential for the limited transcriptional activation of heat shock genes . As Berber et al . did not report whether the residual expression of chaperones they observed requires hsf-1 itself , it is impossible to say whether this modest transcriptional induction is HSF-1 mediated or occurs through some other mechanism . HSF-1 promotes global proteostasis even in the absence of heat stress by regulating the transcription of chaperones , which maintain stability and solubility of the proteome both by assisting in the folding of newly translated polypeptides and by directing chronically misfolded proteins to the proteasome for degradation . The protective functions of the chaperone system are particularly relevant to diseases of the nervous system caused by inappropriate protein aggregation and resulting neural toxicity . Expression of aggregation-prone polyQ transgenic constructs in C . elegans provides both a method for detecting the proteostatic stress level of tissues and a means of identifying protective and risk factors for aggregation diseases . Inactivation of hsf-1 , for instance , causes a premature accumulation of polyglutamine-YFP puncta in muscle cells[80] . When we tested whether HPK-1 displays a similar protective effect on polyQ aggregate formation , we found that HPK-1 is as , if not slightly more important than , HSF-1 in its ability to delay insoluble aggregate formation and associated paralysis in Q35:YFP animals ( see S3 Table ) . Moreover , hsf-1 RNAi did not increase the number or toxicity of polyQ aggregates when combined with the hpk-1 ( pk1393 ) mutation , showing that HSF-1 confers its protective effects entirely under the regulatory umbrella of HPK-1 . Strikingly , HPK-1 overexpression exerted a potent protective effect against polyQ proteotoxicity by dramatically reducing the rate of foci formation and paralysis . There has been significant effort invested in defining the mechanism ( s ) of activation of the heat shock transcription factor HSF-1 , an effort complicated by the large number of post-translational modifications on HSF-1 . Biochemical analysis of HSF-1 regulation is an extremely active area of research from yeast to mammals . Most studies of HSF-1 activation have been performed in tissue culture and ex vivo models , but there is very little information about the regulation of HSF-1 by post-translational modification in living animals . In this study , we describe a pair of HSF-1 modifications in C . elegans that correlate with reduced transcriptional activity of HSF-1 during aging , and in a manner dependent on hpk-1 . Since HPK-1 is a protein kinase that co-localizes into subnuclear foci with HSF-1 , the simplest model is that HPK-1 stimulates HSF-1 activity through direct phosphorylation . There are multiple examples in mammals that phosphorylation of a transcription factor/co-factor directly prevents subsequent sumoylation , thereby increasing the activation potential of that transcription factor; examples include PML protein , p53 , and c-Jun[81 , 82] . While we were unable to resolve a phospho-isoform of HSF-1 attributable to HPK-1 activity , not all phosphorylation events produce a mobility shift . Alternatively , our antibody may not recognize the relevant phospho-isoform of HSF-1 . It would also be informative to test whether HSF-1 regulation requires an active HPK-1 kinase domain . There are at least 19 phosphorylation sites on human HSF1 , a subset of which are conserved in C . elegans ( S5 Table and[41] ) . Consistent with the possibility of direct interaction between HPK-1 and HSF-1 , S . cerevisiae Yak1 directly phosphorylates Hsf1 in response to glucose stress , albeit within a region not conserved in multicellular eukaryotes [36] . Alternatively , HPK-1 may act indirectly to prevent HSF-1 sumoylation . In mammals both sumoylation and acetylation occurs at HSF-1 K298[49] , the latter through the opposing acetyltransferase activities of p300 and SIRT1[41] . Consistent with the possibility of indirect interaction , mammalian HIPK2 directly phosphorylates SIRT1 to restrict activity after DNA damage[83] . Given the complexity of post-translational modification to HSF-1 , determination of whether HPK-1 directly phosphorylates HSF-1 , and identification of the relevant residue ( s ) will require analysis using mass spectrometry . Sumoylation of transcription factors is often associated with reduced transactivation activity [84] . Specifically , sumoylation of HSF family members has previously been reported in mammals , and these isoforms possess decreased transcriptional activity[39 , 85] . Our findings that smo-1 ( RNAi ) leads to an increase in HSF-1 chaperone induction in response to heat shock support a hypothesis in which sumoylation of HSF-1 is inhibitory in C . elegans as well . Moreover , our data support a role for HPK-1 in preventing HSF-1 sumoylation ( and its coupled phosphorylation event ) through the end of development , at which point HPK-1 becomes restricted to low level expression in neurons . This timing for the loss of systemic HPK-1 expression correlates with the onset of transcriptional silencing at stress loci[30] and a decline in multiple additional protein quality control mechanisms ( reviewed in[29] ) . Opposition of sumoylation by HPK-1 may provide a molecular mechanism for a long-standing question in the aging field concerning the decline of HSF-1 activity in aging animals . In mammals , HSF1 DNA binding activity and chaperone expression levels both decline with aging , while the abundance of HSF1 protein does not[41] . We propose that sumoylation of HSF-1 is inactivating , and that increased accumulation of sumoylated HSF-1 with reproductive age explains the decline of basal HSF-1 activity and the resulting decay of proteostasis in aging animals . We also propose that a critical longevity function of HPK-1 is to protect HSF-1 against age-dependent inactivation—by delaying/preventing sumoylation or by driving de-sumoylation . We describe earlier in the discussion how we initially identified a role for hpk-1 in longevity control , and provide the rationale for our conclusion that the pro-longevity functions of hpk-1 belong to pathway ( s ) acting in parallel to the ILS/FOXO signaling pathway . We have shown that hpk-1 promotes HSF-1 mediated transcription—within the context of the heat shock pathway and basal unstressed conditions—to promote global proteostasis via chaperone systems . We also wondered if hpk-1 might function more broadly to regulate stress-responsive transcription factors . We suspected that hpk-1 might modulate nutritional stress responses controlled by the TOR signaling pathway . First , a wealth of data in mammals and C . elegans places daf-15 ( Raptor , a TORC1 subunit ) in a pathway parallel to ILS based on distinct activities within each pathway that modulate longevity[60] . However , daf-2/ILS and daf-15/TOR do not always function in isolated linearity with respect to each other , and under certain conditions these kinases converge on common transcription factors necessary for extension of longevity , including SKN-1 ( Nrf2 ) , HSF-1 , MXL-2 , and DAF-16[21 , 50 , 63 , 72 , 73 , 86–88] . Only a subset of transcriptional targets is shared . For example , TORC1 inactivation does not stimulate DAF-16 nuclear translocation and activation ( as daf-2 inactivation does ) [73] . Though the importance of HIPKs in orchestrating stress responses has only begun to emerge , several pieces of evidence suggest that HIPKs may reside at the crossroads of ILS and TOR signaling . In mammals , multiple mechanisms link insulin/IGF ( insulin-like growth factor ) levels to activation of mTORC1[89] . Knockdown of HIPK1 , 2 or 3 attenuates insulin secretion in response to glucose and HIPKs bind to insulin promoters , suggesting that HIPKs may activate insulin expression[37] . Glucose deprivation can also activate Hipk1[75 , 90] . In budding yeast , the HIPK homolog Yak1 is activated by rapamycin treatment[91] and by glucose depletion[92] . For these reasons , it seemed worth testing if hpk-1 might function within the TOR pathway , potentially as a point of regulation between TOR growth signaling and the nutrient deprivation transcriptional programs repressed by TOR . First , we found that inactivation of daf-15 stimulated neuronal expression of HPK-1 , while inactivation of daf-2 did not . HPK-1 induction in response to daf-15 RNAi was restricted to the nervous system , in contrast to the broader induction of HPK-1 in response to heat shock that also included HPK-1 expression in seam cells and the intestine ( weakly ) . When we examined the transcriptional programs controlled by TOR , we found that HPK-1 was necessary for the induction of autophagy genes bec-1 and atg-8 following TOR inactivation , but that hpk-1 had no effect on the transcription of translation initiation factor genes induced by TOR activation . In further support that HPK-1 can stimulate autophagy , we observed that the induction of autophagosome formation ( LGG-1::GFP ) in response to DR required hpk-1 while interestingly , hsf-1 was not required . Stimulation of autophagy by hpk-1 in response to DR must therefore involve the activation of other transcription factor ( s ) that are not HSF-1 . Autophagy is thought to occur in at least two separate phases: a rapid increase in autophagic flux that occurs entirely through post-translational protein modifications within minutes to hours after stress , which is followed by a subsequent extended phase reliant on the activation of transcriptional programs[17 , 64] . Dietary restriction and TOR inactivation induce protein turnover in C . elegans by stimulating autophagy , which requires multiple DR-responsive transcription factors , including: pha-4 ( FoxA ) , mxl-2 ( MLX ) , hlh-30 ( TFEB ) and nhr-62 ( NHR4α ) . In addition to transcriptional controls , the TORC1 complex inhibits autophagy directly by inhibitory phosphorylation of autophagy components necessary for initiation of autophagy . Because hpk-1 functions biologically by activation of transcription factors , it seems likely that HPK-1 acts during the “extended phase” of autophagy induction by activating one or more of the transcription factors required for autophagy induction in response to DR . Our results are consistent with a general model in which HPK-1 promotes protein homeostasis by two separate mechanisms , each of which can function under basal as well as stressed conditions . HPK-1 stimulates transcription of molecular chaperones via HSF-1 , which decreases the physiological load of misfolded proteins in vivo . In parallel , HPK-1 stimulates autophagy via PHA-4 and MXL-2 , allowing existing proteins to be catabolized into useable metabolites , also reducing the physiological protein load . Under conditions of heat shock , the requirement for HPK-1 becomes more pronounced as the need for a proteostasis compensation mechanism is dramatically increased in response to protein unfolding . Under conditions of nutrient stress , proteostasis is not directly compromised , but the demand for metabolic building blocks must be met using existing cellular resources . In this case , protein turnover by autophagy may be an effective way to supply essential metabolites; at the same time that it improves proteostasis by clearing protein aggregates . A recent study identified a role for HSF-1 in the induction of autophagy after hormetic heat shock or ectopic over-expression of hsf-1[70] . Our finding that hpk-1 is essential for the beneficial effect of hormetic heat shock on lifespan ( S4G Fig ) would be consistent with the notion that hpk-1 may also be essential for the induction of autophagy gene expression conferred by either hsf-1 overexpression or hormetic heat shock , as well as after nutrient stress . Induction of autophagosome formation may also be regulated differently in different tissues: we find that hpk-1 but not hsf-1 is required for autophagosome formation in hypodermal seam cells , while heat mediated autophagosome formation in muscle , intestinal , and cells within the nerve ring require hsf-1[70] . Upon nutrient deprivation , hlh-30 ( TFEB ) , pha-4 ( FoxA ) , the nuclear hormone receptor nhr-62 , and the Myc family transcription factors mml-1 ( Mondo A/ChREBP ) and mxl-2 ( Mlx ) have all been shown independently to be essential for at least one aspect of the induction of autophagy[21 , 22 , 64] . TORC1 inactivation is thought to increase lifespan by inducing autophagy and by decreasing global protein translation ( reviewed in[65] ) . Dietary restriction reduces TOR signaling[93] . We and others have shown that either conditions of dietary restriction or inactivation of TORC1 depend upon hpk-1 , pha-4 and mxl-2 to extend longevity ( this study and[21 , 67 , 72 , 73] ) . Similarly , we show that the extension of longevity conferred by overexpression of hpk-1 is suppressed by pha-4 and mxl-2 , but not hlh-30 or nhr-62 ( Fig 13 ) , suggesting that autophagy contributes to the extended longevity of animals overexpressing HPK-1 in addition to the induction of chaperone genes driven by HSF-1 . HPK-1 could represent the crucial regulation point between TORC1 and its target transcription factors PHA-4 and MXL-2 . We have demonstrated that TOR inhibits expression of hpk-1 mRNA ( Fig 10E ) and protein levels of HPK-1 ( Figs 10D , S8 ) . Mammalian Hipks are unstable and HPK-1 could be inhibited by TOR post-translationally through regulation of its stability . A particularly exciting possibility is that HPK-1 acts as a nutritional switch operated by TOR . For instance , phosphorylation of HPK-1 by TOR may destabilize HPK-1 and/or target it for degradation . Under replete conditions , autophagy gene transcription would be OFF because HPK-1 protein levels are being maintained at only low levels by TOR kinase . Inactivation of TOR kinase by dietary restriction would lead to stabilization of HPK-1 , allowing protein levels to accumulate . HPK-1 could in turn phosphorylate and activate downstream transcription factors ( PHA-4 and MXL-2 ) that stimulate expression of genes necessary for autophagosome biogenesis . A complication of this model is that it only provides a sequence of action for TOR , HPK-1 and PHA-4/MXL-2 when they are functioning within the same tissue . We have shown that TORC1 ( RNAi ) stimulates HPK-1 expression in neurons , while the Q35::YFP transgene is only expressed in muscle . Therefore , it is likely that endocrine signals arise downstream of HPK-1 in order to regulate cellular functions like autophagy in non-neuronal tissues . Precedent for the neuroendocrine relay of stress responses has arisen in recent years , including the unfolded protein responses of the ER UPRER , mitochondria ( UPRmito ) , and the heat shock response[25 , 94–96] . Activation of either of these stress response pathways in neurons stimulates activation of the self-same UPR in non-neuronal tissues , and a recent report suggests that the transcription factor actors of these responses are required both “upstream” ( in neurons ) and “downstream” ( in non-neuronal tissues ) in the neuroendocrine circuit . A specific example is stimulation of the ERUPR in non-neuronal tissues by activation of the UPRER in neurons . XBP-1 , a cell autonomously acting transcription factor is required in both neurons and non-neuronal tissues in order for the UPRER activation in neurons to elicit the UPRER in distal tissues . Aside from evidence that serotonergic signaling is involved , the mechanism ( s ) by which specific cellular stress response pathways send or receive endocrine signals between tissues in order to generate systemic responses has yet to determined[97 , 98] . In addition to evaluating the requirement for autophagy transcription factors in mediating the increased lifespan arising from HPK-1 overexpression , we tested the possibility that PHA-4 , MXL-2 and HLH-30 were targets of HPK-1 regulation in the proteostatic context of polyQ aggregation . First we quantitated the dynamics of Q35::YFP aggregate formation and toxicity in the presence or absence of pha-4 , mxl-2 and hlh-30 using RNAi . We found that inactivation of all three transcription factors significantly accelerated the accumulation of Q35::YFP aggregates . Inactivation of pha-4 and mxl-2 accelerated the rates of toxicity-induced paralysis . Curiously , hlh-30 RNAi partially rescued the paralysis phenotype of the hpk-1 ( pk1393 ) mutant . In the Q35::YFP; hpk-1 ( pk1393 ) background , inactivation of pha-4 , mxl-2 or hlh-30 did not increase aggregate formation above the already elevated level caused by the absence of hpk-1 . Similarly , pha-4 ( RNAi ) and mxl-2 ( RNAi ) had no increased effect on the accelerated rates of paralysis observed in the hpk-1 ( pk1393 ) mutant strain relative to wild type . Our finding that inactivation of pha-4 and mxl-2 causes an increase in aggregate formation suggests that autophagy can in fact mitigate the formation of toxic Q35::YFP aggregates in muscle cells . This supports our model: extension of longevity depends specifically on the stimulation of autophagy by hpk-1—via activation of the pha-4 and mxl-2 transcription factors—and suggests that HPK-1 , PHA-4 , and MXL-2 may comprise a larger transcriptional circuit to regulate autophagy gene expression under specific conditions of nutrient stress . Aging is highly dependent upon the cellular processes that maintain proteostasis[99 , 100] , and diverse age-related neurodegenerative diseases ( e . g . Alzheimer , Parkinson and Huntington diseases ) are linked to compromised autophagy and the decline of proteostasis[99–102] . Proteostasis is achieved by the careful balancing of rates of protein synthesis with chaperone-mediated protein folding and turnover of unfolded polypeptides via proteasome- and autophagy-mediated degradation pathways . When proteostability is compromised , HSF-1 promotes restoration of proteostasis by induction of molecular chaperones and proteosomal degradation of misfolded proteins[49] . In contrast , nutrient rich conditions challenge proteome stability by activation of the TOR pathway , which stimulates protein translation and inhibits protein turnover . When nutrients are limited , TOR kinase is inactivated , prompting protein degradation through autophagy into constituent metabolites , which serves the dual function of reducing the overall concentration of proteins that must remain soluble while providing biosynthetic building blocks that can be remobilized towards pro-survival processes[103] . It is worthy of notice that HPK-1 and TOR act in physiological opposition; HPK-1 promotes catabolic processes that decrease proteotoxic stress , while TOR promotes an anabolic physiological state represented by an overall increase in cellular protein . Overexpression of HPK-1 or inhibition of the TOR pathway promotes proteome re-stabilization and concomitant extension of longevity . Most neurodegenerative diseases are characterized by an age-dependent onset , which share the common molecular feature of abnormal protein aggregation leading to neuronal cell death . A number of studies have linked loss of either HSF1 or autophagy to the promotion of neurodegenerative diseases[104 , 105] . We have discovered that HPK-1 functions as a central regulatory node linking these two complementary proteostatic mechanisms that act to preserve the proteome ( Fig 15 ) , and justifies consideration of HPK-1 as an attractive intervention point for the improvement of healthspan and protection against proteostatic diseases . RNAi clones originated from early copies of E . coli glycerol stocks of the comprehensive RNAi libraries generated in the Ahringer and Vidal laboratories . Each RNAi colony was grown overnight in Luria broth with 50 μg ml–1 ampicillin and then seeded onto 24-well RNAi agar plates containing 5 mM isopropylthiogalactoside to induce dsRNA expression overnight at room temperature . All of the RNAi clones used in this study were verified by DNA sequencing and subsequent BLAST analysis to confirm their identity . Fusion of promoter and gene sequences to reporters were carried about using the SOEing technique [106] . Pooled PCR products were microinjected to generate transgenic lines as follows . Phpk-1::GFP: 2kb upstream of ATG start for the longest isoform of hpk-1 was amplified from fosmid WRM066aH12 and fused to GFP ( from pPD95 . 75 ) ( Addgene ) using the following primers: 30ng of fused PCR product was injected with 120ng of pRF4 ( rol-6 ) as a coinjection marker . Phpk-1::hpk-1::GFP: C-terminal fusion of full-length hpk-1 to GFP was generated using the fosmid template WRM066aH12 and pPD95 . 75 as above with following primers: 5ng of PCR product was injected into wild type and hpk-1 ( pk1393 ) animals with 150ng of pRF4 as coinjection marker . Phpk-1::hpk-1::tdtomato C-terminal fusion of full-length hpk-1 to GFP was generated using the fosmid WRM066aH12 and pDS9 ( tdTomato ) as templates For visualizing co-localization of HSF-1::GFP with HPK-1::tdTomato , pAH71 ( P-hsf-1::hsf-1::GFP ) was obtained from Dr . Hsu ( University of Michigan , Ann Arbor ) and injected at a concentration of 15ng/ul either with 5ng/ul or 30ng/ul of P-hpk-1::hpk-1::tdtomato . Psur-5::hpk-1::CFP: 1 . 5kb promoter region of sur-5 amplified from genomic DNA was fused to hpk-1 genomic fragment and CFP at the C-terminus in a nested PCR reaction . CFP was amplified from pPD134 . 96 . 5ng/ul of the PCR fusion construct , 5ng/ul of Pmyo-2::m-cherry ( pCFJ90 ) , and Log2 DNA ladder ( NEB N3200S ) was included as carrier DNA . Lifespan assays were performed essentially as described [107] . Synchronized L1 animals were seeded onto plates and allowed to develop to L4 stage at 20°C , with two exceptions: for the hormesis experiments ( S5G and S5H Fig ) animals were shifted to 25°C at L4 and maintained at this elevated temperature until day 3 of adulthood and then shifted back to 20°C; and for the intestinal specific RNAi ( S2D Fig ) animals were maintained at 25°C during development ( this strain is a pha-1 rescue strain that must be maintained at 25°C during development to ensure maintenance of the extrachromosomal transgene[108] ) . In all cases , at L4 2’ fluoro– 5’deoxyuridine ( FUDR ) was added to a final concentration of 400 uM to prevent progeny production ( defined as day 0 adulthood ) . Viability was scored every day or every other day as indicated in each figure . Survival analysis was performed for all the experiments using the Kaplan-Meier estimator and Peto & Peto’s generalized Wilcoxon test , both as implemented in the R package survival version 2 . 38 ( http://CRAN . R-project . org/package=survival ) . P-values were corrected for multiple testing with the Benjamini-Hochberg FDR approach[109] . To determine temporal requirements in longevity control ( S2A–S2C Fig ) , synchronized L1 animals were seeded on control or hpk-1 RNAi according to the following conditions: for constitutive RNAi L1 animals exposed to each of the RNAi conditions throughout lifespan , for developmental gene inactivation animals were fed E . coli expressing dsRNA to hpk-1 or control ( empty vector ) from L1-L4 after which animals were moved to dcr-1 ( dicer-1 ) RNAi , and for adult inactivation animals were maintained on control RNAi plates until L4 and then moved to either control , hsf-1 or hpk-1 RNAi plates . Detailed information on all lifespan trials , including number of animals scored can be found in S1 Table . Time course survival assays at high temperature were conducted using the replica set method as previously described [72] . In brief , synchronized L1 animals were allowed to develop at 20°C; FUDR was added at the L4 stage and at day three adulthood animals were moved to 35°C for the indicated time period , allowed to recover for 1–2 hours at 20°C , and viability was scored every 2 hours for up to 14 hours . Significance of pairwise comparisons between genotypes was determined by fitting the data to logit curves with pooled data from all three trials and randomized permutation testing with 100 , 000 iterations to determine p-values , as previously reported in [38 , 72] . For single time point thermotolerance assays , synchronized day 1 adult animals were heat shocked at 35°C for 9–10 hours and viability was scored the next day . 3–4 technical replicates were included in each of two independent trials . Statistical testing for single time point assays was performed with ANOVA followed by Tukey’s HSD post-hoc test , and p-values were corrected to account for multiple testing . Synchronized Punc-54::polyQ35::YFP L1 progeny were treated with the indicated RNAi and FUDR was added at L4 stage to prevent progeny production . To quantify fluorescent foci formation , between 15–50 animals per technical replicate were scored blind daily from days one to three of adulthood . To assess paralysis , prodded animals that responded with head movement ( and were therefore still alive ) but no backward or repulsive motion were scored as paralyzed ( as described in [72] ) . Statistical testing between pairs of conditions for differences in paralysis rates was performed with Cox modeling and the Wald test . Rabbit polyclonal antibodies to C . elegans HSF-1 were generated using purified , recombinant C . elegans HSF-1 , which was recovered from E . coli expressing His6-HSF-1 from a modified version of the pET-28a ( + ) vector by affinity purification and subsequent on column thrombin cleavage to remove the His6 tag . Polyclonal rabbit antibodies to the full-length , untagged protein were generated by Covance ( http://www . covance . com/services/lead-optimization/immunology/custom-antibody-development . html ) through injection of two rabbits , of which the one with the least background immunogenicity was chosen . Approximately 1000 synchronized animals were collected with M9 , washed , and pelleted animals were snap frozen in liquid nitrogen . Pellets were lysed in RIPA buffer ( 150 mM NaCl , 1 . 0% NP-40 , 0 . 5% sodium deoxycholate , 0 . 1% SDS , 50 mM Tris ( pH 8 . 0 ) ) with 40 mM N-Ethylmaleimide ( NEM , Thermo #PI23030 ) ; per 1 mL RIPA , 40 μL of 100x Halt Protease & Phosphatase Inhibitor ( Thermo #78446 ) , and 40 μL 100x EDTA ( 0 . 5 M ) was added . Briefly , extracts were vortexed with Zirconin 2 . 0 mm beads ( #11079124ZX , BioSpec ) and lysate was separated from pellet by centrifugation . Samples were normalized using a Bradford assay and was resolved by 6% SDS-PAGE . The antibodies used to probe membranes of the immunoblots were anti-HSF-1 ( 1:500 ) , and anti-beta-actin ( 1:1000 ) ( Cell Signaling #4967 ) . The ratio of HSF-1 unmodified ( 75kD ) to modified ( 90-95kD , sumoylated and sumoylated plus phosphorylated ) was determined based on the quantification of band intensity in Image J . Protein levels of unmodified ( 75kD ) , modified ( 90-95kD ) HSF-1 , and beta-actin were also quantified in ImageJ from western blots . HSF-1 expression was first normalized to matched beta-actin levels and then relative fold-change compared to N2 was calculated . The S . E . M . from quantified extracts/western blots was calculated and in both cases significance was determined by Student’s t-test . For detection of HSP-16 . 2 protein and GFP reporter expression on RNAi , 100 worms grown at 16°C on RNAi were collected in 25 μL dH2O . An equal volume of 2x SDS-PAGE loading dye was added followed by 15 min boiling . A 10–20% Tris-HCl gel ( Bio-Rad ) was loaded with 15μL ( ~30 worms ) after another 15 min boiling , run , transferred to nitrocellulose , and blocked in 2% milk in 1X TTBS ( 1M Tris , 150 mM NaCl , 0 . 1%Tween 20 ) . Membrane was probed simultaneously for HSP-16 . 2 ( 1:5000 , rabbit , #5506 R120; kind gift of Chris Link , UC Boulder ) , GFP ( 1:1000 , mouse , Roche 7 . 1 and 13 . 1 ) and β-actin ( 1:2000 , mouse , Sigma AC-15 ) at 4°C overnight . Blots were washed in 0 . 1% milk 1X TTBS three times . Secondary antibodies were anti-mouse HRP and anti-rabbit HRP ( 1:5882 dilution both , Amersham ) . Blots were visualized with Amersham ECL Western Blotting System ( RPN2108 ) . For HSF-1 immunoprecipitations , L4 C . elegans pellets were collected as per sample collection for immunoblotting . C . elegans pellets were lysed in 20 mM Tris-HCl ( pH 8 . 0 ) , 170 mM KCl , 0 . 5% NP-40 ( IGEPAL CA-360 ) , 2 mM EDTA , with 40 μL of 100x Halt Protease & Phosphatase Inhibitor , 40 μL 100x EDTA ( 0 . 5 M ) ( Thermo #78446 ) per mL of lysis buffer . Lysates were prepared and quantified as above , and then 1 . 5 mg of lysate was precleared by the addition of 10 μL of normal rabbit serum ( Invitrogen # 016101 ) and incubated with rotation for 1 h at 4°C , at which time the equivalent of 120 μL of drained Protein-A-Sepharose was added to immunoprecipitate non-specific interactions and incubation continued an additional 2 h . The proteins were eluted from the nonspecific pellet fraction with the addition of equal volume 2x SDS reducing buffer ( 1x SDS reducing buffer: 62 . 5 mM Tris ( pH 6 . 8 ) , 10% glycerol , 0 . 02% SDS , and 5% β-mercaptoethanol ) and boiled for 5 min . Precleared lysate was evenly divided , antibodies were added , and lysates were incubated with rotation for 2 h at 4°C . The antibodies used in various immunoprecipitations were as follows: for HSF-1 1 μL of HSF-1 immune serum , and for negative controls either 1 μL of commercially available NRS ( Invitrogen # 016101 ) or pre-immune serum from the rabbit prior to injection with recombinant C . elegans recombinant HSF-1 . The equivalent of 50 μL of drained beads of Protein A-Sepharose was added , and lysates were incubated with rotation for an additional 2 h at 4°C . Immunoprecipitates were washed three times with lysis buffer , resuspended in 2x SDS reducing buffer , boiled for 5 min , and resolved by 6% SDS-PAGE . Subsequent immunoblotting for HSF-1 was carried out as described above . Prior to elution after immunoprecipitation , beads of Protein A-Sepharose were washed three times with 1x SUMO Protease Buffer ( 50 mM Tris-HCl ( pH8 . 0 ) , 0 . 2% Igepal ( NP-40 ) , 150 mM NaCl , 1 mM DTT ) , and resuspended in 100 μL of 1x SUMO protease buffer with or without 5 μL of SUMO protease ( ULP1; Invitrogen #12588–018 ) . Reactions were performed at 20°C for 1 hour with nutating . Immunoprecipitate was eluted with the addition of equal volume 2x SDS reducing buffer and boiled for 5 min . Extracts were prepared as above but with HALT complete protease inhibitor cocktail ( Thermo #87785 ) without EDTA . Protein dephosphorylation was carried out with lambda Protein Phosphatase ( λPP , New England Biolabs #P0753S ) treatment . For λPP treatment 25 μg total protein was incubated for 1 h at 30°C in 1x NEBuffer for MetalloPhosphatases ( PMP ) , 1 mM MnCl2 , and 400U of λPP . As a negative control mock λPP treatments used enzyme that had been inactivated by 10 min of boiling prior to addition and samples contained Halt Protease & Phosphatase Inhibitor ( Thermo #78446 ) . In all cases , after 1 h of incubation SDS reducing buffer was added and samples were boiled 5 minutes . For measurement of hsp gene induction in response to heat shock , late L4 animals were heat shocked at 37°C for one hour and allowed to recover for two hours before harvesting in M9 buffer . For measurement of atg-18 , bec-1 , ifg-1 , iftb-1 , hsp-16 . 2 , and hsp-70 mRNA levels after daf-15 inactivation , synchronized populations were grown until L4 at 20°C on RNAi plates and animals were harvested on day 1 of adulthood . RNA was isolated using Trizol reagent ( Life Technologies ) . RNA concentration was measured using a Nanodrop , and RNA preparations were reverse transcribed into cDNA using the BioRad cDNA synthesis kit ( #1708890 ) as per manufacturer’s protocol . For all qRT-PCR primer pairs , ( - ) RT test reactions were run to confirm mRNA purity and target specificity . Quantitative real time PCR was performed using SYBR green supermix ( Biorad ) with three biological and two technical replicates for each condition . Primer sets with at least one primer spanning the exon were used to amplify the gene of interest . cdc-42 mRNA levels were used for normalization of autophagy and translation gene expression . act-1 mRNA levels were used for normalization of hsp gene expression . Fold change in mRNA levels was determined using ΔΔ Ct method [110] . Primer sequences used were as follows: This study did not employ Human Subject Research or Animal Research that would require IACUC oversight or approval .
|
Aging is the gradual and progressive decline of vitality . A hallmark of aging is the decay of protective mechanisms that normally preserve the robustness and resiliency of cells and tissues . Proteostasis is the term that applies specifically to those mechanisms that promote stability of the proteome , the collection of polypeptides that cells produce , by a combination of chaperone-assisted folding and degradation of misfolded or extraneous proteins . We have identified hpk-1 ( encoding a homeodomain-interacting protein kinase ) in the nematode C . elegans as an important transcriptional regulatory component of the proteostasis machinery . HPK-1 promotes proteostasis by linking two distinct mechanisms: first by stimulating chaperone gene expression via the heat shock transcription factor ( HSF-1 ) , and second by stimulating autophagy gene expression in opposition to the target of rapamycin ( TOR ) kinase signaling pathway . HPK-1 therefore provides an attractive target for interventions to preserve physiological resiliency during aging by preserving the overall health of the proteome .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion"
] |
[
"cell",
"death",
"invertebrates",
"autophagic",
"cell",
"death",
"rna",
"interference",
"classical",
"mechanics",
"caenorhabditis",
"gene",
"regulation",
"regulatory",
"proteins",
"cell",
"processes",
"dna-binding",
"proteins",
"mechanical",
"stress",
"neuroscience",
"animals",
"animal",
"models",
"caenorhabditis",
"elegans",
"model",
"organisms",
"sumoylation",
"experimental",
"organism",
"systems",
"transcription",
"factors",
"epigenetics",
"research",
"and",
"analysis",
"methods",
"genetic",
"interference",
"proteins",
"thermal",
"stresses",
"animal",
"cells",
"gene",
"expression",
"physics",
"biochemistry",
"rna",
"cellular",
"neuroscience",
"eukaryota",
"cell",
"biology",
"nucleic",
"acids",
"post-translational",
"modification",
"neurons",
"genetics",
"nematoda",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"physical",
"sciences",
"organisms"
] |
2017
|
The homeodomain-interacting protein kinase HPK-1 preserves protein homeostasis and longevity through master regulatory control of the HSF-1 chaperone network and TORC1-restricted autophagy in Caenorhabditis elegans
|
Computational methods attempting to identify instances of cis-regulatory modules ( CRMs ) in the genome face a challenging problem of searching for potentially interacting transcription factor binding sites while knowledge of the specific interactions involved remains limited . Without a comprehensive comparison of their performance , the reliability and accuracy of these tools remains unclear . Faced with a large number of different tools that address this problem , we summarized and categorized them based on search strategy and input data requirements . Twelve representative methods were chosen and applied to predict CRMs from the Drosophila CRM database REDfly , and across the human ENCODE regions . Our results show that the optimal choice of method varies depending on species and composition of the sequences in question . When discriminating CRMs from non-coding regions , those methods considering evolutionary conservation have a stronger predictive power than methods designed to be run on a single genome . Different CRM representations and search strategies rely on different CRM properties , and different methods can complement one another . For example , some favour homotypical clusters of binding sites , while others perform best on short CRMs . Furthermore , most methods appear to be sensitive to the composition and structure of the genome to which they are applied . We analyze the principal features that distinguish the methods that performed well , identify weaknesses leading to poor performance , and provide a guide for users . We also propose key considerations for the development and evaluation of future CRM-prediction methods .
Cis-acting transcriptional regulation involves the sequence-specific binding of transcription factors ( TFs ) to DNA [1] , [2] . In most cases , multiple transcription factors control transcription from a single transcription start site cooperatively . A limited repertoire of transcription factors performs this complex regulatory step through various spatial and temporal interactions between themselves and their binding sites . On a genome-wide scale , these transcription factor binding interactions are clustered as distinct modules rather than distributed evenly . These modules are called cis-regulatory modules . On DNA sequences , promoters , enhancers , silencers and others , are examples of these modules . The appropriate transcription factors compete and bind to these elements , and act as activators or repressors . Generally a CRM ranges from a few hundred basepairs long to a few thousand basepairs long; several transcription factors bind to it , and each of these transcription factors can have multiple binding sites ( Figure 1 ) . Berman et al . [3] demonstrated the feasibility of identifying CRMs by sequence analysis . They scanned the Drosophila genome for clusters of potential binding sites for five gap gene transcription factors that are known to , together regulate the early Drosophila embryo . They found more than a third of the dense clusters of these binding sites correspond to be CRMs regulating early embryo gene expressions , characteristic of genes regulated by maternal and gap transcription factors . Similarly , exploiting the property that many CRMs contain clusters of similar transcription factor binding sites ( TFBSs ) , Schroeder et al . [4] computationally predicted CRMs over the genomic regions of genes of interest with gap or mixed maternal-gap transcription factors , and identified both known and novel CRMs within the segmentation gene network . Recent study has confirmed the importance of CRM functions , and revealed how subtle changes to the original arrangement of module elements can affect its function . Gompel et al . [5] found that modifications to cis-regulatory elements of a pigmentation gene Yellow can cause a wing pigmentation spot to appear on Drosophila biarmipes similar to that seen in Drosophila melanogaster , thus showing that mutations in CRMs can generate novelty between species . In a later study [6] they showed the creation and destruction of distinct regulatory elements of same gene can lead to a same morphological change . Williams et al . [7] investigated the genetic switch whereby the Hox protein ABD-B controls bab expression in a sexually dimorphic trait in Drosophila . They discovered the functional difference of this case lies not only in the creation and destruction of the binding sites , but also in their orientations and spacings . There is also evidence showing that disruption of cooperations within a specific CRM can lead to malformation and disease . One example is given by Kleinjan et al . [8] . The deletion of any distal regulatory elements of PAX6 changes its expression level and causes congenital eye malformation , aniridia , and brain defects in human . Methods attempting to identify CRMs in the genome face a challenging problem: a module is a mixture of signals – transcription factor binding sites and other sequence features – and these signals are spatially clustered within a specific genomic interval and are frequently , but not universally , conserved between related species [9] . Searching for a cis-regulatory module consists of searching for two properties: a set of signals , and the spatiotemporal relationships between this set of signals . In order to identify CRMs , one must first define and build a model . Except for a small number of specific , well-characterized , interactions , the vast majority of spatiotemporal relationships between transcription factors remain unknown . This information deficit limits most CRM prediction methods to defining CRMs based on their general features: their spatial constraints ( i . e . a close distance between binding sites within a CRM ) , their phylogenetic constraints ( i . e . a CRM is a conserved block between species ) [10]–[12] , or both . Therefore , pre-compiled binding site profile libraries and multiple genome alignments are required by many CRM prediction methods . The search strategies for the existing methods can be roughly classified into four families . Window clustering involves significant clustering of high densities of binding sites within a sequence window . Probabilistic modelling consists of identifying sequences that resemble a statistical model of a binding site cluster more than a model of background DNA . Phylogenetic footprinting searches for high density regions of binding sites conserved between closely related species . Discriminative modelling seeks to identify set of signals on regulatory regions that can maximize the differences between regulatory regions and non-regulatory regions ( Figure 2 ) . Many methods are hybrids of two or more strategies . We wished to understand the performance of CRM prediction methods and , if possible , identify an optimal method . We also hoped to locate the key features that distinguish a good method and the reasons behind it . More specifically , we would like to answer these questions: ( 1 ) Which search strategy best predicts CRMs ? ( 2 ) What types of CRMs are easy or difficult to predict ? ( 3 ) What causes false positives and false negatives , and how they can be reduced in the future ? In order to examine all of these features of CRM prediction methods , we selected twelve representative methods from the above four search strategies families: MSCAN [13] , MCAST [14] , ClusterBuster [15] , Stubb [16] , StubbMS [17] , MorphMS [18] , CisModule [19] , MultiModule [20] , CisPlusFinder [21] , phastCons score [22] ( http://hgdownload . cse . ucsc . edu/goldenPath/dm2/phastCons9way/ ) , Regulatory Potential [23] and EEL [24] . These twelve methods cover almost all the possible combinations of CRM representations , information resources used and search strategies available , as shown in the the summary table ( Figure 3 ) . Their operational principles are summarized ( Table 1 ) . Among these twleve methods , Regulatory Potential and EEL only have results available for the human genome . Therefore the other ten methods were applied to predict the CRMs in the Drosophila CRM database REDfly [25] to assess their general predictive power . Next , three optimal methods from the REDfly prediction result , together with Regulatory Potential and EEL , were applied to the human ENCODE regions , to assess the utility of these methods when dealing with different genomes of various compositions and structures . The family of window clustering methods , such as MSCAN , MCAST and CisPlusFinder , represent a CRM in a most naïve form as a statistically significant clustering of high affinity transcription factor binding sites . MSCAN and MCAST scan a motif library against a single genome . CisPlusFinder takes the perfect local ungapped sequences as potential transcription factor binding sites , then searches for a high density of multiple such short sequences that are conserved between closely related species . The family of probabilistic modelling methods , ClusterBuster , Stubb , StubbMS , MorphMS , CisModule , and MultiModule , all implement a hidden Markov model ( HMM ) and they model a CRM sequence as being generated by a combination of a set of binding sites . The difference between them is ClusterBuster , Stubb and CisModule are based on a single genome while StubbMS , MorphMS and MultiModule are based on a pair or multiple orthologous genomes . Morever , the difference between StubbMS and MorphMS lies on their first step of aligning their input orthologous sequences: StubbMS uses Lagan [26] that produces a fixed alignment according to the sequence similarity . On the contrary , MorphMS aligns sequences by probabilistically summing over all possible alignments by their matches to the potential binding sites . CisModule and MultiModule are unique from the rest methods of this family by predicting both binding sites and CRMs in one step . CisModule encodes binding sites and a CRM into one hierarchical mixture model and follows Bayesian inference to predict both the location of CRM and the location of the binding sites within the CRM simultaneously . MultiModule follows the same model but improves on CisModule by incorporating information from comparative genomes . Among the above two families of methods , the methods using multiple alignments: CisPlusFinder , StubbMS , MorphMS and MultiModule are also members of the phylogenetic footprinting family . Among these ten methods , CisModule , MultiModule and CisPlusFinder are the three methods that do not rely on the prior information of a motif library . To further check how well the functional CRMs can be predicted without additional binding site knowledge , we applied a method based purely on sequence conservation – as represented by phastCons score [22] – as an independent calibration . PhastCons score is calculated by a phylogenetic hidden Markov model considering the evolutionary distance between species . It assigns each nucleotide position a score which represents the conservation degree of that position . We followed the approach used by King et al . [27] and took continuous windows with a mean phastCons score over an optimized phastCons score threshold as a potential CRM ( see Materials and Methods ) . We also identified a few interesting methods which we were unable to include in this assessment due to incompatibility with the experimental design of this study or unavailability of required data . For example HexDiff [28] , a method in the discriminative modelling family , learns a set of over-represented hexamers in known CRM sequences , and discriminates CRM sequences from non-CRM sequences by searching for the highest frequency hexamers . Such a method requires correctly annotated positive and negative datasets of known CRMs to assess its performance . Regulatory Potential [23] is another type of discriminative method , which learns the abundant hexamers and the first order dependency relationships between columns of aligned position from known regulatory regions . Similar to MorphMS are EEL [24] , PhylCRM [29] and EMMA [30] , which aim to better use multiple genome information by implementing binding site-based alignment methods . EEL considers the potential secondary structure of a DNA-protein complex by weighting the difference in the distance between adjacent binding sites between the two aligned species . PhylCRM uses MONKEY [11] directly to predict true functional binding sites in its first step . MONKEY uses multiple alignments and models the binding sites of each transcription factor with a specific evolutionary model . Thus , the binding sites predicted by MONKEY are enriched for true conserved functional sites among those gained , lost and turned over . EMMA takes a similar approach as MorphMS but incorporates binding site gains and losses . However , this makes its computational cost increase exponentially with the number of transcription factors considered , and limits EMMA to more focused problems , rather than genome-wide studies . There are a number of studies that search for tissue specific or stage specific CRMs based on a set of co-regulated genes . Some studies also include information from microarray expression data , such as LRA [31] , ClusterScan [32] , Composite Module Analyst [33] , and ModuleMiner [34] . Other methods scan only for regions where a small set of user defined transcription factors bind but do not predict novel CRMs , such as STORM & MODSTORM [35] , ModuleScanner [36] , Target Explorer [37] , and CisModScan [38] . These types of methods are not included in this assessment because we focus on genome wide novel CRM prediction methods . Several previous publications have reviewed different aspects of some of these methods . Gotea et al . [39] studied the problem on a small scale up to 10kb upstream of sets of co-expressed genes; Aerts et al . [40] performed a genome-scale target genes prediction for individual transcription factors; King et al . [27] compared methods using comparative genomics in different ways; Wang et al . [41] experimentally validated predictions based on the hypothesis that the combination of high Regulatory Potential and existence of a conserved known binding motif is a good predictor for functional CRMs; Halfon et al . [42] , Chan and Kibler [28] and Pierstorff et al . [21] , each compared the performance of several CRM prediction methods . However , their results are based on several small sets of data and the number of methods compared is limited .
The results of the ten selected methods on REDfly are plotted as a receiver operating characteristic ( ROC ) curve , where sensitivity is plotted as a function of specificity at different cut-off thresholds . Sensitivity is proportional to the true positive rate indicating how many true CRMs are found from all the annotated CRMs , ( Sensitivity = TP/P = TP/ ( TP+FN ) ) . Specificity depends on the true negative rate indicating how many true introns , exons or intergenic regions are found from the negative dataset ( Specificity = TN/N = TN/ ( TN+FP ) ) . The ten methods applied are in ten different colours . Each method has two ROC curves , one is for window size 200 bp , and another one is for window size 500 bp . The ROCs of the methods' ability to distinguish CRMs from short introns are plotted ( Figure 4A ) . All methods show a positive predictive power , except MCAST whose prediction power is close to random . The results show two clear clusters: the methods based on a single genome , and the methods based on multiple genomes . Among the single-genome methods , the best performing one is ClusterBuster . Among the multiple-genomes methods , the best performing one is MorphMS . Among all the ten methods , StubbMS and MorphMS outperform the other methods clearly . The fact that MorphMS performs better than StubbMS suggests that a probabilistic alignment strategy based on binding sites does capture the functional element information better than the conventional alignment strategy based on nucleotides , as stated in Sinha and He [18] . CisPlusFinder and MultiModule are based on multiple genome alignments and do not show any dramatic improvement over the single-genome methods . CisPlusFinder performs well while its CRM score threshold is high , but it deteriorates as the threshold is reduced . This might be due to the specific type of CRM targets of CisPlusFinder: CisPlusFinder defines a CRM as a cluster of so called perfect local ungapped sequences – multiple copies of over-represented binding sites in a single sequence . Each set of perfect local ungapped sequences is a homotypical clustering of binding sites of one transcription factor , and a cluster of these sequences refers to the CRMs containing multiple homotypical clusterings of binding sites . Thus the CRMs containing only binding sites of a single transcription factor , or a heterotypical cluster of several single binding sites , will be missed . Another factor that may affect their performance is that these two methods do not use a motif library , unlike StubbMS and MorphMS , as predicting both the transcription factor binding site and the CRM simultaneously is a more challenging task . Unexpectedly , the simple peak phastCons score window method outperforms all the more complex methods . When evolutionary conservation is used as an independent feature to distinguish the true CRMs from the intronic sequences , its performance is nearly perfect . The ROCs of the methods distinguishing CRMs from exons are plotted ( Figure 4B ) . This result shows a dramatic reversal of the curves of those methods based on multiple alignments , indicating that these methods are driven heavily by the conservation feature of the given sequences and do not have the ability to distinguish conserved regulatory elements from conserved protein-coding sequences . This also indicates that there are many false positive hits of transcription factor binding sites on exon regions as well , as a motif library of known transcription factor binding sites is not able to compensate for the high level of sequence conservation . The more a method relies on the conservation factor when predicting CRMs , the worse it performs at distinguishing CRMs from exons . That is why the peak phastCons score window method performs the worst in this case . The only exception is CisPlusFinder , which does not fall completely into the bottom right space . CisPlusFinder requires a candidate CRM sequence not only to be conserved , but also has the inter-relationships between the adjacent perfect local ungapped sequences . Only a cluster of the local ungapped sequences can be the CRM candidate . This condition reduces the likelihood of conserved exon sequences being recognized falsely as functional regulatory sequences . However , it still loses its prediction power as the score threshold goes down . On the contrary , the methods based on a single genome stay at a similar level to their results on distinguishing the CRMs from the introns , and the optimal one is still ClusterBuster . To summarize , for the ROC curves above , an Area Under ROC Curve ( AUC ) score is calculated as a representation of the prediction power of a method . Then the methods are ranked by their AUC scores according to their results of distinguishing the CRMs from the short introns ( Figure 4C ) . The top three performing methods are all multiple alignments based: phastCons , MorphMS and StubbMS . However , they all show a weak predictive power against exons . ClusterBuster ranks fourth for its predictions against short introns . Compared to the first three methods , its performance is similar against both short introns and exons . Given an unannotated genome , such a method will provide more reliable predictions . For most of the selected methods under this experimental setting , their predictions are not very sensitive to the window size 200 bp or 500 bp settings . The probabilistic modelling methods , especially the ones using multiple genomes , such as StubbMS , are less sensitive than the window clustering methods , such as CisPlusFinder . CisPlusFinder performs better when its window size is set to be 500 bp instead of 200 bp: a longer region is more prone to have multiple homotypical clusterings as CisPlusFinder targets for . The slightly preferred window size for majority of methods is 500 bp , which is similar to the average length 635 bp of predicted human and mouse CRMs of the database PReMod [49] , [50] , and the average length 760 bp of fly CRMs of the database REDfly [51] . We also obtained the prediction results of these ten methods on a medium length intron dataset ( Figure 5A ) and an intergenic dataset ( Figure 5B ) ( Figure 5C . the AUC scores of the assessed methods ) . All methods except MorphMS and MCAST , show a clear performance deterioration compared to the short intron dataset . This is not surprising considering that the medium length introns and the intergenic regions are more likely to contain actual transcription factor binding sites than the short introns , and the intergenic regions are the most contaminated among these three regions [52] , [53] . This is illustrated clearly by the performance changes of the methods relying on clusters of binding sites only , such as ClusterBuster , Stubb and MSCAN . The phastCons score window method performed much worse on these two datasets than on the short intron dataset . The gap between the predictions of the window size 200 bp setting and the prediction of the window size 500 bp setting is significantly larger than their difference on the short intron and the exon datasets . The result of the 500 bp window size is superior to 200 bp . It is known that introns can mediate gene expression in various ways [54] . The intron length is connected to alternative splicing events ( http://www . sdbonline . org/fly/aimain/6rna-ooc . htm ) and functional introns tend to be the larger ones [55] . The conserved intergenic regions are also known to play regulatory roles [56] . Therefore it is very likely that there are conserved functional regions existing in the medium length introns and intergenic regions , and some of them can span around 200 bp . CRMs can be distinguished from these functional regions by a larger window size setting of 500 bp . Apart from above differences , these results agree with those obtained from the short intron dataset in terms of ranking among the methods and similar performance between the two window size settings for each method . Based on the prediction score of REDfly CRMs given by each method , we normalized the scores of each method to the same scale between 0 to 1 , by dividing each score by the maximum possible of that method . We then calculated the correlation coefficients between all pairs of methods ( Figure 6A ) . For each method , the results with 200 and 500 bp window sizes correlate closely . Particularly for MorphMS , a very high correlation exists between the two predictions of these window sizes . This further confirms the previous results that these methods are not very sensitive to the window size parameter setting under this experimental design . One exception is CisPlusFinder , which shows a stronger prediction power with 500 bp window size compared to 200 bp . The other exception is CisModule , where the 200 and 500 bp window size results form two separate clusters . This might be explained by the fact that CisModule follows a non-deterministic algorithm and each run returns a slightly different result . The high correlation coefficients show the agreement between these representative methods . Those methods with the same underlying CRM representations and which require the same prior information are clustered together as expected ( e . g . MorphMS and StubbMS , CisPlusFinder and MSCAN , and ClusterBuster and Stubb ) . Unexpectedly , CisPlusFinder performs more similar to the multiple alignments probabilistic modelling methods StubbMS and MorphMS when its window size is set to be 500 bp . These three methods from two different families all have strong predictive power with significant agreement , despite their different underlying mechanisms . Another exception is MultiModule , which is clustered into the single genome probabilistic modelling family together with ClusterBuster and Stubb . MultiModule itself is a generative probabilistic model , similar to a hidden Markov model . However , the information from the double alignment does not improve the performance of MultiModule over the methods using a single genome only . Pairwise complementarity of methods is checked by summing the normalized scores given by each pair of methods for both their predictions on the CRMs and their predictions on the short intron negative dataset . The increase or decrease of the AUC scores of the new pairs over the maximum of the individual methods is shown ( Figure 6B ) . Most methods deteriorate when the predictions of two different window size settings are summed together . This is clearly the case for MorphMS and the peak phastCons score window method . At the same time several other methods show an opposite effect , such as StubbMS for which the summed result brings its prediction power from AUC score 0 . 893 and 0 . 888 to 0 . 996 ( Figure 6C ) . The new result is equivalent to the prediction power of phastCons score and is nearly perfect . Amongst these methods , the window clustering family methods CisPlusFinder and MSCAN , especially with the window size 200 bp setting , are highly complementary to probabilistic modelling family methods StubbMS , CisModule and MultiModule . The performances of these pairs of methods are better than any individual method . One possible reason might be the different approaches of these methods to defining the candidate binding site profiles . CisPlusFinder is not constrained to the prior knowledge of binding site profiles and therefore has the potential to search for unknown transcription factor binding sites . Another reason might be that they focus on different length CRMs: the probabilistic modelling family methods tend to find short CRMs , while CisPlusFinder tend to find long CRMs . For example , the first quartile and the third quartile of the lengths of the predicted clusters by ClusterBuster with window size 200 bp setting are 149 bp and 790 bp accordingly; in the results of CisPlusFinder with window size 200 bp setting , there are only two predicted CRMs shorter than 200 bp , and the first quartile and the third quartile of the lengths of the predicted clusters are 677–1643 bp accordingly . To understand what properties of a CRM make it distinctive , and what features of a negative sequence cause false positive predictions , we checked the correlation coefficients between sequence features of the CRMs , the short introns and the exons , and the scores given by each method . The sequence features considered include its average conservation degree measured by phastCons score and its length ( Figure 7 ) . The predictions of StubbMS and MorphMS are heavily affected by the average conservation degree of a sequence . This confirms that the high average sequence conservation is the key feature these two methods rely on , and it contributes both the true positives and false positives . The peak phastCons score window method , searching for continuous windows over a threshold , does not rely on this feature of CRMs for its prediction . The phastCons score window method predicts CRMs better than MorphMS and StubbMS , showing that searching for peak conservation regions on a sequence can capture more regulatory elements than counting the average sequence conservation . For the correlations between the prediction and the sequence length , which is equivalent to the CRM length in this experimental design , nine out of ten methods show a correlation to a certain degree . Especially MultiModule , Stubb and ClusterBuster , the members of the probabilistic modelling family , have correlation coefficients over 0 . 5 . Among all , MCAST is the method driven by the sequence length most . Basically , a long sequence brings a high scored prediction . This bias causes false positives of all the methods except the peak phastCons score window method , which does not rely on this feature of CRMs for its prediction . We sorted the CRMs by the summed scores of all ten methods and excluded the CRMs having a 0 score by any method , and then checked the properties of the CRMs commonly found by the ten selected methods . These CRMs tend to be long sequences , but not always very conserved . The correlation coefficient between the predictions and the sequence lengths is high , while the same figure for the average sequence conservation is low . For the same reason , the false positive predictions from the short intron and the exon datasets also tend to be long sequences , and the correlations between the prediction and the sequence length are high . The peak phastCons score window method is the one least biased from these sequence features . In summary , for most methods , long length and general conservation of a short intron or exon sequence contribute the most to both true and false positives . A continuous peak conserved window is a more distinctive and unique feature of a CRM , and can be used to identify the real CRMs as the success shown by the peak phastCons score window method . Among all the CRM sequences , 19 sequences are annotated with known transcription factors , and their transcription factor binding sites are experimentally validated and annotated by the Drosophila DNase I footprint database FlyReg [57] . This provides us a chance to further check how CRM properties affect the prediction of each method , based on the known information so far . We checked for how these methods are prone to the abundance of transcription factor binding sites , the number of transcription factors , and the composition of homotypical clustering , by calculating the correlation between the CRM properties and the prediction scores on the 19 annotated sequences ( Figure 8 ) . Different CRM representations and search strategies rely on different CRM properties . The predictions of the ClusterBuster , CisPlusFinder with window size 200 bp setting and MSCAN are significantly correlated with the total number of transcription factor binding sites of a CRM . CisPlusFinder also shows a strong correlation with the number of transcription factors a CRM contains . Indeed , it predicts the CRMs with multiple transcription factors only . The CRMs containing large homotypical clustering of multiple transcription factor binding sites are more likely to be found by ClusterBuster and MSCAN . For MultiModule , the density of transcription factor binding site on a sequence is critical for its prediction . Some types of CRMs are easier to be predicted and some types of CRMs do not have very distinctive features ( Table S1 ) . The CRMs with multiple transcription factor binding sites of known transcription factors are easier to be predicted , such as CRM Ubx_basal_promoter containing 20 transcription factor binding sites of seven known transcription factors including Ubx and zen . Most methods score it high , especially ClusterBuster and CisPlusFinder with window size 200 bp setting . On the contrary , short CRMs with a few transcription factor binding sites are easily missed by most prediction methods . For example , for the 227 bp long ninaE_distal_enhancer with only two gl binding sites , ClusterBuster with window size 200 bp setting scores it very low because of there is not a profile of the gl transcription factor binding site supplied . CisPlusFinder scores it 0 for another reason: this CRM is composed of only one homotypical clustering . For the short CRMs with few transcription factor binding sites , the peak window phastCons score method will not miss it . For this particular CRM , phastCons with window size 200 bp setting scores it high as 0 . 991 . The peak phastCons score method does not always pick up the real CRMs . There are cases where the probabilistic modelling methods predict correctly while the peak phastCons score method does not . For example , CRM Dpp_BS1 . 0 contains five transcription factor binding sites of transcription factor en within a 246 bp distance . The peak phastCons score window method scores it relatively low , while probabilistic modelling methods such MorphMS score this sequence high . The reason leading to this phenomenon could be the binding sites on this sequence are conserved but the sequence between them are not . Therefore there is not a continuous peak conserved window as the peak phastCons score method requires . MorphMS is able to detect such shifted conservation by aligning sequence by the location of transcription factor binding sites . Unexpectedly , there are also cases where CisPlusFinder misses out genuine CRMs with multiple homotypical clusterings: such as Ance_race_533 , a 533 bp long CRM annotated with nine transcription factor binding sites of three transcription factors including Mad and zen . Both CisPlusFinder with 200 bp window setting and with 500 bp window setting score this sequence as 0 . The perfect local ungapped sequences defined by CisPlusFinder cannot always represent real binding sites accurately . The above success of using pure conservation scores to predict CRMs suggests that searching for appropriately sized conserved blocks is sufficient to distinguish true CRMs from the REDfly database from short introns and exons . This may not be surprising considering the Drosophila genome is relatively small and compact , and its regulatory regions are closely packed together [58] . REDfly is principally composed of developmental enhancers and these elements are known to be generally very conserved [59] , [60] . However , the dramatic contrast of the performance of these multiple alignment based methods depending on whether introns or exons are used as representative negative sequences leads us to question whether the level of conservation seen in the CRMs collected by REDfly is representative of typical CRMs . To further investigate this possibility and to check if these methods are sensitive to the composition and structure of the genome , we applied the optimal methods among the prediction on REDfly: ClusterBuster , MorphMS and the peak phastCons score ( http://hgdownload . cse . ucsc . edu/goldenPath/hg18/phastCons17way/ ) window method , plus the peak Regulatory Potential score ( http://hgdownload . cse . ucsc . edu/goldenPath/hg18/regPotential7X/ ) window method ( see Materials and Methods ) and the prediction results of EEL ( http://www . cs . helsinki . fi/u/kpalin/EEL/ ) , to human ENCODE regions . The human genome is more diverse on its conservation degree of regulatory elements . Specifically , 30 out of 44 ENCODE regions were picked by the ENCODE consortium according to their non-exonic conservation levels ( 1 . 1–6 . 2% , 6 . 3–10 . 2% , 10 . 7–18 . 6% . ) and gene densities ( 0–1 . 7% , 2 . 0–3 . 6% , 4 . 4%–10 . 6% ) [61] ( http://genome . ucsc . edu/ENCODE/regions . html ) . We used these 30 ENCODE regions to make sure that the sequences are diverse in their converstaion degrees and thus eliminate the possibility of any bias caused by conservation . Firstly we compared the conservation degree of transcription factor binding sites , CRMs , and noncoding regions of both Drosophila genome and human ENCODE regions ( Figure 9 ) . For human ENCODE regions , we used ENCODE regulome DNase I hypersensitive sites of human lymphoblastoid cells GM06990 [62] ( http://hgdownload . cse . ucsc . edu/goldenPath/hg18/encode/database/encodeRegulomeDnaseGM06990Sites . txt . gz ) as the potential CRMs which mark the chromatin regions having high accessibility to transcription factors . We expect the CRMs are less conserved than the transcription factor binding sites because CRMs contain less constrained sequences between transcription factor binding sites . The probability density shows that , for Drosophila , the REDfly CRMs are more conserved than the transcription factor binding sites . For human ENCODE regions , the transcription factor binding sites are more conserved than the DNaseI hypersensitive sites . This confirms that the REDfly CRMs are more conserved than expected . Comparing between the two genomes , the entire Drosophila genome and their regulatory regions are more conserved than their equivalents on the human ENCODE regions . Next , we applied the five selected methods on the ENCODE regions , and their performances were evaluated by their overlaps with the DNaseI hypersensitive sites . If a prediction overlapped – even partially – with any DNaseI hypersensitive site , it was counted as a true positive . A prediction not overlapping with any DNaseI hypersensitive site was counted as a false positive . The missed DNaseI hypersensitive sites were counted as false negatives . Because these methods need to scan large ENCODE regions therefore it is not sensible to define a fixed-sized true negatives . For this reason , instead of specificity , positive prediction value was calculated to show the methods performance . The results of these methods were plotted in a pseudo ROC plot , where sensitivity is plotted against positive prediction value ( PPV ) : sensitivity = TP/ ( TP+FN ) , indicating how many true CRMs are found among all the DNaseI hypersensitive sites , and PPV = TP/ ( TP+FP ) , indicating the percentage of true CRMs among all the predictions ( Figure 10 ) . Among all the methods , the peak Regulatory Potential score window method significantly outperforms the rest of the methods . This suggests that the information learnt from the known regulatory regions is very helpful indeed . Unexpectedly , EEL does not pick up any positive signals and is at the bottom of the chart . This might be due to the public available prediction results of EEL were produced with a high cut-off threshold , while the other methods' cut-off thresholds were deliberately set to be their lowest in this assessment to allow the maximum number of predictions . Overall , their performance ranks them from top to bottom in this order: the peak Regulatory Potential score window method , MorphMS , ClusterBuster , the peak phastCons score window method , and EEL . This result shows a different prediction power of some methods from their previous prediction performances of the REDfly CRMs: the peak conservation phastCons score approach does not outperform probabilistic modelling methods in this case . For the window size setting ( Figure S1 ) , both ClusterBuster and MorphMS predictions with 500 bp window setting discovered slightly more CRMs than their predictions with 200 bp window setting , with a price paid by vastly increased computational time for MorphMS . We also increased the window size of the peak phastCons score window method and the peak Regulatory Potential score window method from 100 bp to 200 bp , 500 bp , 1000 bp and 1500 bp . The increase of the window size universally increased the performance of these two methods . Perhaps understandably , the optimal window size setting of these methods tuned for the human genome tend to be larger than the ones for the Drosophila genome . Upon summarizing the above results , it is clear that the application of the prediction methods on the Drosophila genome and the human genome need to be treated differently . Not only are the composition [63] and the structure [64] of the Drosophila and human genomes different , but the evolutionary distance between the given alignment: Drosophila melanogaster and Drosophila pseudoobscura , human and mouse , are different too . The nucleotide conservation levels between the Drosophila melanogaster genome and the Drosophila pseudoobscura genome are ∼70% for coding sequences , ∼40% for introns [65] . The corresponding figures between the human and the mouse genomes are: ∼85% for coding regions , ∼35% for introns [66] . These might all contribute to the different performances of the prediction methods .
The two most frequently used types of genome information resources: conservation and transcription factor binding site profiles , and the four families of search strategies , are applied in numerous ways . Any subtle change in the combination or the order may yield different results . Therefore the existing methods can show a great variety of results given the same data . Although there is not a universal optimal method suitable for all situations , several key strategies applied in the existing methods do show their values on improving predictions . The advantage of MorphMS over StubbMS for predicting REDfly CRMs supports the view that aligning multiple genomes by locations of conserved transcription factor binding sites can perform better than conventional alignment according to the nucleotides . CisPlusFinder can complement several methods . This brings our attention to neighbourhood relationships between homotypical clusters of sites for multiple factors . The success of the peak Regulatory Potential score window method shows the importance of the information learnt from the known regulatory elements , particularly , the novel strategy of considering the alignment pattern: the first order dependent relationship between the conserved columns within a transcription factor binding site . However , there remain some clear problems with CRM prediction . Firstly is the fundamental problem of modelling functional CRMs: the majority of existing CRM prediction methods target regions rich in clustered and conserved transcription factor binding sites , and while this does work to a degree , it remains a relatively poor proxy for identifying functional regulatory elements ( Figure 11 ) . The fact is that the distance and conservation features of a sequence are not sufficient to accurately deduce its function . In addition , not all CRMs are tightly packed or highly conserved . At the same time , a fragment of a CRM , or overlapping regions shared by more than one CRMs , could be predicted as one complete CRM . Clearly , the current CRM prediction methods are only a first step towards accurately predicting true CRMs . Secondly , the general CRM properties are not universally applicable . There are also exceptional cases where some real regulatory functional sites are not more conserved than the background sequences [67] . At the same time , not all the clustered conserved elements are cis-regulatory elements - they can be conserved non-functional noncoding regions [68] , or other conserved signals which have other functions other than being an enhancer , e . g . microRNA . In addition , some transcription factors , such as E2F1 , do not require a canonical binding site [69]; while for some other cases , for a same consensus , several transcription factors can compete each other on binding on it . Further more , the interactions between DNA and transcription factor , and the interactions between factors and factors form 3D complexes; this makes identifying the members indirectly involved even more difficult . Obviously , the information of binding affinity , the distance and the conservation , are far from being enough to identify a functional module . Thirdly , the CRM prediction methods development and evaluation lacks genome-scale standardization and benchmarking . Most development and comparison on the CRM prediction methods were based on either a small set of genes or REDfly . King et al . [27] used HBB gene complex; Wang et al . [41] used the mammalian genes expressed in read blood cells; or Aerts et al . [40] and Sinha and He [18] used REDfly , which is the only experimentally confirmed genome wide CRM database available . A small set of co-expressed genes tends to have a limited number of similar CRMs made of a few transcription factors , and we showed that the CRMs in the REDfly database are very conserved and therefore might not able to represent the general CRMs on other genomes . A method tuned on the maximum performance on these sequences can be biased toward the extreme properties of the data itself and therefore is not suitable to be universally applied to another set of sequences or another genome . The human ENCODE regions have a wider range of sequence conservation compared to the Drosophila , and the DNase I hypersensitive sites are not biased toward developmental enhancers . These regions have been heavily studied in the past few years so there are plenty of annotations and there are going to be more . Our results show how the performances of some methods change depending on the composition and structure of genomes . This suggests that a method developed for a general purpose , regardless the genome , needs to be tested on multiple genomes to show its general applicability . Certainly this assessment and analysis are also only based on the available annotations , such as the cell type dependant DNaseI hypersensitive sites [70] we used as potential CRMs , which mark the chromatin regions accessible to all types of proteins but not only limited to transcription factors . There is no direct equivalent CRM database to REDfly in mammals . In addition , the parameter settings of the methods are their defaults , and might not be the optimal settings for some methods to show their peak performance . The major difficulty of modelling CRMs comes from the fact that the majority of direct and indirect interaction relationships between transcription factors remain unknown . These subtle but critical transcriptional regulatory codes might only be decoded on a smaller scale: such as using expression microarrays or RNA-seq to identify the co-regulated genes then extracting the common patterns from the upstream of these co-regulated genes , or identifying the interaction relationship within a module through a gene regulatory network analysis . Even with the interaction relationships known , the dynamic information at different conditions are needed to really understand the regulation machinery . The transcriptional logic code is sensitive to conditions . Depending on the context , cis-regulatory elements can be active for function or not , and can perform different roles too: either as transcription factor binding sites , or as facilitated steps for CRM scanning along the sequence or looping and tethering intervening DNA [71] . So far , among all the methods studied in this assessment , only EEL takes DNA structure of a sequence into consideration . Recently , other types of information have been used to assist the CRM prediction , such as the DNA double helix structure profile [72] , chromatin structure and histone modification [73] , and chromatin immunoprecipitation followed by microarray analysis ( ChIP-chip ) [74] or chromatin immunoprecipitation followed by high-throughput sequencing ( ChIP-seq ) [75] . In anticipation of a large-scale analysis , one of the most intriguing projects , ENCODE Pilot Project , is scaled up to a production phase to annotate the entire human genome . This ongoing project will systematically and comprehensively identify transcription factor binding sites , map the histone modifications , and mark the methylation status of CpG-rich regions ( http://www . genome . gov/10005107 ) . In addition , the modENCODE project will identify these regulatory elements on the Drosophila and worm genomes [76] . During this process , the existing technologies including DNaseI hypersensitivity assays and chromatin immunoprecipitation followed by high-throughput sequencing are applied , whereas more advanced high-throughput computational and experimental methods are in great demand . To answer this request , novel analysis strategies and prediction methods that integrate sequence information and chromatin signatures could be a major step forwards . For instance , Won et al . [77] integrated strong Histone H3 Lys 4 methylations ( H3K4me1/2/3 ) signals together with sequence affinity for transcription factor binding sites into one hidden Markov model to characterize regulatory regions on mouse embryonic stem cells . We believe with the assistance of new technologies , novel analysis strategies , and more complete functional annotations , next generation CRM prediction methods will aim to recreate a dynamic picture of transcription regulation interactions in three-dimensional space . Beyond identifying CRM locations , the future focus will also turn to measuring and predicting spatio-temporal cis-regulatory activity [74] , [78] , [79] . For the Drosophila genome , based on the results of the REDfly database , which possibly promotes bias toward methods relying on sequence conservation , MorphMS produces the most successful and stable predictions when dealing with the non-exonic regions . The peak phastCons score window method with 500 bp setting can be a good choice too but users may need to double check to confirm the predicted regions are indeed functional as CRMs . Other methods can be used here to provide this information by checking which transcription factors bind there . ClusterBuster is the best choice for single genome , or MorphMS for multiple genomes . However , users need to be aware that the predefined motif library limits the performance of both ClusterBuster and MorphMS . They cannot predict successfully on a region with unknown transcription factor binding sites . Even for the known transcription factor binding sites , there might be a disagreement between the transcription factor binding site profile provided and the genuine transcription factor binding sites on the sequence . For those regions with unknown transcription factor binding sites , CisPlusFinder appears to offer a solution , by searching for multiple conserved , locally overrepresented sequences as potential binding sites . Therefore there will not be any constraints due to lack of prior knowledge of these binding sites . One condition for CisPlusFinder to locate a potential CRM is the existence of multiple homotypical clusters . This causes CisPlusFinder to miss all CRMs interacting with only one transcription factor , or a single binding site of every transcription factor it contains . Another issue is that a real transcription factor binding site signal might not be abundant in one particular CRM; therefore the perfect local ungapped sequences might not be able to represent all the transcription factor binding sites . For this reason CisPlusFinder can be used combined with ClusterBuster or MorphMS to discover every CRM candidate . These two different families methods are not only complementary to each other on searching for the unknown transcription factor binding sites , but also on searching for different lengths CRMs: the probabilistic modelling family methods tend to find short CRMs , while CisPlusFinder tend to find long CRMs . For the mammalian genome , the peak Regulatory Potential score window method is the best way to locate CRM regions . ClusterBuster and MorphMS can be used in addition to identify which transcription factors bind there .
REDfly version 2 . 0 is a curated collection of known Drosophila transcriptional cis-regulatory modules and transcription factor binding sites . It contains all experimentally verified Drosophila regulatory elements along with their DNA sequences , their associated genes , and the expression patterns they direct . There are in total 665 CRMs and 941 transcription factor binding sites annotation . The first and the third quartile of the length of these CRMs are 907 bp and 2967 bp , and the median is 1520 bp . Because the boundaries of these CRMs are not certain , each CRM region was extracted including its core sequence plus 200 bp flanking regions on both upstream and downstream . Multiple alignments of 12 Drosophila species were extracted for each REDfly CRM region . These raw multiple alignments for comparative analysis were produced by Colin Dewey in Lior Pachter's group at UC Berkeley by their multiple-sequence aligner – MAVID [47] , based on the first freeze of all the comparative assemblies of 12 Drosophila genomes in December 2005 and January 2006 [80] . To make the dataset compatible with all the selected methods requirements , among the 665 CRM sequences , we chose 244 non-redundant CRMs satisfying the following two requirements: The four negative sequence datasets: short introns , exons , medium lengh introns and intergenic regions , were extracted from the Drosophila Melanogaster genome sequences , where no regulatory elements are supposed to exist . These negative sequences would differ from CRM sequences in their compositional contents , conservation rates , GC content and other features . The intron dataset was assembled from introns between 12 bp to 81 bp in length . The exon dataset was assebmled from randomly selected exons . For each short intron or exon sequence , 6 bp was removed from its 5′ end and 3′ end to avoid any consensus splice donor sites ( GTA/GAGT for intron and G/A for exon ) and any consensus splice acceptor sites ( C/TAG for intron and C/AAG for exon ) [81] . The sequences of each type of source were then randomly selected and randomly extracted , then were concatenated into 244 sequences with the same lengths as the 244 CRMs . The medium length intron dataset was assembled from introns between 300 bp and 1 kb in length . For each sequence , 150 bp was removed from its both 5′ and 3′ ends to minimize the risk of contamination with any splice regulatory sequences . The integenic dataset was assembled from those integenic regions between 2 kb and 100 kb in length . For each intergenic sequence , 1kb was removed from both its 5′ end and its 3′ end to avoid any promoter sites and post-transcriptional modification sites . For those methods based on multiple genomes , pairwise alignment of Drosophila Melanogaster on Drosophila Pseudoobscura of both positive and negative datasets were extracted from MAVID . The alignments of human and mouse were downloaded from the UCSC genome browser . It is from the December 2007 ENCODE Multi-Species Sequence Analysis ( MSA ) sequence freeze , which consists of orthologous sequences in mouse to the human ENCODE regions The peak phastCons score window method and the peak Regulatory Potential score window method follow the window size settings and the threshold cut-off settings as described in [27] . For phastCons score , a 100 bp window having an average score over 0 . 13 is counted as a positive; continuous overlapped positive windows are counted as a regulatory region . Same process is applied to Regulatory Potential score , with the cut-off threshold set to be 0 . For the Drosophila genome , the conservation degrees were checked for the ChIP-on-chip verified transcription factor binding sites of four transcription factors ( http://furlonglab . embl . de/data/download ) : Mef2 [82] , Twist [82] , Bagpipe and Biniou [83]; the REDfly CRMs; and the entire Drosophila Melanogaster non-coding regions . For the ENCODE regions , the conservation degree were checked for the ENCODE Yale/UC-Davis/Harvard TFBSs by ChIP-seq of eight transcription factors ( http://genome . cse . ucsc . edu/cgi-bin/hgTrackUi ? db=hg18&g=wgEncodeYaleChIPseq ) : c-Fos , c-Jun , c-Myc , GATA-1 , JunD , Max , NF-E2 and ZNF263 [84]; the ENCODE Regulome DNase I hypersensitive sites and the entire ENCODE non-coding regions .
|
Transcriptional regulation involves multiple transcription factors binding to DNA sequences . A limited repertoire of transcription factors performs this complex regulatory step through various spatial and temporal interactions between themselves and their binding sites . These transcription factor binding interactions are clustered as distinct modules: cis-regulatory modules ( CRMs ) . Computational methods attempting to identify instances of CRMs in the genome face a challenging problem because a majority of these interactions between transcription factors remain unknown . To investigate the reliability and accuracy of these methods , we chose twelve representative methods and applied them to predict CRMs on both the fly and human genomes . Our results show that the optimal choice of method varies depending on species and composition of the sequences in question . Different CRM representations and search strategies rely on different CRM properties , and different methods can complement one another . We provide a guide for users and key considerations for developers . We also expect that , along with new technology generating new types of genomic data , future CRM prediction methods will be able to reveal transcription binding interactions in three-dimensional space .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"computational",
"biology/sequence",
"motif",
"analysis",
"computational",
"biology/transcriptional",
"regulation",
"computational",
"biology/comparative",
"sequence",
"analysis",
"computational",
"biology/evolutionary",
"modeling",
"evolutionary",
"biology/bioinformatics"
] |
2010
|
Assessing Computational Methods of Cis-Regulatory Module Prediction
|
Microbial diversity is typically characterized by clustering ribosomal RNA ( SSU-rRNA ) sequences into operational taxonomic units ( OTUs ) . Targeted sequencing of environmental SSU-rRNA markers via PCR may fail to detect OTUs due to biases in priming and amplification . Analysis of shotgun sequenced environmental DNA , known as metagenomics , avoids amplification bias but generates fragmentary , non-overlapping sequence reads that cannot be clustered by existing OTU-finding methods . To circumvent these limitations , we developed PhylOTU , a computational workflow that identifies OTUs from metagenomic SSU-rRNA sequence data through the use of phylogenetic principles and probabilistic sequence profiles . Using simulated metagenomic data , we quantified the accuracy with which PhylOTU clusters reads into OTUs . Comparisons of PCR and shotgun sequenced SSU-rRNA markers derived from the global open ocean revealed that while PCR libraries identify more OTUs per sequenced residue , metagenomic libraries recover a greater taxonomic diversity of OTUs . In addition , we discover novel species , genera and families in the metagenomic libraries , including OTUs from phyla missed by analysis of PCR sequences . Taken together , these results suggest that PhylOTU enables characterization of part of the biosphere currently hidden from PCR-based surveys of diversity ?
A central goal of ecology and evolution is to understand the forces that shape biodiversity - the variety of life on Earth . It is becoming increasingly clear that global biodiversity is mostly microbial . It is estimated that there are millions of microbial species on the planet , relatively few of which have been isolated in culture [1]–[2] . Despite the recognized importance of microorganisms , we still know little about the magnitude and variability of microbial biodiversity in natural environments relative to what is known about plants and animals . This is a major knowledge gap , given that microbes are critical components of our planet , responsible for key ecosystems services including the production of agriculturally critical small molecules , the degradation of environmental contaminants , and the regulation of human host phenotypes . Biodiversity science has traditionally focused on comparing species richness across space , time and environments . Out of necessity , microbial diversity studies usually examine the richness ( i . e . number ) of operational taxonomic units ( OTUs ) , where OTUs are sequence similarity based surrogates for microbial taxa , which can be difficult to define . In addition to richness , OTUs have been used to characterize the abundance , range , and distribution of microbes , thereby improving our understanding of both natural ecosystems and human health [3]–[6] . OTUs are commonly identified by aligning sequences of the small subunit of ribosomal RNA ( SSU-rRNA ) from one or more samples and identifying groups of related sequences using a hierarchical clustering algorithm . This clustering is based upon a measure of distance between all pairs of sequences , which is typically defined using some variant of the percent sequence identify ( PID ) ( e . g . [3] , [7]–[8] ) . For example , researchers traditionally cluster sequences that are no more than 3% diverged into the same OTU . This designation has been proposed as being roughly equivalent to a species-level classification [9] , though evidence suggests that it may result in an underestimate of the true number of species [10] . The SSU-rRNA sequences for OTU identification are traditionally amplified from a sample via polymerase chain reaction ( PCR ) using universal primers . Each PCR product is then individually sequenced . One of the biggest drawbacks of this targeted sequencing approach is that it leverages PCR , which has been shown to exhibit sequence-based biases at the level of priming and extension [11]–[13] . In addition , the so-called ‘universal’ PCR primers used in such assays will fail to amplify sequences sufficiently diverged from those used to design the primers . The result is that some taxa may be disproportionately amplified or even missed [14] . Metagenomic approaches eliminate this bias by sequencing randomly-sheared fragments ( i . e . , shotgun sequencing ) of environmental DNA , and , despite having their own sources of bias [15] , may therefore provide a potentially more accurate characterization of microbial diversity . For example , the analysis of metagenomic data from a relatively simple microbial community revealed the presence of low-abundance acidophilic Archaea overlooked by PCR-based surveys of diversity [16] . Because of the fragmentary nature of shotgun sequencing , metagenomic reads frequently exhibit minimal , if any , sequence overlap . PID-based evaluations using metagenomic data are thus restricted to the subset of reads that mutually overlap and can therefore be aligned to one another ( e . g . , [17] and [18] ) . Alternative approaches have been adopted to describe microbial diversity from non-overlapping metagenomic sequences , including the binning of reads into a reference taxon by comparing each read against reference sequence databases ( e . g . , [17] , [19] and [20] ) and using de novo sequence assemblers to build SSU-rRNA contigs ( e . g . , [21] ) . While these approaches have substantially advanced the field of microbial biodiversity , they exhibit significant limitations . The former is limited by the diversity encoded in sequence databases , most of which was obtained via targeted sequencing studies . The latter is restricted to the subset of high-confidence assemblies , which can be difficult to produce in many environments given that contig assembly may result in chimeric SSU-rRNA sequences from complex communities [22] ) . Despite the rapidly growing metagenomic data in microbial ecology and human microbiome studies , no method currently provides a means of characterizing microbial diversity directly from non-overlapping metagenomic data . There is a great need for new approaches that identify OTUs using metagenomic data . We present PhylOTU , the first method that enables automated identification of microbial OTUs directly from non-overlapping metagenomic sequence reads . PhylOTU leverages a phylogenetic tree of metagenomic SSU-rRNA reads , constructed using probabilistic sequence profiles built from full-length SSU-rRNA sequences from completed genomes , to identify and characterize phylogenetic distances between SSU-rRNA reads in metagenomic data sets . This phylogenetic distance ( PD ) , rather than PID , is then used to cluster reads into OTUs in a fashion similar to that utilized for targeted sequencing data . Because the enormous volume of sequence in most metagenomic libraries presents substantial challenges in the form of sequence-alignment quality and the rate of computational through-put , we developed and implemented within PhylOTU a series of data quality control filters and efficient data structures . We also developed an error rate metric for the analysis of clustered data and used simulated sequences to quantify the accuracy of PhylOTU . These investigations enabled us to derive corrections for biases in phylogenetic methods , producing a tool with similar accuracy to existing PID-based methods . We used PhylOTU to describe microbial diversity in the global open ocean by processing the 10 , 133 , 846 shotgun reads in the Global Ocean Survey sequence library [21] . In addition , we compared the OTUs identified by PCR-generated sequences to those identified by shotgun sequences from the same samples . We find that analysis of shotgun sequences reveals a novel part of the biosphere missed by analysis of PCR-generated sequences . PhylOTU is freely available for download at github ( https://github . com/sharpton/PhylOTU ) and BioTorrents ( http://www . biotorrents . net ) [23] .
Traditionally , OTUs are identified from a PCR-generated targeted sequence library by aligning all pairs of sequences , calculating each pair's PID-based distance , and using this distance to group sequences using agglomerative hierarchical clustering . Due to the fragmentary nature of shotgun metagenomic reads , this traditional approach is limited to the subset of overlapping sequences; non-overlapping reads cannot be directly aligned to one another . Even when reads can be aligned ( e . g . , to full-length reference sequences ) , one still cannot calculate PID for sequences that do not overlap . To overcome these limitations , we designed PhylOTU , which uses a probabilistic sequence profile to align reads and a phylogenetic tree to infer their similarity . The general strategy PhylOTU employs is to leverage full-length reference sequences to construct a probabilistic sequence profile of SSU-rRNA . The profile is used to align metagenomic reads and reference sequences , and this alignment is in turn used to compute the phylogenetic distance between every pair of reads for input into the clustering algorithm . A general workflow schematic of our method is illustrated in Figure 1 . First , probabilistic profiles that encode the evolutionary diversity and secondary structure of the SSU-rRNA sequence from Bacteria and Archaea [24] are constructed via high-quality reference alignments of full-length SSU-rRNA sequence [25] . These profiles are pre-computed for use in different metagenomic analyses . For a given metagenomic data set , SSU-rRNA homologous reads are identified from the shotgun sequencing data via a BLAST search of every metagenomic read against the small but phylogenetically diverse SSU-rRNA STAP databases [26] . This relatively fast search allows one to accurately differentiate SSU-rRNA homologs of Archaea from those of Bacteria , which in turn accelerates and improves downstream alignment and phylogenetic analysis . Multiple sequence alignments of metagenomic reads are created by aligning each SSU-rRNA read to the appropriate Bacterial or Archaeal SSU-rRNA profile , using profile alignment methods [24] . This read alignment is then mapped onto the reference alignment used to build the profile , resulting in a multiple sequence alignment of full-length reference sequences and metagenomic reads . The final step of the alignment process is a quality control filter that 1 ) ensures that only homologous SSU-rRNA sequences from the appropriate phylogenetic domain are included in the final alignment , and 2 ) masks highly gapped alignment columns ( see Text S1 ) . We use this high quality alignment of metagenomic reads and references sequences to construct a fully-resolved , phylogenetic tree and hence determine the evolutionary relationships between the reads . Reference sequences are included in this stage of the analysis to guide the phylogenetic assignment of the relatively short metagenomic reads . While the software can be easily extended to incorporate a number of different phylogenetic tools capable of analyzing metagenomic data ( e . g . , RAxML [27] , pplacer [28] , etc . ) , PhylOTU currently employs FastTree as a default method due to its relatively high speed-to-performance ratio and its ability to construct accurate trees in the presence of highly-gapped data [29] . After construction of the phylogeny , lineages representing reference sequences are pruned from the tree . The resulting phylogeny of metagenomic reads is then used to compute a PD distance matrix in which the distance between a pair of reads is defined as the total tree path distance ( i . e . , branch length ) separating the two reads [30] . This tree-based distance matrix is subsequently used to hierarchically cluster metagenomic reads via MOTHUR into OTUs in a fashion similar to traditional PID-based analysis [31] . As with PID clustering , the hierarchical algorithm can be tuned to produce finer or courser clusters , corresponding to different taxonomic levels , by adjusting the clustering threshold and linkage method . To evaluate the performance of PhylOTU , we employed statistical comparisons of distance matrices and clustering results for a variety of data sets . These investigations aimed 1 ) to compare PD versus PID clustering , 2 ) to explore overlap between PhylOTU clusters and recognized taxonomic designations , and 3 ) to quantify the accuracy of PhylOTU clusters from shotgun reads relative to those obtained from full-length sequences . We sought to identify how PD-based clustering compares to commonly employed PID-based clustering methods by applying the two methods to the same set of sequences . Both PID-based clustering and PhylOTU may be used to identify OTUs from overlapping sequences . Therefore we applied both methods to a dataset of 508 full-length bacterial SSU-rRNA sequences ( reference sequences; see above ) obtained from the Ribosomal Database Project ( RDP ) [25] . Recent work has demonstrated that PID is more accurately calculated from pairwise alignments than multiple sequence alignments [32]–[33] , so we used ESPRIT , which implements pairwise alignments , to obtain a PID distance matrix for the reference sequences [32] . We used PhylOTU to compute a PD distance matrix for the same data . Then , we used MOTHUR to hierarchically cluster sequences into OTUs based on both PID and PD . For each of the two distance matrices , we employed a range of clustering thresholds and three different definitions of linkage in the hierarchical clustering algorithm: nearest-neighbor , average , and furthest-neighbor . To statistically evaluate the similarity of cluster composition between of each pair of clustering results , we used two summary statistics that together capture the frequency with which sequences are co-clustered in both analyses: true conjunction rate ( i . e . , the proportion of pairs of sequences derived from the same cluster in the first analysis that also are clustered together in the second analysis ) and true disjunction rate ( i . e . , the proportion of pairs of sequences derived from different clusters in the first analysis that also are not clustered together in the second analysis ) ( see Methods and Figure S1 ) . PhylOTU exhibits high true conjunction and true disjunction rates at commonly employed PID thresholds ( e . g . , 0 . 03 , 0 . 06 ) , demonstrating that PD-based clustering accurately recapitulated PID-based clustering at the same threshold ( Figure S2 ) . On the other hand , when applying the same clustering threshold to both distance matrices , PID-based clustering produces a higher richness estimate ( i . e . , total number of OTUs ) than PD-based clustering ( Table S1 ) . Comparing the pairwise distance distributions obtained from the PID- and PD- based approaches finds that at relatively short distances ( e . g . , 0–0 . 03 ) , PD-based pairwise distances are shorter than the corresponding PID-based distances , while at relatively long distances ( e . g . , greater than 0 . 1 ) , PD-based pairwise distances are longer than the corresponding PID-based distances ( Figure S3 ) . These findings suggest that differences in richness estimates result from the fact that PD-based clustering tends to merge some clusters that are found to be distinct , but closely related , by PID-based clustering . However , the overall composition of the clusters is very similar: merging of closely related clusters results in a significant reduction in estimated richness , but can produce a relatively small number of conjunction and disjunction errors . We subsequently investigated whether we could both maintain accuracy of PD-based clustering , while at the same time obtaining richness estimates more similar to PID-based results , which are thought to approximately correspond to the number of distinct microbial taxa in an environmental sample . First , we considered changing the hierarchical clustering algorithm . It has been shown that the choice of nearest-neighbor , average , or furthest-neighbor linkage in hierarchical clustering algorithms results in substantially different estimates of taxonomic richness , with average-linkage clustering performing the best for PID-based approaches [33] . In agreement with these earlier studies , we observed different OTU richness estimates when these three different linkage methods were employed in PhylOTU , with furthest-neighbor clustering producing richness estimates most similar to PID-based clustering for a given threshold ( Table S1 ) . But there is a trade-off: employing a different clustering algorithm generally reduces the accuracy with which PhylOTU clusters recapitulate PID-based OTUs , implying that while our estimate for richness may be improved by varying the clustering algorithm , we might be finding the right number of ‘wrong’ OTUs . We reach a similar conclusion if we lower the PD-clustering threshold . We naturally find a greater number of OTUs with a lower threshold , so a threshold that produces a PID-like OTU richness estimate can be identified . However , the accuracy of PD clustering relative to PID clustering becomes systematically lower as the PD threshold deviates from the PID threshold . Given these results , PhylOTU implements average-linkage and a threshold of 0 . 03 as default settings when clustering full-length SSU-rRNA sequences into OTUs ( Table 1 ) . Overall , our results imply that PhylOTU finds OTUs very similar to PID-based methods in terms of cluster composition , but that recapitulating PID-based clusters with high accuracy will generally result in a lower richness estimate . We consider the accurate clustering of sequences to be more critical than matching OTU richness , given that an equal number of clusters may be optimized between two methods while the accuracy of cluster member composition is simultaneously low . Therefore , we recommend using the default PhylOTU settings , which optimize similarity to PID-based clusters , with the caveat that lower OTU richness estimates may be produced . Next , we looked at how well PhylOTU clusters full-length sequences relative to taxonomy-guided clusters . We obtained the GenBank taxonomy information for each of the 508 full-length reference sequences and clustered them into taxonomic groups at the species level . We find that PhylOTU clusters sequences into their proper taxonomic group with high true conjunction ( 96 . 5% ) and true disjunction ( 99 . 4% ) rates at a clustering threshold of 0 . 03 ( Table S2 ) . However , similar to the results observed in the comparison with PID-based OTUs , PhylOTU tends to underestimate richness relative to GenBank taxonomy . To provide a reference for understanding these results , we conducted a similar comparison of PID-based OTUs and taxonomic groups . PID and PD clustering recapitulate taxonomic groups with similar accuracy at a clustering threshold of 0 . 03 . But , PID clustering produces a slightly closer approximation of richness relative to the taxonomy clusters , consistent with our direct comparison between PhylOTU and PID-based OTUs ( Table S2 ) . The similarity between taxonomy and PID-based OTUs is not surprising given the fact many bacterial taxa were defined via PID-based clustering of SSU-rRNA sequences ( see Discussion ) . To investigate the performance of PhylOTU on metagenomic reads versus full-length sequences , we generated 25 distinct simulation data sets using metaPASSAGE ( Riesenfeld et al . , unpublished communication ) , a recently developed , highly parameterized simulation pipeline which expands the function of the MetaSim program [34] . For each simulation , 50 of the 508 reference SSU-rRNA sequences were drawn at random to represent taxa detectable in the sample . These 50 sequences are termed “source sequences” because they are used to generate the simulated metagenomic data . Since most taxa in nature do not have full-length SSU-rRNA sequences in current databases , we used only the remaining 458 non-sampled sequences as the reference sequences for each simulation . We designated the 50 source sequences as full-length PCR products to simulate a targeted sequencing study for each simulated sample . To simulate metagenomic sequencing of the same sample , we generated in silico shotgun reads from the 50 source sequences with a read length distribution chosen to be similar to a 454-sequence library ( see Methods ) . We simulated exactly one read per source and did not simulate sampling or PCR bias to enable direct comparison of full-length and shotgun PhylOTU results . For each in silico sample , we separately applied PhylOTU to the 50 metagenomic reads and the 50 full-length sequences . We used two metrics to quantify the performance of PhylOTU on metagenomic reads: 1 ) similarity between the read and full-length sequence distance matrices , and 2 ) accuracy at which the algorithm clusters reads into OTUs relative to clusters built from full-length sequences . Comparing the PD matrices from metagenomic and full-length data sets , we observe a strong correlation between the pairwise distances computed on reads and full-length sequences . For each of the 25 simulated samples , the read and corresponding full-length-sequence distance matrices show a positive and significant correlation ( Mantel test , p<0 . 05; Figure S4 ) . Having established that pairwise PD measurements are on average similar between metagenomic reads and full-length sequences , we next investigated whether specific properties of individual metagenomic reads systematically generate errors in metagenomic PD estimates compared to full-length PD measurements . We hypothesized that PD error might be higher in shorter reads , which contribute less phylogenetic information than longer sequences , and in reads from hyper-variable regions in the SSU-rRNA locus , which will have higher than expected substitution rates . To explore these hypotheses , we calculated , for each read , a measure of the relative contribution by that read to the total PD error ( see Methods ) . This measure is designed to detect whether certain reads are placed on particularly poorly estimated parts of the phylogeny . We compared this relative error to read length , location within the SSU-rRNA locus ( mapped through a read's midpoint position in the multiple sequence alignment ) , and the amount of alignment overlap the read shares with other reads . We detected no significant correlation between relative PD error and rate variation or alignment depth . We did find a slightly negative , but significant , correlation between relative PD error and read length , suggesting that short reads may contribute more error than long reads ( Spearman's rho = −0 . 088 , p = 0 . 0028 ) . This signal disappeared when reads less than 100 base pairs ( bp ) were removed from the analysis . As a result , we incorporate a 100 bp read length cutoff in our method . Further analyses are required to comprehensively study the effects of read length and other attributes on PD estimates . Next , we compared the OTUs produced from metagenomic and full-length sequences , using PhylOTU with identical clustering settings . As illustrated in Figure 2 , this analysis reveals that even at low false conjunction rates ( meaning that few reads whose corresponding full-length sequences are in separate OTUs are clustered together ) , PhylOTU tends to correctly put reads from the same OTU in the full-length analysis into the same cluster . This indicates that PhylOTU accurately discriminates between sequence-pair conjunctions: false conjunctions do not need to be tolerated at a high rate to identify true conjunctions . Additionally , PhylOTU clusters reads substantially better than randomly permuting reads into OTU clusters . We then determined whether the performance of PhylOTU on metagenomic data could be improved by tuning the parameters of the clustering algorithm . Taking the OTUs from full-length sequences at a given clustering threshold as a gold standard , we explored how the true conjunction rate and true disjunction rate vary as functions of the threshold used to cluster the reads . There exists a tradeoff between the true conjunction and true disjunction rates as the threshold changes: at small threshold values , PhylOTU accurately separates reads into distinct OTUs , while at high threshold values , the algorithm accurately clusters sequences into the same OTU ( see Figure S5 ) . Maximizing the true disjunction rate subject to a minimum true conjunction rate of 80% , we observe that increasing the read threshold relative to the full-length sequence threshold greatly improves the agreement between the two sets of OTUs . Interestingly , we find a nearly linear relationship between the most accurate read clustering threshold and the full-length sequence threshold ( Figure S6 ) . This relationship and the accuracy of PhylOTU remains consistent up to relatively large full-length sequence clustering thresholds ( e . g . , 0 . 29 , Figure S7 ) . The linear relationship between read and full-length sequence thresholds enabled us to identify adjusted thresholds for metagenomic reads that accurately recapitulate OTUs from full-length sequences ( Table S3 ) . PhylOTU obtains 80% accuracy ( true conjunction rate = 80% , true disjunction rate = 99 . 58% ) at a read threshold of 0 . 09 , and 90% accuracy ( true conjunction rate = 90% , true disjunction rate = 98 . 73% ) at a threshold of 0 . 18 . Thus , simulations enabled us to select tuning parameters of the hierarchical clustering algorithm in PhylOTU so that the OTUs generated from shotgun read data closely resemble those that would be identified if full-length PCR products were available for each SSU-rRNA sequence in the read library . Given this insight into the accuracy with which PhylOTU clusters metagenomic reads under relatively simple simulation parameters , we evaluated how PhylOTU performs using more rigorous parameters that are reflective of situations encountered during real studies . First , in some environmental samples , the average read may be quite diverged from its closest reference sequence . Second , in many studies the number of reads will be greater than the number of reference sequences . To investigate these two issues , we first used our simulated sequences to evaluate the relationship between the mean phylogenetic distance from each read to its nearest reference sequence ( e . g . , read-to-reference distance ) and the true conjunction rate . We found no significant correlation ( Spearman's test ) . Next , we conducted additional simulations based on sampling reads from full-length Bacterial SSU-rRNA sequences in the SILVA database [35] . This investigation allowed us to generate data sets with more reads than reference sequences and where read-to-reference distances exceeded those in our primary simulations . The latter property is important because of known phylogenetic sampling biases , especially for sequenced genomes [36] . For each of 15 independent simulations , we randomly sampled 1 , 000 SSU-rRNA sequences from the SILVA database , reflecting the approximate number of SSU-rRNA reads expected when performing one run of next-generation sequencing on a shotgun library . These 1 , 000 source sequences were then used to simulate metagenomic reads as described above . Reference sequences were pruned from both the source and simulation phylogenies and full-length source sequences and simulated reads were then clustered into OTUs . In these simulations , the average distance between each read and its nearest source is an order of magnitude greater than that observed in our previous simulation analysis ( 0 . 182 versus 0 . 010 mean read-to-reference distance ) , which is expected given that the SILVA database is highly populated and comprised of phylogenetically diverse sequence data . Evaluating the accuracy of PhylOTU under these conditions reveals high true disjunction rates , similar to those observed in the RDP reference library based simulations . True conjunction rates are somewhat lower , but still meet our accuracy standards . For example , at a read threshold of 0 . 15 , PhylOTU clusters metagenomic reads with an 80% true conjunction rate and a 98 . 8% true disjunction rate ( Figure S8 , Table S4 ) , when compared to full-length sequences clustered at a threshold of 0 . 03 ( corresponding to an 86 . 8% true conjunction rate and a 98 . 8% true disjunction rate under RDP reference library based simulation parameters ) . This suggests that read library size and phylogenetic novelty do have a small impact on the accuracy of PhylOTU , but that they can generally be compensated for by appropriately tuning the clustering cutoff . To demonstrate the utility of PID-based clustering of metagenomic data , we analyzed the pooled Global Ocean Survey ( GOS ) metagenomic read library [21] with PhylOTU . This data set represents the most extensive publicly available metagenomic sequence library generated to date , with the exception of the Illumina library generated by Qin et . al , which contains reads that are too short to process via PhylOTU [37] . Additionally , many of the GOS sampling sites were also explored with deep , targeted sequencing of the SSU-rRNA locus enabling comparisons of shotgun and PCR libraries . Despite the use of Sanger sequencing , the mean SSU-rRNA metagenomic read length is roughly similar to that used in our simulation analysis ( 518 bp ) . Thus , the GOS read library represents the best opportunity to explore PhylOTU's ability to discover novel taxa from metagenomic data . Of the 10 , 133 , 846 Sanger sequenced reads in the library , PhylOTU identifies 14 , 320 Bacterial SSU-rRNA homologs , of which 12 , 020 passed the method's filters and could be used for OTU discovery . Previous work using the same library was constrained to analysis of 4 , 125 high-confidence SSU-rRNA assemblies [21] , the difference resulting from the fact that many of the SSU-rRNA reads identified by PhylOTU were either assembled in this prior analysis or excluded from this early work given assembly constraints . PhylOTU clusters the 12 , 020 SSU-rRNA reads into 833 OTUs at a PD threshold of 0 . 15 , which , according to our SILVA-based simulation analysis , corresponds to a full-length threshold of 0 . 03 . Applying a cutoff of 0 . 09 , which was identified as the appropriate corresponding cutoff from the RDP reference library based simulations , identifies 1 , 078 OTUs . We also identify 192 Archaeal SSU-rRNA sequences , 79 of which pass the quality control filters and cluster into 7 OTUs when using the 0 . 15 threshold and 10 OTUs when using a threshold of 0 . 09 . This compares to the 811 total OTUs identified by Rusch et . al . via analysis of assembled SSU-rRNA reads at the 97% identity level . We have made our designation of OTUs derived from GOS metagenomic reads and PCR sequences available at BioTorrents [23] . This comparison reveals the ability of assembly-free methods such as PhylOTU to identify novel taxa missed by approaches that rely upon assembled contigs . The GOS project also generated 6 , 413 full-length SSU-rRNA sequences via targeted sequencing of PCR products from six of the 73 geographical sites surveyed [38] . We evaluated the ability of PhylOTU to discover novel taxa in shotgun data by comparing the OTUs identified from metagenomic reads to those identified from full-length PCR data from these six sites . We applied PhylOTU to both data sets and corrected for the difference in sequence types by adjusting the read threshold relative to the full-length sequence threshold according to our simulation analysis . Specifically , we used a read threshold of 0 . 15 and a full-length sequence threshold of 0 . 03 to evaluate diversity at approximately the species level . We compared the number of OTUs identified per sequence across methods by conducting a rarefaction analysis ( Figure 3 ) [39] . For each method and for subsets of the full data set from one to the observed number of sequences , we drew 100 random subsets of sequences from each data set and calculated the average number of OTUs identified by each method for that number of sequences . This allowed a comparison of the effect of read threshold and sequencing method on the total number of OTUs and rate of OTU accumulation . While there are more PCR SSU-rRNA sequences ( N = 6 , 413 ) and OTUs ( N = 1 , 563 ) than metagenomic SSU-rRNA reads ( N = 1 , 233 ) or OTUs ( N = 242 ) , when normalized for the number of sequences in each library , the number of OTUs identified per sequence are similar for the two libraries ( 0 . 24 for PCR sequences , 0 . 20 for shotgun sequences ) . After normalizing by the average sequence length for each library , however , the shotgun sequence data generates three times as many OTUs per sequenced SSU-rRNA base relative to PCR-generated sequences ( 4 . 63×10−4 and 1 . 66×10−4 OTUs per sequenced base , respectively ) . Evaluating the intersection of OTUs identified by the two libraries when they were pooled together and processed by PhylOTU reveals a shared set of OTUs as well as unique OTUs missed by each method ( Figure 4 ) . Because this pooled data set contains both full-length sequences and shotgun reads , we evaluated the distribution of sequences across OTUs for a range of thresholds ( Figure S9 ) and made comparisons between OTUs obtained at thresholds appropriate for full-length sequence ( 0 . 03 ) and shotgun reads ( 0 . 15 ) . Specifically , at the 0 . 15 threshold , the metagenomic library contains 80 OTUs that are not revealed through analysis of the PCR library , while the PCR library contains 1 , 254 unique OTUs at the 0 . 03 threshold . Normalizing the number of unique OTUs by the number of sequences per library finds that the PCR-based sequences encode more unique OTUs per sequence ( 0 . 19 ) than shotgun sequences ( 0 . 06 ) . However , comparing the change in the number of OTUs uniquely identified by shotgun sequence data to the change in the number of OTUs uniquely identified by PCR sequence data across thresholds suggests that shotgun sequences reveal unique OTUs that are highly diverged from those identified using PCR-based sequences ( Figure S9 ) . Despite the amount of sequencing conducted , the steep slopes of the rarefaction curves indicate that sampling has not been saturated at these geographical sites . Thus , deeper sequencing through either method is warranted and may either increase or reduce the number of unique OTUs . We compared the sequences from the novel OTUs identified from metagenomic reads to the Greengenes SSU-rRNA sequence database to determine if any other PCR-based study revealed the existence of these taxa [40] . Using traditional percent identity cutoffs and the Greengenes database as a reference of nearest neighbor percent identity ( e . g . , DNAML distance ) , we find that many of the metagenomic read OTUs represent novel species , genera and families . We further characterized the taxonomic distribution of these novel OTUs via taxonomic classification through comparison of the sequences to the RDP database . OTUs unique to the metagenomic reads are predominantly members of the Alpha- ( 19% ) and Gamma-proteobacteria ( 11% ) , Actinobacteria ( 15% ) , and Bacteroidetes ( 12% ) . We also find that the Bacteroidetes , Verrucomicrobia , Firmicutes , and Delta-proteobacteria are enriched in the OTUs unique to shotgun sequences relative to OTUs unique to PCR data or shared between metagenomic and PCR data ( Table S5 ) . In addition , several clades , including TM7 , Planctomycetes , OD1 , and WS3 were only identified via analysis of metagenomic sequence . Reasoning that the universal PCR primers traditionally employed in most targeted sequencing studies ( i . e . , 8F , 27F , 1525R , 1429R [41]–[43] ) , may inefficiently amplify or fail to amplify the SSU-rRNA sequences uniquely identified via shotgun sequences , we searched SSU-rRNA reads that overlap the universal priming sites for the presence of sequence complementary to universal SSU-rRNA primers . Of the shotgun reads that overlap a universal priming site ( N = 6 ) , we find two that share a unique point mutation relative to the remaining overlapping reads and the 8F and 27F primer sequences ( Figure S10 ) . Prior work demonstrated that differences between the primer and template sequences can result in PCR amplification bias [43] . Our findings support the use of universal-primer-sequence variants that include degenerate positions , such as those described in [43] , to improve the resolution of lineages harboring this variant through PCR-based investigations . For the remaining reads that do contain a universal priming site , we do not know if the sequence they were generated from contains the anti-sense priming site because these reads do not span the length of the SSU-rRNA locus . Alternatively , these reads may have been obtained from discontinuous rRNA , such as the rRNA sequence found in the mitochondria of Chlamydomonas [44] . Should the priming sites be located in relatively disparate parts of the genome , discontinuous rRNA may fail to amplify even if the universal primer sites are highly conserved .
We have developed a novel method that enables comparison of non-overlapping metagenomic SSU-rRNA reads and their assignment into OTUs . This is the first automated procedure that identifies OTUs directly from non-overlapping metagenomic reads , which facilitates the identification of taxa potentially overlooked by targeted sequencing studies and leverages the vast quantities of shotgun sequencing data currently being produced by environmental and microbiome studies . The key innovation allowing us to compare non-overlapping reads is our use of phylogenetic distance ( PD ) to cluster reads into OTUs in place of PID . Building a phylogenetic tree requires that at least some of the sequences within the input alignment overlap . Thus , we incorporate high-quality , full-length reference sequences into the SSU-rRNA sequence alignment to guide the phylogenetic placement of metagenomic reads . The accuracy of this approach is constrained , at least in part , by the phylogenetic diversity of the reference sequences and the means by which the phylogenetic algorithm processes missing data . For example , it is challenging to assess distances between non-overlapping shotgun reads derived from a similar place in the phylogeny , even via comparison to full-length reference sequences . We determined the robustness of our method by evaluating the OTU assignment accuracy of simulated metagenomic reads relative to their full-length sources , finding that the relative PD between a pair of reads is on average highly consistent with the relative PD between full-length sources . This result indicates that metagenomic reads can be assigned to OTUs with high accuracy by simply scaling the clustering threshold . We also tested whether clustering based on PD could accurately recapitulate clustering based on PID for full-length reads where both methods may be applied . Processing 508 full-length reference sequences via both algorithms reveals that PD accurately assigns sequences into OTUs when compared to the PID OTUs . However , this analysis also reveals that PD results in lower richness estimates relative to PID . This phenomenon appears to be due to a difference in the relative distances between sequences . Specifically , the phylogenetic approach appears to shorten the estimated distance between closely related sequences , relative to the PID approach . This is likely due to the fact that the PD approach employs a weighted substitution model when calculating distances , while the PID approach treats all substitutions with equal weight . Thus , while the hierarchical structure of the clusters is generally consistent between the two methods , as revealed by the cluster composition accuracy analysis , sister OTUs in the PID analysis tend to be merged together via the PD approach . For this reason , it may be necessary to take into account this systematic difference in order to compare the diversity results from a PD-based study with a PID-based study . A similar pattern is observed when the PD-based and PID-based OTUs are compared to OTUs constructed from GenBank taxonomy terms . Specifically , both methods accurately cluster the 508 full-length reference sequences at the species and genus level . Both methods also tend to underestimate the richness , though PID produces an estimate more in line with the taxonomy-guided clusters . Though this analysis serves as a useful benchmark , a more thorough investigation of richness estimation may be warranted in future work for several reasons . First , GenBank taxonomy terms do not necessarily recapitulate the true taxonomic signal or correspond to monophyletic clades . Second , there are known errors in taxonomic assignment and annotation of GenBank sequences [45]–[46] . In addition , many of the taxonomy terms found in GenBank were identified by using the PID approach to classify sequence data . As a result , the reference used in this comparison is necessarily biased towards the PID approach . Regardless , this analysis exemplifies the fidelity with which PhylOTU clusters sequences relative to a commonly adopted interpretation of taxonomy . Having demonstrated the accuracy with which sequences , both full-length and shotgun , are clustered into OTUs using PD , we applied PhylOTU to the Global Ocean Survey ( GOS ) metagenomic library . Previous characterizations of SSU-rRNA diversity found in the GOS library were limited to full-length sequences amplified via PCR and full-length contigs produced from high-confidence read assemblies [21] . To demonstrate the ability to discover novel taxa directly from metagenomic data , we compared the PD-based OTUs from full-length PCR sequence to those identified from metagenomic reads . Several conclusions can be drawn . First , targeted sequencing produces more SSU-rRNA sequence per sequenced base ( since much of the metagenomic library targets other genes ) , but fewer OTUs per sequenced SSU-rRNA base compared to metagenomic sequencing . Second , metagenomic sequences analyzed via PD reveal taxa missed by the targeted sequencing study . In particular , PhylOTU clusters metagenomic reads into OTUs belonging to several Bacterial Phyla overlooked by the PCR-generated sequences . We were not able to detect the presence of completely conserved universal PCR priming sites for some of these sequences , which supports the theory that some faction of the microbial biosphere may be hidden from the view of PCR-based investigation . Deeper sequencing of either library could erode the signal of library-specific OTUs . Nonetheless , the distinct taxonomic composition of the metagenomic-only OTUs compared to the shared and PCR-only OTUs ( Figure 4 , Figure S9 , and Table S5 ) supports the hypothesis that the shotgun libraries would continue to contain unique diversity even after deeper sequencing of both libraries . Thus , we conclude that there are real differences in the identified diversity and composition of these communities depending on the sequencing method employed . Metagenomic sequencing is an increasingly common means of investigating microbial communities . We expect methods , such as PhylOTU , which enable analysis of unassembled , non-overlapping reads to play an important role in the progress of this field . Future developments will include robust characterization of sources of phylogenetic error to improve methodological accuracy , optimization of PD-based richness estimations in conjunction with optimized cluster composition , and the inclusion of more sophisticated phylogenetic algorithms . Additionally , because the output of PhylOTU includes estimates of abundances for the resulting OTUs , future developments will explore the possibility of using PhylOTU to conduct weighted analyses of community structure by incorporating these abundance estimates . We also anticipate that our phylogenetically-based framework can be expanded beyond its current application to improve OTU identification in several ways , including the incorporation of phylogenetic structure and the utilization of multiple loci when designating of OTUs . When coupled with PCR-based sequencing investigations , this type of bioinformatic analysis of metagenomic data should result in a more comprehensive view of microbial biodiversity .
|
Microorganisms comprise the majority of the biodiversity on the planet . Because the overwhelming majority of microbes are not readily cultured in the laboratory , researchers often rely on PCR-based investigations of genomic sequence to characterize microbial diversity . These analyses have dramatically expanded our understanding of biodiversity , but due to methodological biases PCR-based approaches may only reveal part of the microbial biosphere . Shotgun sequencing of environmental DNA , known as metagenomics , avoids the biases associated with targeted amplification of genomic sequence and can provide insight into the diversity hidden from traditional investigations . However , the fragmentary , non-overlapping nature of shotgun sequence data makes it intractable to analyze with existing tools . Here , we present PhylOTU , a novel computational method that enables accurate characterization of microbial diversity from metagenomic data . We process over 10 million metagenomic sequences obtained from the global open ocean to identify novel Bacterial taxa and reveal the presence of microorganisms overlooked by investigation of PCR-based sequences from the same samples . These results suggest that to fully characterize microbial biodiversity requires a novel bioinformatics toolbox for analysis of shotgun metagenomic data .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion"
] |
[
"microbiology/environmental",
"microbiology",
"genetics",
"and",
"genomics/microbial",
"evolution",
"and",
"genomics",
"evolutionary",
"biology/microbial",
"evolution",
"and",
"genomics",
"ecology/marine",
"and",
"freshwater",
"ecology",
"computer",
"science/applications",
"marine",
"and",
"aquatic",
"sciences/ecology",
"microbiology/microbial",
"evolution",
"and",
"genomics",
"computational",
"biology/protein",
"homology",
"detection",
"ecology/community",
"ecology",
"and",
"biodiversity",
"ecology/environmental",
"microbiology",
"marine",
"and",
"aquatic",
"sciences/bioinformatics"
] |
2011
|
PhylOTU: A High-Throughput Procedure Quantifies Microbial Community Diversity and Resolves Novel Taxa from Metagenomic Data
|
Cellular differentiation entails reprogramming of the transcriptome from a pluripotent to a unipotent fate . This process was suggested to coincide with a global increase of repressive heterochromatin , which results in a reduction of transcriptional plasticity and potential . Here we report the dynamics of the transcriptome and an abundant heterochromatic histone modification , dimethylation of histone H3 at lysine 9 ( H3K9me2 ) , during neuronal differentiation of embryonic stem cells . In contrast to the prevailing model , we find H3K9me2 to occupy over 50% of chromosomal regions already in stem cells . Marked are most genomic regions that are devoid of transcription and a subgroup of histone modifications . Importantly , no global increase occurs during differentiation , but discrete local changes of H3K9me2 particularly at genic regions can be detected . Mirroring the cell fate change , many genes show altered expression upon differentiation . Quantitative sequencing of transcripts demonstrates however that the total number of active genes is equal between stem cells and several tested differentiated cell types . Together , these findings reveal high prevalence of a heterochromatic mark in stem cells and challenge the model of low abundance of epigenetic repression and resulting global basal level transcription in stem cells . This suggests that cellular differentiation entails local rather than global changes in epigenetic repression and transcriptional activity .
Resetting of the transcriptional program is the key driver for cell type specification during organismal development [1] , [2] . While embryonic stem ( ES ) cells bear the fascinating ability to acquire very diverse fates , derived somatic stages are usually irreversible under physiological conditions . This unidirectionality has been suggested to depend in part on epigenetic repression of lineage unrelated genes [3] , [4] . Accordingly , ES cell plasticity was suggested to rely on a low prevalence of heterochromatin and coinciding promiscuous low-level expression of many genes in stem cells [5]–[11] . In line with this model , distinct changes in nuclear staining had previously been observed by electron microscopy during cellular differentiation [12] , [13] . Further , a subset of promoters was shown to become DNA methylated [14]–[16] and the repressive histone modifications H3K27me3 and H3K9me3 were reported to locally expand in differentiated cells [9] . Here , we set out to test the model of widespread heterochromatinization via monitoring of the differentiation-coupled dynamics of H3K9me2 , a repressive epigenetic modification , which appears to be the most abundant heterochromatic modification and has recently been reported to cover large domains in differentiated cells [17] . Unexpectedly , we found that H3K9me2 is not only highly abundant in terminally differentiated cells , but already occupies large parts of the genome in pluripotent stem cells . In this cellular state , H3K9me2 occupies most genomic regions devoid of transcription and certain histone modifications . While our analysis revealed discrete local changes particularly at gene bodies , we observed little global increase in H3K9me2 during differentiation . This unexpected finding motivated us to revisit the model of promiscuous low-level gene expression in undifferentiated cells by quantitative RNA sequencing . Remarkably , we found the actual number of low-level expressed genes , postulated hallmarks of stem cells to be equal between both developmental states . Together , our findings challenge the model of promiscuous basal gene expression as a distinct property of pluripotency and a widespread increase of heterochromatin during cellular differentiation .
To asses differentiation associated dynamics of the repressive histone modification H3K9me2 we made use of a highly pure and robust murine in vitro neurogenesis model [18] , which we previously used to profile histone and DNA methylation [14] . Here , we generated profiles for H3K9me2 in pluripotent embryonic stem cells and derived terminally differentiated pyramidal neurons . We made use of custom tiling arrays covering 10% of the mouse genome including all well-annotated promoters , several large multi-gene loci and the complete chromosome 19 ( see Figure S1 and Text S1 ) . The chromosomal profiles for H3K9me2 revealed domains of enrichments that upon visual inspection were highly comparable between stem cells and the neuronal state ( Figure 1A ) , which is further supported by a high overall pair-wise correlation ( Figure 1B ) . Despite this overall similarity we noticed confined regional differences ( Figure 1A ) , a finding which is consistent with the fact that biological replicates of H3K9me2 are more similar than the patterns between cell states ( Figure 1B ) . We also included in our comparison a recently published dataset for H3K9me2 in a distinct ES cell line [17] , which shows high correlation to our ES cell datasets despite different experimental conditions ( Figure S2 ) . Of note , analysis of the H3K9me2 dataset from Wen et al . [17] revealed that chromosome 19 behaves similar to the other chromosomes ( Figure S3 ) , suggesting that our results can be extrapolated to the entire genome . Together this demonstrates that our H3K9me2 data are reproducible and of high resolution , yet overall patterns appear to be highly similar between a pluripotent and a terminally differentiated state . Visual inspection suggests that H3K9me2 covers large domains in both ES cells and neurons ( Figure 1A ) . To quantitatively define the actual location and sizes of domains we applied a Hidden-Markov-Model ( HMM ) analysis to the microarray data . This unsupervised statistical method is a widely accepted approach for unbiased data segmentation in epigenome analysis [19] , [20] . The HMM analysis not only agreed with and statistically corroborated the visual impression of the raw data , but also yielded robust results under variable settings ( Figure S4 ) . It revealed that over 50% of chromosome 19 is covered by H3K9me2 in ES cells ( Figure 1C ) . Using the same approach for the H3K9me2 data in the ES cell-derived terminally differentiated neurons , we detected a modest yet reproducible 5% increase of genomic regions covered by H3K9me2 ( Figure 1C ) . We conclude that global coverage and size of H3K9me2 domains is nearly identical between ES cells and derived post-mitotic pyramidal neurons . In line with this finding we do not detect a significant change in global H3K9me2 levels by Western blot detection ( Figure 1D ) . Moreover , H3K9me2 domain features of ES cells and neurons show similar size distribution and median length ( Figure S4 ) . Further analysis revealed that changes in H3K9me2 between the two examined cellular states are rare; 88% of H3K9me2 occupied regions in ES cells are also occupied in neurons ( Figure 1F ) . Notably , regions that change in H3K9me2 state tend to be small and are below the average size of invariant domains ( Figure S5 ) . Consistent with the overall increase of 5% in H3K9me2 coverage during differentiation , regions which gain H3K9me2 are more frequent and of larger size than regions showing a loss of the mark ( Figure S5 and Figure S6 ) . Interestingly , most of the larger regions ( >10 kb ) that gain H3K9me2 are located within genes , starting downstream of the promoter region ( Figure 2A and Figure S5 ) . These global findings are fully reproducible in single gene controls ( Figure 2B ) and consistent with a focused comparison of only genic regions ( Figure 2C ) . Importantly , this shows that our experimental and data analysis approach is indeed highly sensitive to detect differences if they do occur . Interestingly , many genes that acquire H3K9me2 show slightly reduced expression in many cases , while others increase expression upon gain of the modification ( Figure 2D and Figure S5 ) . This suggests that the gain of H3K9me2 , while highly selective for gene bodies , cannot simply be explained by the silencing of gene activity . Given the high prevalence of H3K9me2 , we next asked how its presence relates to a distinct repressive chromatin modification , namely trimethylation of H3K27 ( H3K27me3 ) . This mark is set by the Polycomb pathway and often occurs in domains of several kilobases [9] , [21] , [22] . We find that both heterochromatic histone modifications occur mutually exclusive even when in direct neighborhood as illustrated by the sharp boundaries of the H3K9me2 signal next to H3K27me3 peaks ( Figure 3A , 3B and 3C ) . This is consistent with a previous study in human embryonal carcinoma cells that was limited to promoters [23] . We further related H3K9me2 occupancy to regions with transcriptional activity or presence of the active modification H3K4me2 . Active regions are mutually exclusive with H3K9me2 in ES cells but surprisingly to a lesser extent in neurons ( Figure 3D ) . The compatibility of H3K9me2 and gene expression in neurons is however limited to gene bodies and does not occur in the promoters of expressed genes , consistent with the former regions gaining H3K9me2 during differentiation ( Figure 2A and Figure S7 ) . We find the majority of H3K4me2 regions to be mutually exclusive with H3K9me2 in stem cells ( Figure 3D ) . In neurons , a small number of regions become co-occupied , again most of these being within transcribed genes ( Figure 3D ) . Importantly , an HMM independent analysis confirms that regions with high H3K9me2 enrichment do not overlap with transcribed genes in stem cells , yet a subset does in neurons ( Figure 3E ) . We conclude that gain of H3K9me2 during differentiation has only a minor effect on the overall chromosomal coverage of the modification , yet it occurs highly localized and preferentially at genic regions . Our finding of surprising conservation of heterochromatin patterns in a refined model of differentiation let us to revisit the transcriptome in a quantitative manner using high throughput RNA sequencing ( RNAseq ) . RNAseq in ES cells and derived neurons revealed the expected down regulation of stem cell specific genes and induction of neuron specific genes ( Figure 4A ) . Further , when counting RNA molecules after gene mapping and normalization , both cell types displayed a characteristic bimodal distribution . This reflects a group of genes that is in a clear off state with no detectable RNA molecules and a second peak of expressed genes ( Figure 4B ) . Separating these two groups of genes by a stringent cutoff revealed that out of 35′606 transcription units , 45% are expressed in ES cells . Interestingly , we identified a slightly higher number ( 50% ) of expressed genes in terminally differentiated neurons , indicating that differentiation of stem cells is not coinciding with a reduced number of highly expressed genes . This agrees with a recent report that suggested that stem cells and somatic cells do mainly differ in the number of low-level expressed genes due to a global reduction of basal gene activity in the course of lineage-commitment and loss of pluripotency [10] . To test this in our in vitro differentiation system , we grouped the genes that could not clearly be assigned to the on or off state , into a separate class of genes expressed at low to background level ( Figure 4B ) . This analysis reveals that in stem cells 16% of all transcription units show a basal expression level . Surprisingly however , the proportion of genes expressed at such low level ( 14% ) is very similar in neurons . This unexpected finding prompted us to conduct an additional RNAseq experiment in a second fully differentiated somatic murine cell type; primary mouse embryonic fibroblasts ( MEFs ) . Interestingly , also fibroblasts display a similar transcriptional landscape as stem cells , with 46% of all transcription units being highly expressed and 13% being expressed at basal levels . Hence , this qualitative similarity of expression patterns is not specific to the neuronal subtype we generated in vitro , but appears to be a more general property of both undifferentiated and differentiated cells . Thus , while transcripts expressed at low levels show little overlap between stem cells and somatic cells ( Figure S8 ) , their numbers are remarkably similar . Stem cells do also not show an increased number of highly expressed genes . Based on additional analysis we can exclude that this similarity of the transcriptional landscape is a consequence of insufficient sampling ( Figure S9 ) . Moreover , it is not limited to genic regions as the abundance of transcripts generated from diverse classes of endogenous repeat is comparable between stem cells and neurons ( Figure 4C ) .
Embryonic stem cells are characterized by their potential to differentiate into any cell type of the three germ layers in the developing embryo , while somatic cells lose this developmental plasticity upon lineage-commitment . Despite its relevance for our understanding of development and disease , the molecular determinants of pluripotency are still not fully understood and the factors responsible for this uniqueness of stem cells are actively debated [24] . Our study of gene expression and an abundant heterochromatin mark reveal surprising conservation of the transcriptome and epigenome landscape between pluripotent and unipotent cells . H3K9me2 is already highly prevalent in ES cells , arguing that the pathways that mediate H3K9me2 are highly active in stem cells , and serve similar functions as in somatic cells , which only show a slight increase of the mark ( from 53 to 58% ) . Interestingly , the observed gain occurs very localized at gene bodies and does not necessarily coincide with lower transcription of the corresponding gene . The analysis of regions that acquire H3K9me2 during differentiation further revealed a differentiation specific coexistence of H3K9me2 and transcriptional activity , which is not detected in the pluripotent state and which could be involved in modulating expression in the differentiated cell . Nevertheless , despite the subtle increase we detected , an involvement of H3K9me2 in globally regulating cell-type specific gene repression appears unlikely . The limited dynamics as compared to massive transcriptome reprogramming and the limited correlation between expression and gain of H3K9me2 at target genes argue against H3K9me2 as being a major player in setting up gene expression programs . These findings disagree with a recent report that suggested absence of large H3K9me2 domains in ES cells and found a striking increase in differentiated cells [17] . Notably , our ES cell profile for H3K9me2 is similar to the one generated previously ( Figure 1B and Figure S2 ) , making data analysis a likely explanation for the discrepancy . We applied an unbiased approach that is insensitive to variations between arrays and which we show to lead to similar results under various parameter settings ( Figure 3 and Figure S10 ) . As already discussed by Fillion and van Steensel [25] , the previous study relied on defined thresholds , which can be prone to false estimation of differences particular in the absence of biological replicates [25] . Widespread low-level expression in stem cells has previously been reported and interpreted as a sign of pluripotency [5] , [6] , [10] . It has been speculated that this basal promiscuous activity would poise genes for rapid induction upon receipt of differentiation cues [5] , [6] , [10] . Using mRNA sequencing in our differentiation paradigm does not confirm this model . We do not find evidence of elevated transcription throughout the genome or on specific chromosome ( Figure S8; [10] ) . A likely explanation for these discrepancies is that microarrays , which were used in the previous studies , overestimate low level signal due to cross-hybridization [26] . In the present study we utilized RNAseq , which permits an actual counting of RNA molecules and thus enables accurate discrimination between very low and no expression . Notably , RNA sequencing experiments have recently put other findings in question that relied on quantifying small transcriptome differences detected by microarrays . For example , recent RNAseq data challenged the presence of pervasive intergenic transcription [26] and the existence of transcriptional dosage compensation of the single male X chromosome [27] . In addition to the increased sensitivity of RNAseq that can explain the differences to previous studies , the presence of a small fraction of differentiated cells in suboptimal conditions of stem cell culture could similarly contribute to an overestimation of the number of genes that are actually expressed in stem cells . Notably , the ES cell differentiation protocol applied by us is optimized to reduce the number of differentiated cells in the culture [18] . In summary , our analysis suggests to revisit the model of massive heterochromatinization during cellular differentiation via a global increase in repressive histone marks [5]–[9] , [17] and coinciding repression of basal gene activity [5] , [6] , [10] . Our data together with previous reports on dynamics of DNA methylation , H3K9me3 and the Polycomb pathway between pluripotent and somatic cells [9] , [14] , [15] , [28]–[31] support a model whereby repressive chromatin is already highly active in stem cells and that epigenome reprogramming entails localized changes of repressive histone modifications and DNA methylation at regulatory regions that specify and stabilize lineage specification and terminal differentiation [32] . It will be interesting to determine if these local differences account for the observed changes in nuclear morphology [33] . Notably , epigenetic repression can be overcome by the local activity of transcription factors upon strong induction cues during normal differentiation or artificially during generation of induced pluripotent stem cells ( iPS ) [34] and might therefore safeguard rather than actively channel development via direct transcriptome regulation .
Wild-type embryonic stem cells ( 129Sv-C57Bl/6 ) were cultured and differentiated as previously described [14] , [35] . Fibroblasts were isolated from wild-type embryos ( C57Bl/6 ) . Peptide sequences can be found in Table S2 . Western blot analysis was performed with acid extracts using 1/1000 dilutions of either anti-H3K9me2 ( Abcam no . 1220 ) or anti-H4 ( Upstate , no . 07–108 ) antibodies . Blots were developed with ECL reagent ( GE Healthcare ) . ChIP experiments were performed as described before [14] , starting with 70 µg of chromatin and 5 µg of the following antibodies: anti-dimethyl-H3K9 ( LP Bio , no . AR-0108 ) , anti-dimethyl-H3K9 ( Abcam no . 1220 ) , anti-trimethyl-H3K27 ( Upstate , no . 07–449 ) , anti-dimethyl-H3K4 ( Upstate , no . 07–030 ) . H3K9me2 ChIP samples were amplified using the WGA2 kit ( Sigma ) and hybridized to a custom tiling microarray ( NimbleGen Systems Inc . , see below ) . H3K27me3 and H3K4me2 ChIP libraries for Illumina sequencing were prepared with the Illumina ChIP-Seq DNA Sample Prep Kit ( Cat# IP-102-1001 ) according to Illumina's instructions and sequenced on the Genome Analyzer II following the manufacturer's protocols . ChIP-real time PCR was performed using SYBR Green chemistry ( ABI ) and 1/40 of ChIP or 20 ng of input chromatin per PCR reaction . Primers are listed in Table S1 . H3K9me2 ChIP samples were hybridized to custom designed microarrays representing all well-annotated promoters , several large multi-gene loci and the complete chromosome 19 with an average probe spacing of 100 bp and a total of 2 . 1 million features ( HD2 . 1 , NimbleGen Systems Inc ) . Sample labeling , hybridization and array scanning were performed by NimbleGen Systems Inc . according to standard procedures . For analysis , raw fluorescent intensity values were used to calculate log2 of the bound/input ratios for each individual oligo . Subsequently , for comparison all arrays were normalized to a median log2 = 0 and scale normalized to have the same median absolute deviation using the “LIMMA” R/Bioconductor package [36] , [37] . RNA from ES cells , neurons and fibroblasts of two independent biological replicates each was used for cDNA preparation using oligo dT primers followed by sequencing on an Illumina GA II analyzer . Reads were mapped to the Mus musculus transcriptome and normalized to transcript length and sequencing library size ( for details see Text S1 ) . Unless otherwise stated , H3K9me2 enriched regions were identified by HMM and H3K4me2 and H3K27me3 peaks using MACS peak finder [38] . Active regions were defined as RefSeq transcription units with a normalized RNAseq log2 read count above 5 ( for details see Text S1 ) . Microarray design , hybridization and analysis , ChIPseq and RNAseq analysis and additional references are described in Text S1 . Microarray and deep sequencing data were deposited at NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE27866 ( http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? acc=GSE27866 ) .
|
Epigenetic modifications of DNA and bound histones are major determinants of cell type–specific gene expression patterns . A prevalent model in stem cell biology suggests that the loss of pluripotency entails global increase in heterochromatin and coinciding shutdown of lineage unrelated genes . We performed analysis of both H3K9 dimethylation pattern and the global transcriptome in an advanced murine neuronal differentiation model . In this paradigm , we do not find evidence for a global increase in heterochromatic H3K9 dimethylation or reduction of transcriptome complexity as stem cells become terminally differentiated post-mitotic neurons . This suggests that pluripotent embryonic stem cells are not per se unique in regards to heterochromatin abundance and transcriptional plasticity as compared to somatic cells . Instead , focal changes in chromatin might help to stabilize cellular states at any developmental stage .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"developmental",
"biology",
"cell",
"fate",
"determination",
"stem",
"cells",
"embryonic",
"stem",
"cells",
"gene",
"expression",
"genetics",
"epigenetics",
"biology",
"chromatin",
"genetics",
"and",
"genomics",
"cell",
"differentiation",
"dna",
"transcription",
"histone",
"modification"
] |
2011
|
Genomic Prevalence of Heterochromatic H3K9me2 and Transcription Do
Not Discriminate Pluripotent from Terminally Differentiated
Cells
|
Erythema nodosum leprosum ( ENL ) is a debilitating multisystem disorder which complicates leprosy . It is characterised by fever , malaise and painful erythematous cutaneous nodules . ENL is often recurrent or chronic in nature and frequently severe . Patients often require prolonged treatment with high doses of oral corticosteroids . There are no data on the mortality associated with treated ENL . The notes of patients who were admitted , discharged , transferred to another facility or died with a diagnosis of leprosy or a leprosy-related complication for a five year period were reviewed . 414 individuals were identified from the ward database . 312 ( 75 . 4% ) patient records were located and reviewed . Ninety-nine individuals had ENL and 145 had a Type 1 reaction . The median age of individuals with ENLwas 25 years . Eight patients with erythema nodosum leprosum died compared with two diagnosed with Type 1 reaction . This difference is statistically significant ( p = 0 . 0168 , Fisher's Exact Test ) . There is a significant mortality and morbidity associated with ENL in this Ethiopian cohort . The adverse outcomes seen are largely attributable to the chronic administration of oral corticosteroids used to control the inflammatory and debilitating symptoms of the condition .
Leprosy is a chronic granulomatous infection predominantly of the skin and peripheral nerves caused by Mycobacterium leprae [1] . 232 , 857 new cases of leprosy were reported to the World Health Organization ( WHO ) in 2012 [2] . The treatment of the infection with multi-drug therapy ( MDT ) is highly effective however a significant proportion of individuals develop immune-mediated inflammatory states known as reactions . Leprosy reactions are important because they are the major cause of nerve function impairment which leads to leprosy associated disability and its life altering consequences . Reactions may occur before , during and after successful completion of MDT . Type 1 reactions ( T1R ) affect patients with the borderline forms of leprosy causing inflammation in pre-existing leprosy skin lesions and neuritis [3] . Type 2 reactions or erythema nodosum leprosum ( ENL ) affect approximately 10% of those with borderline lepromatous ( BL ) leprosy and 50% of individuals with lepromatous leprosy ( LL ) [4] . It is acknowledged that there is a lack of good epidemiological data on the true incidence of ENL [5] . A further risk factor for developing ENL is a mean bacterial index ( BI ) greater than 4 on slit-skin smear [4] . It is estimated that over 50 , 000 of the new leprosy patients diagnosed each year are at risk of ENL . In Ethiopia 5 . 3% of multibacillary patients enrolled in a field cohort study developed ENL however this cohort includes patients with borderline tuberculoid leprosy who are not at risk of ENL [6] . In Ethiopian patients with BL leprosy and LL 5% developed ENL before or during treatment with 24 months of MDT [7] . ENL is characterised by the development of crops of tender cutaneous and subcutaneous nodules in association with generalised malaise , pain and fever [1] . Other organ systems are often involved and patients may experience iritis , neuritis , rhinitis , arthritis and dactylitis , lymphadenitis , orchitis , hepatitis , peripheral oedema , and renal impairment . The histology of ENL lesions classically shows an intense perivascular infiltrate of neutrophils throughout the dermis and subcutis [8] . Tissue oedema and vessels exhibiting fibrinoid necrosis may also be present . ENL has some features of an immune complex mediated disease . Direct immunofluorescence studies have demonstrated granular deposits of immunoglobulin and complement in the dermis in ENL lesions but not in those of uncomplicated LL disease [9] . There is evidence of T lymphocyte and macrophage activation [10] . In the majority of patients ENL is a chronic condition requiring prolonged immunosuppression [4] . Thalidomide is effective in controlling ENL and is recommended by WHO under strict medical supervision because of its severe teratogenic effects [11] . However it is not available in many leprosy endemic countries including Ethiopia and this means patients have to take large doses of oral corticosteroids often for many years . Patients often require increasing doses of prednisolone due to tachyphylaxis [12] . The clofazimine component of MDT is thought to have a protective effect with respect to ENL but this is lost once MDT is stopped . WHO recommend high dose clofazimine in conjunction with prednisolone in the management of severe ENL [11] but this requires a supply of clofazimine separate to that included in the blister packs of MDT but this is not always available ( personal communication . E . Post ) . A Cochrane review of the treatment of ENL highlighted the paucity of data on which to base treatment decisions and recommended well designed intervention studies [13] . ENL often affects young patients often in their 20 s and 30 s [4] and frequently restricts their ability to work and provide for their families causing financial difficulties . Leprosy workers have long recognised that ENL is associated with a risk of death [14] . There are very few published studies examining the relationship between leprosy reactions and death . A study published in 1963 reported a significantly lower mean age at death in lepromatous patients with “lepra reactions” compared to similar patients without reactions [15] . The authors do not use the term ENL but their description is suggestive of ENL and at the time ENL was considered to be the “classical lepra reaction” [16] . Lepromatous patients with reactions had significantly increased rates of renal disease associated with persistent albuminuria compared to lepromatous patients without reactions . This was often attributed to amyloidosis [15] . The authors do not give any details about the treatments patients received for their leprosy or the “lepra reactions” . An earlier Spanish study reported a higher rate of mortality in patients with reactions compared to those without [17] . There have been no systematic studies of mortality associated with ENL since the introduction of MDT in 1982 . We wished to determine the frequency and causes of mortality associated with ENL at the ALERT Center in Addis Ababa .
ALERT Center in Addis Ababa , Ethiopia is a referral centre for the management of patients with leprosy and other skin diseases . Patients with severe ENL are admitted for control of symptoms . Individuals with milder disease may also be admitted if there are complicating factors . In February 2013 the database of patients admitted to the two dermatology , leprosy and HIV wards was reviewed for a 5 year period between February 2008 and January 2013 . The notes of patients who were admitted , discharged , transferred to another facility or died with a diagnosis of leprosy or leprosy-related complication were retrieved and reviewed using a standard data collection tool . Data were collected on all patients diagnosed with ENL with respect to age , leprosy type , treatment , timing of presentation of ENL , number of episodes of ENL and duration of ENL . Additional details about the final admission of individuals with ENL who had died were also recorded . An episode of ENL was defined as the occurrence of ENL requiring the institution or change of treatment ( such as an increase in dosage or frequency of treatment or the addition of or switching to another drug ) . The nature of ENL was defined as acute for a single episode lasting less than 24 weeks . Recurrent if a patient experienced a second or subsequent episode of ENL occurring 28 days or more after stopping treatment for ENL and chronic if occurring for 24 weeks or more during which a patient has required ENL treatment either continuously or where any treatment free period has been 27 days or less . The data were anonymised , entered in Excel and described using descriptive statistics . The Chi squared test was used to compare differences between groups . The study ( PO09/13 ) was approved by the AHRI/ALERT Ethics Review Committee .
414 individuals were identified from the database and the notes of 312 ( 75 . 4% ) were retrieved . Ninety-nine individuals had been diagnosed with ENL , 147 with T1R , nine patients with neuritis secondary to leprosy , 11 patients had been admitted for a leprosy-related problem not due to a reaction and 46 patients were admitted for a problem other than leprosy . The demographic and clinical features of the patients with ENL are given in Table 1 . There were no significant differences between those individuals with ENL who had died and those who had not . The timing of the occurrence of the first episode was recorded in 98 individuals . Thirty-four ( 34 . 7% ) individuals presented with ENL at the time of their leprosy diagnosis , 39 ( 39 . 8% ) developed ENL during treatment with MDT and 25 ( 25 . 5% ) after having successfully completed a 12 month course of MDT . ENL was acute in 19 ( 19 . 2% ) individuals , recurrent in 10 ( 10 . 1% ) and chronic in 70 ( 70 . 7% ) . The nature of the cutaneous lesions and other organ system involvement is shown in Fig . 1a and b . All patients had cutaneous nodules but pustular , bullous and ulcerated lesions were also seen . ENL-associated neuritis was the most frequently documented extra-cutaneous manifestation . The median number of episodes of ENL experienced was four . The duration of ENL is shown in Fig . 2 . All patients received oral prednisolone with a median starting dose of 60 mg daily . The other drugs that were used in conjunction with prednisolone at some point during the course of ENL were: clofazimine in 61 ( 61 . 2% ) , chloroquine in 6 ( 6 . 1% ) and methotrexate and ciclosporin in one individual each . Three individuals ( 3% ) had a co-morbidity at the time their ENL was diagnosed . One patient had asthma , one strongyloidiasis and the third pulmonary tuberculosis . Following the diagnosis of ENL 50 ( 52 . 1% ) of the remaining 96 individuals had developed aco-morbidity ( Fig . 3 ) . The co-morbidities diagnosed are shown in Fig . 4 and all of them may either be caused or exacerbated by chronic administration of high dose oral corticosteroids . There is an obvious trend in the proportion of individuals with a co-morbidity and the number of ENL episodes experienced ( Fig . 4 ) . Two individuals ( 1 . 4% ) with T1R had died compared with eight ( 8 . 1% ) of those with ENL . This is statistically significant p = 0 . 0168 , ( Fisher's Exact Test ) . All of the patients with ENL who died were HIV negative . Table 2 . gives details of the eight individuals with ENL who died , a brief summary of their clinical course ante-mortem and the cause of death recorded in the notes by the medical staff . In four individuals it was felt that oral corticosteroid therapy was a definite contributory factor in their death and in the remaining four it was possibly contributory . In two individuals it was considered that the cause of death was possibly due to ENL itself . Seven ( 87 . 5% ) of the individuals had chronic ENL which had been present for more than 18 months .
The data from this retrospective study must be interpreted with caution although almost 75% of case notes were available . The data extracted were reliant on the findings recorded by the clinicians at the time the patients were seen . The information about deaths is also reliant on the clinical records as no post-mortem autopsies were performed . This is the first study to report that a significant proportion of Ethiopian patients with ENL are dying and that their deaths appear largely attributable to prolonged treatment with oral corticosteroids . The individuals who succumbed were young and none of the deaths were related to HIV infection . We believe that our data underestimates the mortality associated with ENL and its treatment . The lack of any data on the mortality associated with ENL ( apart from occasional case reports ) [18] since the publication by Brusco and Masanti [15] may be due to better prognosis due to MDT and the use of drugs such as thalidomide and clofazimine . However we believe it is more likely that patients who die are likely to be considered as simply “lost to follow up” or there may be reporting bias due to a reticence on behalf of health workers to report such negative outcomes in patients with ENL . This may give a falsely reassuring picture of ENL . Mortality data from other centres where ENL patients are treated would be useful in further assessing the impact of ENL and understanding the factors that result in patient deaths . The significant difference in the number of deaths in those with ENL compared to individuals with T1R is likely due to the shorter duration of T1R which are usually treated with reducing doses of oral prednisolone over the course of six months [3] . Many patients with T1R also require additional corticosteroids but not for as long as patients with ENL [19] . The adverse effects of prednisolone were examined in the TRIPOD studies which recruited 815 participants and were conducted in Bangladesh and Nepal . These three randomised , double-blind studies examined the role of 16 weeks of prednisolone as: prophylaxis for reactions and neuritis , in the treatment of mild sensory impairment and , in nerve function impairment present for more than six months [20] , [21] , [22] . There were no significant differences in major adverse events between the prednisolone treated and placebo groups [23] . The ENL seen in this cohort is typical of that described by other authors in terms of clinical features , number of episodes and duration [19] , [24] , [25] . Neuritis was the most frequent non-cutaneous manifestation . It is notable that 12 ( 19 . 7% ) men were diagnosed as having orchitis and that fever which is commonly regarded as one of the hallmarks of ENL was only recorded in 12 . 1% of individuals . Forty-seven ( 50 . 5% ) of the 93 patients for whom it was possible to calculate the duration of ENL experienced chronic disease lasting more than 24 months and 13 ( 14% ) had ENL for more than 4 years . This is not surprising given that ALERT Center is a referral hospital and that the methodology of this study relied on a database of patients who had been admitted . However this feature of ENL is described in other cohorts and demonstrates that ENL poses a significant and disproportionate burden on health services . Of the 19 patients who were diagnosed as having acute ENL it is likely that a sizeable proportion may go on to have further episodes and thus become either recurrent or chronic cases as a similar study from India showed that individuals with acute ENL made up only 8% of all ENL cases [4] . The chronic nature of ENL means that patients require long term treatment . There is longstanding experience of the effectiveness of thalidomide in controlling ENL . However there are few alternatives to long term treatment with oral corticosteroids for patients living in places where thalidomide is not available , for those in whom thalidomide is contraindicated or its use is limited by adverse effects and , for those who do not wish to take the drug or cannot afford it . This leads to unacceptably high rates of adverse effects as seen in this cohort in which those patients experiencing more episodes of ENL are more likely to experience severe co-morbidities attributable to corticosteroids . Milder adverse effects of corticosteroids are likely to be underestimated by this study as they are less likely to have been recorded in the patients' case notes . The long term implications for patients with ENL treated with high dose corticostreroids are unclear . Sugumaran reported high rates of adverse effects due to corticosteroids in 249 patients with ENL [26] . He stressed the need to identify agents other than corticosteroids which would be useful in the management of ENL and this sentiment was reiterated by the authors of the Cochrane review [13] and by the participants of an international workshop on ENL [27] . Other drugs may play a role as corticosteroid-sparing agents in ENL or as true alternatives to corticosteroids but it is vital that evidence is gathered to assess their efficacy and safety . There also needs to be wider public debate about the role of thalidomide and how it might be used safely in the management of ENL in those countries where it is not currently available . It is essential that robust evidence-based local guidelines are produced to facilitate the management of leprosy patients with ENL in order to try and minimise adverse outcomes including premature deaths . These guidelines may include the early use of corticosteroid sparing agents and there will need to be an adequate , reliable and affordable supply of such drugs .
|
Erythema nodosum leprosum ( ENL ) is an immunologically mediated complication of leprosy . It is a painful debilitating multi-system condition which can occur before , during or after completion of treatment of the multi-drug therapy for leprosy . ENL is often a chronic condition requiring long-term treatment often with oral corticosteroids or thalidomide . This study shows that in Ethiopia ( where thalidomide is not available ) there is a significant mortality associated with ENL . It appears that much of the mortality is attributable to adverse effects of corticosteroids . This study provides additional evidence of the need for alternative agents to manage ENL particularly when thalidomide is not available , contraindicated or unaffordable .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine"
] |
2014
|
The Mortality Associated with Erythema Nodosum Leprosum in Ethiopia: A Retrospective Hospital-Based Study
|
Deacetylases of the Sir2 family regulate lifespan and response to stress . We have examined the evolutionary history of Sir2 and Hst1 , which arose by gene duplication in budding yeast and which participate in distinct mechanisms of gene repression . In Saccharomyces cerevisiae , Sir2 interacts with the SIR complex to generate long-range silenced chromatin at the cryptic mating-type loci , HMLα and HMRa . Hst1 interacts with the SUM1 complex to repress sporulation genes through a promoter-specific mechanism . We examined the functions of the non-duplicated Sir2 and its partners , Sir4 and Sum1 , in the yeast Kluyveromyces lactis , a species that diverged from Saccharomyces prior to the duplication of Sir2 and Hst1 . KlSir2 interacts with both KlSir4 and KlSum1 and represses the same sets of target genes as ScSir2 and ScHst1 , indicating that Sir2 and Hst1 subfunctionalized after duplication . However , the KlSir4-KlSir2 and KlSum1-KlSir2 complexes do not function as the analogous complexes do in S . cerevisiae . KlSir4 contributes to an extended repressive chromatin only at HMLα and not at HMRa . In contrast , the role of KlSum1 is broader . It employs both long-range and promoter-specific mechanisms to repress cryptic mating-type loci , cell-type–specific genes , and sporulation genes and represents an important regulator of cell identity and the sexual cycle . This study reveals that a single repressive complex can act through two distinct mechanisms to regulate gene expression and illustrates how mechanisms by which regulatory proteins act can change over evolutionary time .
Deacetylases of the Sir2 family are key regulators of lifespan and stress resistance in many organisms ranging from yeast to humans [1] . These enzymes couple deacetylation with hydrolysis of NAD+ and consequently their activity is linked to the metabolic state of the cell [2] . Despite having a well-conserved enzymatic activity , Sir2 family members act on a wide variety of substrates and serve a diverse set of biological functions [3] , [4] . To explore the process by which Sir2 deacetylases have diversified , we examined the evolutionary history of two family members from budding yeast , Sir2 and Hst1 [5] , [6] , which arose in a whole-genome duplication [7] , [8] , [9] , yet have distinct functions . Gene duplication is an important force in evolution because it allows variation to occur without compromising the original function of the gene . Preservation of duplicate genes , or paralogs , is proposed to occur through at least two mechanisms , neofunctionalization and subfunctionalization . In the neofunctionalization model , one duplicate retains the original function , leaving the other gene free of selective constraint and able to evolve a new function [10] . Alternatively , in the subfunctionalization model , if the ancestral gene had multiple functions , duplicated genes could each lose one of the original functions and together retain the entire set of ancestral functions [11] . Only a few studies have characterized the path by which paralogs have diverged [12] , [13] , [14] , [15] . To investigated how Sir2 and Hst1 diverged , we have characterized the function of a representative non-duplicated Sir2 from Kluyveromyces lactis , a budding yeast species that diverged from S . cerevisiae prior to the whole-genome duplication [16] . The functions of Sir2 and Hst1 in S . cerevisiae are well understood . Sir2 interacts with the histone-binding proteins Sir3 and Sir4 , and together these proteins generate an extended silenced domain at the telomeres and cryptic mating-type loci , HMLα and HMRa [17] . The HM loci are flanked by silencers that recruit Sir proteins through DNA binding proteins to initiate the formation of silenced chromatin . The telomere repeats also recruit Sir proteins . Sir2 , Sir3 , and Sir4 spread from sites of recruitment through a sequential deacetylation mechanism that is independent of DNA sequence [18] , [19] , [20] . Sir2 deacetylates nearby nucleosomes , creating high affinity binding sites for Sir3 and Sir4 , which bind preferentially to deacetylated tails of histones H3 and H4 . Sir3 and Sir4 then recruit additional Sir2 to newly deacetylated nucleosomes . As Sir proteins spread , they generate a specialized chromatin structure that is restrictive to transcription . Unlike Sir2 , Hst1 does not spread . It is part of the SUM1 complex that represses over fifty genes that are involved in sporulation , NAD+ biosynthesis , and α-cell identity [21] , [22] , [23] , [24] . Sum1 is a DNA binding protein that associates with a conserved sequence , the middle sporulation element , found in the promoters of target genes [21] , [23] , [25] . Hst1 deacetylates the tails of histones H3 and H4 [26] , [27] , and this deacetylation is thought to be important for its repressive function . The third member of the complex , Rfm1 , mediates the interaction between Sum1 and Hst1 [22] . Genes regulated by Sir2 and Hst1 are critical to cell identity as well as the sexual cycle , and consequently these deacetylases have the potential to coordinate the timing of the life cycle with NAD+ availability . Hst1 plays a role in cell-type identity by repressing several α-specific genes [24] . Hst1 also represses a number of mid-sporulation genes , and this repression must be relieved for completion of the sexual cycle [23] . The mating-type of haploid yeast cells , which can be a or α , is determined by the MAT locus , which encodes transcription factors that regulate cell-type specific genes [28] . These transcription factors are also encoded at HMLα and HMRa , but are silenced by the SIR complex and serve as repositories for mating-type switching . Sir2 maintains cell identity by preventing the cell from simultaneously expressing both a- and α-specific transcription factors . Compared to ScSir2 and ScHst1 , the biological function of the non-duplicated KlSir2 is less understood . KlSir2 is thought to have properties similar to both Sir2 and Hst1 , as it complements both sir2Δ and hst1Δ mutations in S . cerevisiae [26] , [29] . In K . lactis , KlSir2 represses the HM loci [30] , [31] , and a sir2Δ mutation results in reduced mating and sporulation defects [29] . Prior to this study , it was not known whether KlSir2 regulates sporulation genes as ScHst1 does . Few studies have investigated silencing in K . lactis , yet the mechanism differs substantially from that in S . cerevisiae . KlSir2 and the histone binding protein KlSir4 contribute to the silencing of HMLα [30] , [31] . However , there is no distinct Sir3 protein in K . lactis . Additionally , the silencer elements that recruit silencing factors are not conserved between K . lactis and S . cerevisiae [32] . Silencers in S . cerevisiae consist of binding sites for ORC , Rap1 , and Abf1 , whereas in K . lactis , binding sites for these factors have not been identified at the HM loci . Instead the only defined silencer consists of a KlReb1 binding site and two other uncharacterized DNA sequences [32] . In this study , we examined the functions of the non-duplicated KlSir2 and found that it interacts with both KlSir4 and KlSum1 . However , the SIR and SUM1 complexes in K . lactis do not function exactly as the analogous complexes do in S . cerevisiae . The KlSum1-KlSir2 complex contributes to silencing at both HM loci as well as sporulation and cell-type specific genes and achieves repression by both long-range and promoter-specific mechanisms . In contrast , KlSir4 only contributes to silenced chromatin at HMLα , but not at HMRa . This study enhances our understanding of the process by which duplicated genes diverge and provides insights into the connections between promoter-specific and regional silencing .
To determine whether the non-duplicated KlSir2 has functions analogous to both ScSir2 and ScHst1 , we first identified its binding partners in K . lactis ( described in Table 1 ) . If KlSir2 functions similarly to ScSir2 , it should associate with KlSir4 , and if it has a function analogous to ScHst1 it should associate with KlSum1 . Trans-species complementation experiments previously demonstrated that KlSir2 associates with both ScSir4 and ScSum1 in S . cerevisiae [26] , suggesting that analogous interactions occur in K . lactis . We created a K . lactis strain with alleles of KlSIR2-HA , KlSIR4-Flag and myc-KlSUM1 integrated at their chromosomal locations . All three tagged proteins were detectable by immunoblotting ( Figure 1 ) and maintained wild-type function , as assessed by RT-PCR analysis of genes repressed by these proteins ( data not shown ) . If KlSir2 associates with both KlSir4 and KlSum1 , it should co-precipitate with these proteins , and indeed , KlSir2 did co-precipitate with both KlSir4 and KlSum1 ( Figure 1A ) . In S . cerevisiae , the association of ScSum1 with ScHst1 requires ScRfm1 [22] . To determine if Rfm1 mediates the interaction between Sum1 and Sir2 in K . lactis , we examined whether the co-precipitation between KlSir2 and KlSum1 persisted in the absence of KlRfm1 . There was no observable co-precipitation between KlSir2 and KlSum1 in an rfm1Δ strain ( Figure 1A ) , suggesting that the architecture of the SUM1 complex is conserved between S . cerevisiae and K . lactis . Given the association of KlSir2 with both KlSir4 and KlSum1 , all three proteins might be part of a stable complex . However , a co-precipitation between KlSir4 and KlSum1 was not detected ( data not shown ) , although we could not distinguish whether this result reflected the absence of a complex containing KlSir4 and KlSum1 or simply its instability . Nevertheless , if this complex does exist , the components are not mutually dependent on one another for association , as KlSir2 and KlSir4 still co-precipitated in the absence of KlSum1 ( Figure 1B ) and KlSir2 and KlSum1 co-precipitated in the absence of KlSir4 ( Figure 1C ) . Therefore , KlSir2 forms independent associations with both KlSir4 and KlSum1 , a finding consistent with KlSir2 having functions analogous to those of both ScSir2 and ScHst1 . We next investigated whether the Sir4-Sir2 and Sum1-Sir2 complexes have the same repressive functions in K . lactis as they do in S . cerevisiae . If these functions are conserved , deletion of KlSIR4 should derepress the HM loci , deletion of KlSUM1 should derepress mid-sporulation genes , and deletion of KlSIR2 should derepress both HM loci and mid-sporulation genes . We first examined silencing at HMLα , which is known to be repressed by KlSir2 and KlSir4 [30] , [31] . To extend this previous result and address the role of KlSum1 at HMLα , we isolated RNA from MATa wild-type , sir2Δ , sir4Δ , sum1Δ , and rfm1Δ strains and examined the expression of HMLα1 , HMLα2 and HMLα3 by quantitative RT-PCR . All three genes were significantly derepressed in the absence of KlSir2 and modestly derepressed in the absence of KlSir4 ( Figure 2 ) , consistent with previous reports . Surprisingly , deletion of KlSum1 resulted in derepression of HMLα to a similar extent as observed in the sir2Δ strain . In contrast to KlSum1 , deletion of KlRfm1 had very little effect on the transcription of HMLα . This result suggests that KlSir2 does not require KlRfm1 to act at HMLα and therefore may act independently of KlSum1 . In this case , a sir2Δ sum1Δ double deletion might disrupt silencing to a greater extent than either single deletion . However , there was no difference in transcription of HMLα in a sir2Δ sum1Δ strain compared to a sir2Δ or sum1Δ strain ( Figure 2 ) . To confirm that these phenotypes resulted from the deletions of the intended genes , plasmids expressing the wild-type KlSIR2 , KlSIR4 and KlSUM1 genes were introduced into the corresponding deletion strains . In all cases , repression was restored ( data not shown ) . These results reveal that KlSum1 , in addition to KlSir2 and KlSir4 , contributes to the silencing of HMLα . Thus , KlSum1 behaves differently than its ortholog in S . cerevisiae , as the deletion of ScSum1 does not alter the expression of ScHMLα [33] . It is interesting to note that in both the sir2Δ and sum1Δ strains the induction of HMLα3 was modest compared to HMLα1 or HMLα2 , suggesting that HMLα3 may be regulated differently than the other two genes at HMLα . The α3 gene , which is specific to Kluyveromyces , is proposed to be a MULE family DNA transposase [34] and is required for mating [30] . The modest derepression of the HMLα locus observed in the sir4Δ strain suggested that another protein might compensate for KlSir4 in its absence . The SIR4 gene was duplicated in tandem prior to the whole-genome duplication , and each of the tandem duplicates was retained as a single gene after the whole-genome duplication [7] . This ancient duplicate of Sir4 , Asf2 ( Anti-Silencing Factor 2 ) , reduces silencing when over-expressed in S . cerevisiae [35] . The SIR4 and ASF2 genes are rapidly evolving , making it difficult to determine which K . lactis gene is orthologous to which S . cerevisiae gene ( Figure S1 ) . Gene KLLAOF14320g has been designated KlSIR4 based on functional studies [31] , and therefore we refer to the other gene ( KLLA0F13398g ) as KlASF2 . To determine whether its common ancestry with KlSir4 enables KlAsf2 to silence HMLα in the absence of KlSir4 , we constructed both asf2Δ and asf2Δ sir4Δ strains and examined expression of the HMLα genes . The lack of KlAsf2 resulted in the further repression of all three genes to less than one-tenth the level of the wild-type strain , and the double deletion of asf2Δ and sir4Δ resembled the single sir4Δ deletion ( Figure S2 ) . Therefore , KlASF2 does not have a SIR4-like function . In fact , KlASF2 , like ScASF2 , is antagonistic to silencing . Given the surprising result that KlSum1 affects the expression of HMLα , it was important to investigate whether KlSum1 acts directly at HMLα to silence transcription . We also examined the association of KlSir2 and KlSir4 with HMLα , as the association of these proteins with HMLα had not been assessed previously . We used chromatin immunoprecipitation to map the distributions of KlSir2 , KlSir4 , KlSum1 and KlRfm1 across HMLα . We observed a robust enrichment of all four proteins across the entire HMLα locus ( Figure 3A ) , demonstrating that not only KlSir2 and KlSir4 , but also the components of the SUM1 complex , KlSum1 and KlRfm1 , spread across this locus . Therefore , KlSum1 contributes directly to transcriptional silencing at HMLα . The enrichment of KlSir2 , KlSir4 and KlSum1 peaked at a previously identified silencer ( [32] , represented as an aqua bar in Figure 3 ) , suggesting that this sequence may stabilize the association of silencing proteins with chromatin . Three other peaks were also observed ( indicated by asterisks in Figure 3A ) : one in the intergenic region in which the α2 and α3 genes converge , one in the α3 promoter , and a smaller peak on the centromere proximal side of HMLα . These peaks could represent additional silencers or proto-silencers . Curiously , two of the peaks coincided with sequences that are conserved between the transcriptionally silent HMLα locus and the transcriptionally active MATα locus . If these peaks represent binding sites for silencing factors , then these factors might be recruited to MATα . To examine this possibility , we constructed a strain in which the α-cassette at HML was replaced with an a-cassette , so that the only α-cassette in the genome was at the MAT locus . Using this strain , we investigated whether KlSir2 , KlSir4 or KlSum1 associated with the MAT locus . All three proteins associated with control loci ( data not shown ) . However , we observed no significant enrichment of KlSum1 , KlSir2 or KlSir4 anywhere along the MATα locus ( Figure 3B ) . Therefore , the peaks of silencing proteins at the α3 promoter and the α2–α3 intergenic regions are specific to the HMLα locus , and these sequences cannot recruit silencing proteins independently . Sir2 deacetylases lack DNA-binding and histone-binding domains and consequently are recruited to chromatin through adaptor proteins such as Sum1 , a DNA binding protein , or Sir4 , a histone binding protein . To determine whether KlSir4 and/or KlSum1 recruit KlSir2 to HMLα , we examined the association of KlSir2 with HMLα in strains lacking these proteins . In a sir4Δ strain , the enrichments of KlSir2 and KlSum1 were significantly reduced over the silencer and across the open reading frames of α1 , α2 , and α3 ( Figure 3C ) . However , the associations of KlSir2 and KlSum1 with the promoter of α3 and centromere-proximal side of HMLα were unchanged . Thus , there may be different requirements for the assembly of silenced chromatin on the two sides of the HMLα locus . On the telomere-proximal side , containing the known silencer , KlSir4 is important for the recruitment and spreading of silencing proteins . However , on the centromere-proximal side , the recruitment of KlSum1 and KlSir2 is independent of KlSir4 . The ability of KlSir2 to associate with the centromere-proximal side HMLα in the absence of KlSir4 suggests that another protein is recruiting KlSir2 to this region . To determine whether KlSum1 is required for the recruitment or spreading of KlSir2 and KlSir4 , we examined the associations of these proteins with HMLα in a sum1Δ strain . The deletion of KlSum1 caused a reduction in the association of KlSir2 at the α2–α3 intergenic region , the α3 promoter and on the centromere-proximal side of the HMLα locus . There was no observable difference in the association of KlSir4 with HMLα ( Figure 3D ) . These results suggest that KlSum1 is important for stabilizing the association of KlSir2 with the HMLα locus , particularly at the α3 promoter and centromere-proximal regions , but that it is not absolutely required for the recruitment or spreading of either KlSir2 or KlSir4 . Together , these results indicate that neither KlSir4 nor KlSum1 is solely responsible for the recruitment of KlSir2 to HMLα . This finding is consistent with the independent interactions of KlSir2 with KlSir4 and KlSum1 ( Figure 1 ) . The greater level of transcription of HMLα in a sir2Δ strain compared to an rfm1Δ strain ( Figure 2 ) suggests that KlRfm1 is not critical for the recruitment of KlSir2 or other silencing proteins . In fact , in the absence of KlRfm1 , all three silencing proteins , KlSir2 , KlSir4 and KlSum1 , still associated with the entire HMLα locus ( Figure 3E ) . The enrichment of KlSum1 was indistinguishable between the wild-type and rfm1Δ strains , indicating that its association with HMLα does not require KlRfm1 and may be an inherent property of the Sum1 protein . Interestingly , the enrichments of both KlSir2 and KlSir4 were significantly enhanced in the rfm1Δ strain compared to the wild-type strain , although the overall pattern , with peaks of association at the silencer , α2–α3 intergenic region , α3 promoter and centromere-proximal side of HMLα , was maintained . Perhaps in the absence of KlRfm1 , KlSir2 is better able to associate with KlSir4 . In S . cerevisiae , the deacetylase activity of Sir2 is required for the spreading of Sir3 and Sir4 [18] , [19] , [20] . To determine whether a similar requirement exists in K . lactis , we examined the associations of KlSir4 and KlSum1 with HMLα in a sir2Δ strain . KlSir4 and KlSum1 were reduced over the silencer and the three open reading frames ( Figure 3F ) . However , both silencing proteins remained strongly associated with the α3 promoter , and KlSir4 displayed a more robust enrichment with this region in the absence of KlSir2 . This pattern of association is similar to the distribution of KlSum1 and KlSir2 in the sir4Δ strain ( Figure 3C ) . Therefore , KlSir2 may contribute to the assembly of silenced chromatin on the telomere-proximal side of HMLα , but it is not required to assemble these factors at the α3 promoter . Given that KlSum1 is a DNA-binding protein , we were curious whether it binds directly to a sequence at HMLα . The mid-sporulation element ( MSE ) consensus sequence , to which Sum1 binds in S . cerevisiae , appears to be conserved in K . lactis , as it occurs at the promoters of a number of sporulation genes ( data not shown ) . However , a match to the MSE consensus sequence was not found in the known telomere-proximal silencer ( aqua box ) or the rest of the HMLα locus . Moreover , the observation that the enrichment of KlSum1 was significantly reduced on the telomere-proximal side of HMLα in the absence of KlSir4 or KlSir2 ( Figure 3C and 3F ) makes it unlikely that KlSum1 binds directly to this side of the locus . Furthermore , KlSum1 did not associate with the MATα locus ( Figure 3B ) , indicating that the sequences conserved between MATα and HMLα are unable to recruit KlSum1 directly . It remains possible that KlSum1 binds directly to a non-MSE sequence on the centromere-proximal side of the HMLα , and KlSum1 did associate with this region of HMLα in the absence of both KlSir2 and KlSir4 ( Figure S3 ) , indicating that the recruitment of KlSum1 to HMLα is independent of KlSir2 and KlSir4 . However , it is also possible that another , unidentified protein recruits KlSum1 to this region . We next investigated the roles KlSir2 , KlSir4 , KlSum1 and KlRfm1 have in regulating the other cryptic mating-type locus , HMRa . In S . cerevisiae , both HM loci are silenced by the same set of Sir proteins . However , in K . lactis , deletion of KlSir4 had little effect on the expression of the a1 or a2 genes found at HMRa ( Figure 4A ) . Furthermore , deletion of KlAsf2 , the paralog of KlSir4 , either singly or in conjunction with KlSir4 did not result in derepression of HMRa ( Figure S4 ) . In contrast , deletion of KlSir2 or KlSum1 resulted in a substantial derepression of HMRa1 and HMRa2 , whereas deletion of KlRfm1 resulted in very little change in HMRa1 or HMRa2 expression ( Figure 4A ) . These results suggest that only a subset of the proteins that contribute to the silencing of HMLα also repress HMRa . To determine whether KlSir2 and KlSum1 act directly at HMRa , we examined their association by chromatin immunoprecipitation . We observed an asymmetric distribution of KlSir2 and KlSum1 , as well as KlRfm1 , with the HMRa locus . A substantial peak of enrichment was observed on the centromere-proximal side of HMRa , and a shoulder extended across the open reading frames ( Figure 4B ) . The peak likely indicates the location of a silencer element . In contrast to KlSir2 and KlSum1 , there was no significant association of KlSir4 with any part of HMRa , consistent with the deletion of SIR4 resulting in no change in the transcription of HMRa1 and HMRa2 . These results indicate that KlSum1 and KlSir2 , but not KlSir4 , are responsible for repressing HMRa . Thus , the mechanisms of silencing at HMRa and HMLα are distinct . Curiously , KlRfm1 associated with HMRa ( Figure 4B ) , yet was not required for repression of the HMRa1 and HMRa2 genes ( Figure 4A ) . We examined the association of KlSum1 and KlSir2 with HMRa in a rfm1Δ strain and found that KlSum1 was only slightly reduced at the proposed silencer ( Figure 4C ) . Intriguingly , KlSir2 was still able to associate with HMRa in the absence of KlRfm1 , despite the fact that it no longer co-precipitated with KlSum1 ( Figure 1A ) . We propose that the absence of KlRfm1 may enable KlSir4 to interact with KlSir2 and KlSum1 , thereby stabilizing the association of KlSir2 with HMRa . To test this hypothesis , we assessed whether KlSir4 associated with HMRa in an rfm1Δ strain , and indeed , KlSir4 associated with HMRa ( Figure 4C ) . This result is reminiscent of the increase in KlSir4 at HMLα in the absence of KlRfm1 ( Figure 3E ) . To determine whether KlSum1 and KlSir2 depended on one another for association with HMRa , we performed chromatin immunoprecipitation experiments in the absence of KlSum1 or KlSir2 . In the absence of KlSum1 , KlSir2 no longer associated with any region of the HMRa locus ( Figure 4D ) , and therefore KlSum1 was required for recruitment of KlSir2 to HMRa . This result contrasts with what was observed in the absence of KlRfm1 ( Figure 4C ) . Deletion of KlSir2 , like deletion of KlRfm1 , resulted in a reduced association of KlSum1 with the proposed silencer at HMRa . Despite this reduction , KlSum1 still spread across HMRa ( Figure 4D ) . Thus , the association and spreading of KlSum1 does not require KlSir2 or KlRfm1 . In S . cerevisiae , the Sum1-Hst1 complex represses mid-sporulation genes . To assess whether KlSir2 regulates mid-sporulation genes in a manner similar to ScHst1 , we isolated RNA from wild-type , sir2Δ , sum1Δ and rfm1Δ strains and examined expression of the K . lactis orthologs of the mid-sporulation genes CDA2 , SPR3 , SPS4 , and SPS2 that are repressed by ScHst1 in S . cerevisiae [23] . Deletion of KlSir2 , KlSum1 and KlRfm1 all resulted in derepression of CDA2 , SPS4 , and SPR3 , but not SPS2 ( Figure 5A , note the different scales of the x-axes ) . We also examined whether KlSir4 has a role in regulating transcription of these genes , as KlSir2 and KlSum1 functioned with KlSir4 to regulate HMLα . However , the sir4Δ strain had no effect on the expression of CDA2 , SPS4 , SPR3 or SPS2 ( Figure 5A ) . Therefore , KlSum1 , KlSir2 and KlRfm1 , repress sporulation genes independently of KlSir4 . In addition , many ( CDA2 , SPS4 and SPR3 ) , but not all ( SPS2 ) of the targets of the Sum1-Hst1 complex in S . cerevisiae are also targets in K . lactis . To determine if KlSir2 , KlSum1 , and KlRfm1 repress mid-sporulation genes directly , we used chromatin immunoprecipitation to assess the association of KlSir2 , KlSum1 , KlRfm1 and KlSir4 with the promoters of these genes . KlSir2 , KlSum1 and KlRfm1 were enriched at the promoters of CDA2 , SPS4 and SPR3 ( Figure 5B ) , suggesting that these proteins repress these genes directly , presumably as a complex . In contrast , KlSir4 did not associate with mid-sporulation genes , consistent with the sir4Δ strain having no effect on transcription . To address whether KlSir2 , KlSum1 and KlRfm1 spread at sporulation genes , as they do at HMLα and HMRa , we examined a 3-kb region around the CDA2 promoter and open reading frame . A relatively narrow peak of KlSum1 , KlRfm1 and KlSir2 coincided with an MSE consensus sequence at the promoter of CDA2 ( indicated by the blue bar in the schematic ) , and the association of these proteins diminished significantly in both directions ( Figure 5C ) , suggesting that these proteins do not spread at the CDA2 locus . Therefore , the ability of the SUM1 complex to spread differs between the HM loci and mid-sporulation genes . We had observed at HMLα that KlAsf2 was antagonistic to silencing ( Figure S2 ) , and it was possible that KlAsf2 restricts the spreading of the Sum1-Sir2 complex at sporulation genes and therefore accounts for the difference in spreading at HMRa compared to sporulation genes . To test this hypothesis , we assessed the distribution of KlSum1 and KlSir2 at the sporulation gene CDA2 in an asf2Δ strain . We observed no changes in the distribution of KlSir2 and KlSum1 across the CDA2 locus ( Figure S5A ) . Furthermore , the transcription of several mid-sporulation genes was not altered ( Figure S5B ) . Therefore , KlAsf2 only antagonized silencing at HMLα . We discovered that KlSir2 was more dependent on KlRfm1 for recruitment to CDA2 as compared to HMRa . At HMRa , KlSir2 required KlSum1 but not KlRfm1 for recruitment ( Figure 4C and 4D ) . In contrast , the association of KlSir2 with CDA2 was greatly reduced in both sum1Δ and rfm1Δ strains ( Figure 5D and 5E ) . This dependence was similar to what has been observed for the S . cerevisiae SUM1 complex at mid-sporulation genes . One potential explanation for the reduced role of KlRfm1 at the HM loci is the ability of KlSir4 to compensate for the loss of KlRfm1 . For example , at both HMLα and HMRa , the association of KlSir4 increased in the absence of KlRfm1 ( Figure 3E and Figure 4C ) . In keeping with the greater role of KlRfm1 at CDA2 , we observed only a modest increase in the association of KlSir4 ( Figure 5D ) in the absence of KlRfm1 . We also found that the ability of KlSum1 to associate with the promoter of CDA2 was unaltered in the absence of KlSir2 ( Figure 5D ) , and was reduced , but not abolished , in the absence of KlRfm1 ( Figure 5E ) . Thus , KlRfm1 contributes to the ability of the SUM1 complex to associate with DNA . We conclude that the promoter-specific mechanism by which the SUM1 complex represses mid-sporulation genes is conserved between K . lactis and S . cerevisiae . The KlSum1-KlSir2 complex is clearly critical to the regulation of sexual identity and the sexual cycle as it represses both the HM loci and sporulation genes . However , the Sum1-Sir2 complex may have an even broader role in controlling sexual identity . It has recently been shown in both Saccharomyces bayanus and S . cerevisiae that Sum1 represses α-specific genes [24] . To investigate whether the Sum1-Sir2 complex in K . lactis also represses α-specific genes or other cell-type specific genes , we examined whether promoters of cell-type specific genes were associated with KlSir2 . Remarkably some , but not all , α-specific , a-specific and haploid-specific genes were associated with KlSir2 ( Figure 6A and data not shown ) . For example , the α-specific gene MFα1 , the a-specific gene BAR1 , and the haploid-specific gene STE18 were associated with KlSir2 , KlSum1 , and KlRfm1 , but not KlSir4 ( Figure 6A ) . To determine whether the Sum1-Sir2 complex represses these genes , RNA was isolated from both MATa and MATα cells and expression of MFα1 , STE18 , and BAR1 was examined by quantitative RT-PCR . MFα1 encodes α-pheromone and in S . cerevisiae is expressed in MATα cells but not in MATa cells . However in K . lactis , deletion of KlSum1 or KlSir2 resulted in the derepression of MFα1 in both cell types to a comparable extent ( Figure 6B ) . Quantification of cDNA from wild-type cells revealed that MFα1 was repressed to a similar degree in both MATa and MATα cells ( Figure S6 ) . These findings suggest that during vegetative growth , haploid K . lactis cells are not transcribing or producing α-pheromone , regardless of their mating-type identity , and that the Sum1-Sir2 complex contributes to the repression of this gene . STE18 encodes the G protein gamma subunit in the mating signaling pathway and in S . cerevisiae is expressed in both MATα and MATa haploid cells . In K . lactis , STE18 , like MFα1 , was repressed in both MATα and MATa cells ( Figure S6 ) , and deletion of either KlSir2 or KlSum1 resulted in derepression of STE18 in both cell types ( Figure 6C ) . BAR1 encodes an α-pheromone protease that in S . cerevisiae is expressed to a greater extent in MATa than MATα cells . This pattern of gene expression was also found in K . lactis ( Figure S6 ) . However , as for MFα1 and STE18 , deletion of KlSum1 or KlSir2 resulted in the derepression of BAR1 in both MATa and MATα cells ( Figure 6D ) . To verify that we had correctly identified the mating-type of the strains used for these experiments , we analyzed a segment of the MAT locus using mating-type specific PCR primers that yield different sized products in MATa and MATα strains . All strains had the expected genotypes ( Figure 6E ) . Together , these results suggest that the KlSum1-KlSir2 complex represses a variety of cell-type specific genes as well as mid-sporulation genes and the HM loci . Therefore , this complex represents an important regulator of yeast sexual identity and activity .
This study has made the striking discovery that the Sum1-Sir2 complex in K . lactis achieves repression through several distinct mechanisms ( Figure 7 ) . In S . cerevisiae , the Sum1-Hst1 complex functions primarily as a promoter-specific repressor of mid-sporulation , α-specific , and NAD+-biosynthetic genes , and loss of ScSum1 or ScHst1 do not alter the expression of the HM loci [6] , [33] . In contrast , in K . lactis , the Sum1-Sir2 complex not only uses a promoter-specific mechanism to repress the same sets of genes as in S . cerevisiae ( Figure 7 , top panel ) , it also has a major role in silencing the HM loci by forming extended chromatin structures ( Figure 7 , middle and lower panels ) . Interestingly , the KlSum1-KlSir2 complex acts differently at HMLα ( lower panel ) , where it works in conjunction with KlSir4 , compared to HMRa ( middle panel ) , where KlSir4 is not normally present . Thus , the mechanism by which HMRa is silenced is unlike the mechanism employed at HMLα . The absence of KlSir4 at HMRa is surprising , as the spreading of silencing proteins is thought to require a histone-binding protein , such as KlSir4 , and neither KlSum1 nor KlSir2 is known to have this capacity . An important subject for future studies will be to determine how the spreading capacity of the KlSum1-KlSir2 complex is modulated at different genomic locations . It is possible that factors associated with the HM loci promote the spreading of KlSum1-KlSir2 . For example , silencers may recruit additional proteins that facilitate the spreading process . We have recently found that the HMR-E silencer in S . cerevisiae can promote the assembly of silenced chromatin through a mechanism that is independent of recruitment [36] , and it is possible that silencers in K . lactis have similar properties . Alternatively , factors associated with the promoters of mid-sporulation genes may limit or disable the spreading of KlSum1-KlSir2 . This study also revealed that , although the KlSum1-KlSir2 and KlSir4-KlSir2 complexes cooperate at HMLα , they have distinct contributions to chromatin assembly and transcriptional repression . For example , the KlSir4-KlSir2 complex was critical for assembly of silencing proteins on the telomere proximal side of HMLα . However , silenced chromatin on the centromere-proximal side did not depend on KlSir2 or KlSir4 , but was affected by the loss of KlSum1 . These results suggest that the chromatin structure differs on the two sides of HMLα , perhaps due to different types of silencer elements . Another indication that the KlSum1-KlSir2 and KlSir4-KlSir2 complexes have independent properties is the observation that the associations of KlSir4 and KlSir2 increased at HMLα and HMRa in the absence of KlRfm1 . This result suggests that KlSir4 and KlRfm1 may compete for association with KlSir2 . One puzzling observation was that the absence of KlSir4 resulted in a relatively modest induction of the HMLα1 and HMLα2 genes despite a significant decrease in the associations of both KlSir2 and KlSum1 with the α1–α2 promoter . Conversely , the absence of KlSum1 resulted in a large increase of transcriptional activity yet had seemingly little effect on the associations of KlSir2 and KlSir4 with HMLα . These results are reminiscent of observations that , in some situations , Sir proteins in S . cerevisiae associate with HM loci but do not achieve repression [37] , [38] . We speculate that the presence of the KlSum1-KlSir2 complex at HMLα is more critical for repression than is the presence of KlSir4 . Moreover , KlSum1 and KlSir2 must be able to achieve repression over a distance , because their presence at the HMLα3 promoter is sufficient to repress the HMLα1 and HMLα2 genes . Similarly , KlSum1 and KlSir2 may act at distance at HMRa , as their greatest enrichment is some distance from the promoter . In contrast , the KlSir4-KlSir2 complex appears to be somewhat permissive to transcription in the absence of KlSum1 . Perhaps this chromatin structure serves another biological function , such as preventing illegitimate mating-type switching . While K . lactis is considered to be a homothallic yeast species [39] , an ortholog of the HO endonuclease , which initiates switching in S . cerevisiae , has not been identified [40] , and mating-type switching presumably occurs through spontaneous homologous recombination . These switching events are relatively rare had have not been studied recently [39] . This study was initiated to investigate how the deacetylases SIR2 and HST1 diverged after duplication . Two models , subfunctionalization and neofunctionalization , have been proposed to explain how duplicated genes diverge . We used the non-duplicated KlSir2 as a proxy for the ancestral protein and found that it interacted with both KlSir4 and KlSum1 ( Figure 1 ) , the partners of ScSir2 and ScHst1 , respectively . Furthermore , KlSir2 functioned as a promoter-specific repressor of sporulation genes ( similar to ScHst1; Figure 5 ) and also as a silencing factor that spreads across the HM loci ( similar to ScSir2; Figure 2 , Figure 3 , Figure 4 ) . Therefore , KlSir2 has both Hst1- and Sir2-like functions . The most parsimonious interpretation of these results is that the ancestral deacetylase also had both functions and that subfunctionalization occurred after duplication . This conclusion is supported by the observation that ScSir2 has retained the ability to substitute for ScHst1 in its absence [26] . This is an important contribution to the understanding of the evolution of duplicated genes , as it provides an example of subfunctionalization of protein-protein interactions as opposed to partitioning of expression patterns , which have previously been documented [41] . Previous work provides insight into how the subfunctionalization of SIR2 and HST1 occurred . A chimeric protein consisting of the N-terminus of ScSir2 and the C-terminus of ScHst1 has both Sir2- and Hst1-like functions in S . cerevisiae [26] , [42] . This observation suggests that different regions of the deacetylases are important for specifying interactions with the SIR and SUM1 complexes . It is likely that the ancestral deacetylase used these same domains to interact with the SIR and SUM1 complexes . After SIR2 was duplicated , the two copies likely acquired mutations that reduced their affinities for either the SIR or SUM1 complexes , leading to subfunctionalization . Over the course of evolution it was not simply the deacetylase that subfunctionalized . The proteins associated with Sir2 and Hst1 are used in different ways to achieve repression of essentially the same sets of genes in S . cerevisiae and K . lactis . Other studies have revealed changes in the transcriptional regulatory circuits of yeasts [13] , [43] , [44] . However in previous examples , evidence suggested that promoter elements have changed to bring genes under the control of different regulators or alter their expression patterns . This study expands the scope of adaptations that can lead to modifications in transcriptional networks , as it reveals that the molecular mechanisms by which regulatory proteins act can also change over evolutionary time . In addition to the paralogs SIR2 and HST1 , we investigated a second duplicated gene pair , SIR4 and ASF2 . SIR4 and ASF2 were tandemly duplicated prior to the whole genome duplication and to the divergence of Kluyveromyces and Saccharomyces species . Due to their tandem arrangement and rapid rate of sequence change , it has been difficult to determine which gene is the ortholog of ScSIR4 or ScASF2 . Functional analysis shows that KLLA0F14320g silences HMLα ( Figure 2 , Figure 3 , and [31] ) as thus has a Sir4-like function , whereas KLLA0F13998g antagonizes silencing at HMLα ( Figure S2 ) and thus has Asf2-like function . This experimental evidence seems to contradict phylogenetic analyses implying that KLLA0F13998g is the ortholog of ScSIR4 , as it clusters with SIR4 genes from other yeast species , and that KLLA0F13420g is an ortholog of ScASF2 , as it clusters with ASF2 genes as well as SIR4 genes from Candida glabrata , S . castellii , S . kluyveri and Ashbya gossypii ( Figure S1 and [7] ) . However , this gene tree does not match the species phylogeny , perhaps due to the rapid rate of sequence change and consequently may not accurately reflect the evolutionary relationships among these genes . The observation that KlSum1 spreads at the HM loci provides a new perspective on the perplexing SUM1-1 mutation identified in S . cerevisiae . This mutation was originally isolated as a suppressor of a sir2Δ mutation [45] and results from a single point mutation , T988I . It causes Sum1 to re-localize from mid-sporulation promoters to the HM loci and form an extended chromatin structure [46] , [47] . It had been thought that the SUM1-1 mutation is a gain-of-function mutation that creates the ability to spread de novo , and it was surprising that a single amino acid change could have such a profound effect . However , this study suggests a new interpretation . The ability of both KlSum1 and ScSum1-1 to spread at HM loci suggests that the ancestral Sum1 also had this ability , which was subsequently lost in the Saccharomyces lineage . Consequently , wild-type ScSum1 probably retains most of the properties necessary to spread , and the T988I mutation unmasks this hidden potential . Our knowledge of the mechanism of the SUM1-1 mutation may provide insights into how the spreading of KlSum1 is controlled . Residue T988 of ScSum1 is conserved in KlSum1 , as well as in many other budding yeasts , and is located in the DNA-binding domain . Mutating this residue reduces the affinity of Sum1 for DNA [48] and replacing threonine 988 with isoleucine enables the protein to associate with new partners - ORC ( the Origin Recognition Complex ) and itself [47] , [48] , [49] . These observations led to the hypothesis that the SUM1-1 mutation occurs in an interaction domain , and the switch between threonine and isoleucine causes the protein to interact with different partners [48] . Perhaps this domain of KlSum1 also has the capacity to interact with multiple partners , and the genomic context dictates whether this surface functions as a DNA-binding domain to recruit the Sum1-Sir2 complex to mid-sporulation genes or as a self-associating surface to enable KlSum1 to propagate along the chromatin at the HM loci . The K . lactis Sum1-Sir2 complex plays a critical role as a regulator of sexual identity because it regulates some cell-type specific genes ( Figure 6 ) . Within budding yeasts there has been a transition from positive to negative regulation of a-specific genes . Candida albicans requires an activator to turn on a-specific genes in MATa cells , whereas in S . cerevisiae , a-specific genes are on by default and must be turned off in MATα cells [50] . K . lactis has been proposed to have an intermediate circuitry in regulating cell-type identity [43] , as a-specific gene promoters share features of both C . albicans and S . cerevisiae promoters . In this study we have demonstrated that many cell-type specific genes , including a- and α-specific genes are repressed by the KlSum1-KlSir2 complex in both haploid cell types providing an additional level of regulation to sexual identity . Differences between the life cycles of K . lactis and S . cerevisiae may heighten the importance of the Sum1-Sir2 complex in K . lactis . Vegetative growth of K . lactis occurs predominantly in the haploid phase , and mating occurs in response to nutrient deprivation , leading almost immediately to sporulation [39] , [51] , [52] . In contrast , S . cerevisiae propagates primarily in the diploid phase . Mating occurs shortly after germination in rich nutrient conditions , but sporulation of the resulting diploid cells is delayed until nutrients become scarce . Thus , unlike S . cerevisiae , K . lactis requires a mechanism to suppress mating of haploid cells under nutrient-rich conditions , and perhaps the Sum1-Sir2 complex contributes to this regulation by repressing some of the α-specific , a-specific , and haploid-specific genes required for mating . The use of a repressive complex containing an NAD+-dependent deacetylase may help connect the sexual cycle of K . lactis with nutrient availability .
All K . lactis strains used in this study were grown at 30° in YPD medium containing 1% yeast extract , 2% peptone and 2% glucose . Antibiotic supplements were added to YPD medium at 50 µg/ml of clonNAT and 200 µg/ml of geneticin . Electroporation conditions were as described [53] with the following changes . Cells were washed with LiAc buffer ( 10 mM Tris pH 7 . 5 , 270 mM sucrose , 1 mM lithium acetate ) after initial centrifugation . After treatment with the pre-treating buffer ( YPD , 20 mM HEPES pH 8 . 0 , 25 mM DTT ) , cells were resuspended in LiAc buffer to a final concentration of 2×109 cells/ml and electroporation was performed in a 0 . 2 cm cuvette , with a final at volume between 50 and 55 µl . The settings for electroporation were 1 , 000 V , 25 µF and 300 Ω . Cells transformed with antibiotic resistance markers were grown at 30° in YPD for 3–5 hours before being plated on selective medium . Mating was carried out by mixing equal volumes of overnight cultures of the two parental strains , plating 4–10 µl on malt extract ( ME ) medium ( 2% malt extract , 2% agar ) and incubating at 30° for 2–3 days . Cells were then streaked on media to select for diploids and subsequently transferred to ME plates for sporulation . After 3–4 days , the sporulated culture was suspended in 500 µl water , incubated at 56° for 15 minutes , and plated on media to select for alleles of interest . Genotypes were confirmed by PCR . Strains used in this study were derived from SAY538 ( Table S1 ) . The sir2Δ::KanMX allele was obtained from S . Astrom . The sir2Δ::NatMX , sir4Δ::URA3 , asf2Δ::NatMX , sir4Δ asf2Δ::URA3 , sum1Δ::NatMX and rfm1Δ::URA3 alleles were complete deletions of the open reading frames generated by one-step gene replacement . The replacement markers NatMX and URA3 were derived from pAGT100 [54] and pRS316 [55] , respectively . The HMLa allele was a fortuitous gene conversion event that occurred during the course of crossing a sir2Δ strain . The SIR2-HA , RFM1-HA and SIR4-Flag alleles were constructed by integrating the tag plus a selectable marker at the end of the open reading frame . Tagging cassettes were generated from pAGT105 [54] containing the HA-epitope tag along with the entire open reading frame of NatMX or p3FLAG-KanMX , [56] containing the Flag tag plus KanMX . The myc-SUM1 allele was generated in two steps . First , a myc-URA3-myc-SUM1 construct , derived from p3MPY-3xMyc , [57] was integrated into the K . lactis genome . After correct integration was confirmed by PCR , cells were grown in non-selective media to allow for recombination between the identical myc-tags and cells were plated on 5-FOA to select for the loss of the URA3 marker . In all cases , the correct integration was confirmed by PCR using primers flanking the sites of recombination . To confirm that the tagged proteins were functional , expression of genes regulated by these factors was examined by quantitative RT-PCR . Alleles were moved into various genetic backgrounds ( as described in Table S1 ) through genetic crosses . RNA was isolated from logarithmically growing cultures of each strain using a hot phenol method [58] . Removal of DNA was as previously described [26] . To verify that there was no contaminating DNA , 1 µl of DNAse-treated material was used in a PCR reaction containing primers to amplify the KlACT1 transcript . 1 µg of DNA-free RNA was used for cDNA synthesis as previously described [26] . To quantify the relative amounts of mRNA transcripts , approximately 0 . 025 µg of cDNA was analyzed by real-time PCR in the presence of SYBR Green using a Bio-Rad iCycler . The standard curve was generated with genomic DNA isolated from the wild-type strain ( SAY538 ) . Oligonucleotide sequences are provided in Table S2 . Data were analyzed with iCycler iQ Optical System Software . Transcript levels of queried genes were first normalized to the KlACT1 mRNA for each genetic background . The fold-induction was calculated by normalizing to the wild-type strain . Results represent the average fold induction ( relative to wild-type ) of at least two independent cultures of each strain background . The standard error measurement ( SEM ) was calculated from the differences in fold induction of two or more independent cultures from the mean . Chromatin immunoprecipitation was performed by harvesting approximately 50 OD ( 7×108 ) of logarithmically growing cells , collected at an OD600 = 1 . 4 . Cells were collected , washed twice in PBS , re-suspended in DMA ( 10 mM dimethyl adipimidate , 0 . 1% DMSO , 1× PBS ) and rocked at room temperature for 60 minutes to crosslink . Subsequent to crosslinking , cells were washed twice with PBS , re-suspended in 36 ml PBS and rocked with 1% formaldehyde at room temperature for 60 minutes . The preparation of soluble chromatin and immunoprecipitation was performed as previously described [26] . Chromatin IP samples were analyzed by qPCR using a standard curve prepared from input DNA . The amounts of the immunoprecipitated DNA at experimental loci and a control locus , KlRRP7 , were determined relative to the input DNA , and the relative enrichment of the experimental loci compared to the control locus was calculated . Oligonucleotide sequences are provided in Figure S7 and Table S3 . Results represent the relative immunoprecipitation of two or more independent cultures of each strain background , and the SEM was calculated from differences in the relative enrichment from the mean . No antibody control data represent the average values from multiple chromatin IP experiments using different strains . Co-immunoprecipitations were performed by harvesting approximately 30 OD ( 4 . 2×108 ) of logarithmically growing cells . The preparation of whole-cell lysates was performed as previously described [26] . Whole-cell lysates were incubated overnight at 4°C with 5 µl of α-HA ( Sigma H-6908 ) , α-Flag ( Sigma F-7425 ) or α-myc ( Millipore 06-549 ) antibody . Subsequently , 60 µl of Protein A agarose beads were added and samples were rotated at 4° overnight and protein was eluted in 75 µl 3× protein sample buffer ( 30% glycerol , 15% β-mercaptoethanol , 0 . 006% bromophenol blue , 0 . 1875 M Tris pH 6 . 8 ) for 3 minutes at 95° . 20 µl of IP samples and 7 . 5 µl of whole-cell extracts were electrophoretically fractionated on 7 . 5% polyacrylamide-SDS gels , transferred to nitro-cellulose membranes , and probed with either mouse polyclonal α-HA antibody ( Sigma H-3663 ) , mouse polyclonal α-myc antibody ( Calbiochem OP10 ) , rabbit ( Sigma F-7425 ) or mouse ( Sigma F-3165 ) α-Flag antibodies and detected by chemiluminescence ( GE RPN2135 ) .
|
Sir2 deacetylases are found in organisms ranging from bacteria to mammals . Sir2 from the yeast Saccharomyces cerevisiae deacetylates histones and is part of the SIR complex that spreads across chromatin to repress gene expression . A related histone deacetylase , Hst1 , interacts with a DNA–binding protein , Sum1 , to repress genes in a promoter-specific manner . Hst1 and Sir2 are paralogs , arising from a duplication about 100 million years ago . To understand how Sir2 and Hst1 have diverged , as well as to investigate the evolutionary relationship between spreading and non-spreading mechanisms of gene repression , we have characterized the function of a non-duplicated Sir2 from the yeast Kluyveromyces lactis , a species that diverged from Saccharomyces prior to this duplication . We found that KlSir2 is part of both the SIR and SUM1 complexes , indicating that the ancestral Sir2 had both Sir2- and Hst1-like properties . Interestingly , we found that , in K . lactis , the Sir2-Sum1 complex not only uses a promoter-specific mechanism to repress the same sets of genes as S . cerevisiae , it also forms extended chromatin structures to repress gene transcription . Our results illustrate how mechanisms by which regulatory proteins act can change over evolutionary time .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"molecular",
"biology/molecular",
"evolution",
"molecular",
"biology/chromatin",
"structure"
] |
2009
|
The Sir2-Sum1 Complex Represses Transcription Using Both Promoter-Specific and Long-Range Mechanisms to Regulate Cell Identity and Sexual Cycle in the Yeast Kluyveromyces lactis
|
The family Flaviviridae includes viruses that have different virion structures and morphogenesis mechanisms . Most cellular and molecular studies have been so far performed with viruses of the Hepacivirus and Flavivirus genera . Here , we studied bovine viral diarrhea virus ( BVDV ) , a member of the Pestivirus genus . We set up a method to purify BVDV virions and analyzed their morphology by electron microscopy and their protein and lipid composition by mass spectrometry . Cryo-electron microscopy showed near spherical viral particles displaying an electron-dense capsid surrounded by a phospholipid bilayer with no visible spikes . Most particles had a diameter of 50 nm and about 2% were larger with a diameter of up to 65 nm , suggesting some size flexibility during BVDV morphogenesis . Morphological and biochemical data suggested a low envelope glycoprotein content of BVDV particles , E1 and E2 being apparently less abundant than Erns . Lipid content of BVDV particles displayed a ~2 . 3 to 3 . 5-fold enrichment in cholesterol , sphingomyelin and hexosyl-ceramide , concomitant with a 1 . 5 to 5-fold reduction of all glycerophospholipid classes , as compared to lipid content of MDBK cells . Although BVDV buds in the endoplasmic reticulum , its lipid content differs from a typical endoplasmic reticulum membrane composition . This suggests that BVDV morphogenesis includes a mechanism of lipid sorting . Functional analyses confirmed the importance of cholesterol and sphingomyelin for BVDV entry . Surprisingly , despite a high cholesterol and sphingolipid content of BVDV envelope , E2 was not found in detergent-resistant membranes . Our results indicate that there are differences between the structure and molecular composition of viral particles of Flaviviruses , Pestiviruses and Hepaciviruses within the Flaviviridae family .
The Flaviviridae family includes important human and animal pathogens . Members of this family are enveloped , positive-stranded RNA viruses that share similarities in replication and genome organization . They have been classified into 4 genera , namely Flavivirus , Hepacivirus , Pestivirus and Pegivirus . The Flavivirus genus consists of a large number of arthropod-borne viruses . The Hepacivirus genus includes hepatitis C virus ( HCV ) and recently identified closely related viruses . Members of the Pestivirus genus are animal pathogens including bovine viral diarrhea virus ( BVDV ) , classical swine fever virus ( CSFV ) and border disease virus ( BDV ) of sheep [1] . The Pegivirus genus contains a few HCV-related viruses , formerly known as GB-viruses . Despite similarities in genome organization and replication mechanisms , members of this family have very different modes of transmission: most flaviviruses are transmitted by mosquitoes or ticks , while the mode of transmission of pestiviruses can be oro-nasal and diaplacental and that of HCV is parenteral . These differences in transmission mode are mirrored by differences in the structure of infectious particles and envelope proteins of flaviviruses and hepaciviruses . Flavivirus infectious virions are ~50 nm particles fully coated with 90 dimers of class II envelope protein E [2 , 3] . In contrast , HCV particles have been proposed to display a lipoprotein-like structure [4 , 5] containing high amounts of apolipoproteins , especially apoE [6 , 7] . These different structural organizations denote differences in morphogenesis mechanisms . Envelope proteins appear to be a major driving force of flaviviruses morphogenesis , yielding both infectious particles and non-infectious , capsid-less , sub-viral particles . In contrast , the release of sub-viral particles from pestivirus-infected cells has not been reported . Capsid-less particle release by HCV-infected hepatocytes has been reported [8] . However this most probably reflects the incorporation of E1E2 envelope glycoproteins in apoB-containing lipoproteins [9] , rather than the production of flavivirus-like sub-viral particles . Indeed , HCV appears to contain a limited number of envelope glycoproteins per virion [6 , 7] , precluding any role in driving the budding process . HCV morphogenesis is proposed to be aided by the machinery of VLDL formation of hepatocytes [4] . This results in the production of infectious particles of low buoyant density reflecting a high content in neutral lipids cholesteryl ester [6] and probably also triacylglycerol . Remarkably , recent reports also indicate that the structures of envelope proteins E2 of BVDV and HCV are radically different from each other and from class II proteins of flaviviruses [10–13] . For pestiviruses , much less is known concerning the structure of the virions and the mechanisms of morphogenesis [14] . Electron microscopy images of BVDV-infected cells have revealed the presence of small enveloped ~50 nm viral particles of irregular form [15] budding in the ER membrane [16] . Pestivirus virions produced in cell culture have buoyant densities of 1 . 10–1 . 13 g/cm3 in sucrose gradients [17–19] , intermediate between those of flaviviruses ( 1 . 19–1 . 23 g/cm3 ) and HCV ( 1 . 05–1 . 10 g/cm3 ) [1] . Moreover , it has recently been demonstrated that core , the capsid protein , is not essential for the release of infectious CSFV particles [17] . All these observations suggest that pestiviruses have a virion structure and morphogenesis mechanisms that differ from those of HCV and flaviviruses . To address this question , we set up a method of purification of BVDV virions and analyzed their morphology by cryo-electron microscopy and their protein and lipid content by mass spectrometry .
Viral particles were purified from culture medium of MDBK cells infected with a non-cytopathic strain of BVDV in order to reduce the presence of apoptotic bodies and other cell-derived debris , which are released in high amounts during an infection with a cytopathic strain . Four steps were required in order to obtain a purified virus . Viral particles were precipitated with PEG and separated from heavy membranes and from soluble material by centrifugation through two cushions of 15 and 30% sucrose . The infectious material was found in a fraction located just below the 15–30% interface . The virus was further purified by flotation in a sucrose density gradient . Viral proteins were detected by immunoblotting and the infectivity of each fraction of the gradient was measured . A single peak of C , Erns , E1 and E2 proteins was detected in the gradient ( Fig 1A ) , which matched the distribution of infectious particles ( Fig 1B ) . E1 and E2 were detected as single bands , whereas Erns appeared as a smear or multiple bands , likely representing heterogeneity of Erns-associated glycan structure . C was detected as two bands . However , it is not clear if this actually represents post-translational modification or limited degradation of the protein , because the doublet was not consistently observed . The buoyant density of this peak was measured as 1 . 125 ± 0 . 006 g/cm3 ( n = 5 , mean ± SD ) . This value is consistent with previously reported buoyant density values of infectious BVDV particles from partially purified or unpurified sources [18 , 19] . Finally , most of the remaining membranous contaminants were removed with a chromatography step on cellulose-sulfate beads . At each purification step , viral particles were quantified both by infectious titer and core immunoblotting . Typically , starting from about 900 ml of culture medium titrating approximately 106−107 ffu/ml , this method yielded 50 μl of purified virus with a titer in the range of 1010−1011 ffu/ml . Examination of purified fractions by electron microscopy after negative staining revealed the presence of numerous particles of about 50 nm and a few contaminating objects ( Fig 2A ) . At higher magnification , 50-nm particles appeared more or less spherical ( Fig 2B ) . For a better preservation of BVDV morphology , purified particles were observed by cryo-electron microscopy , a technique that preserves the hydration state of the sample and avoids dehydration artifact induced by negative staining . Again , the more abundant objects were spherical particles with a diameter of about 50 nm ( Fig 2C and 2D and S1 Fig ) . These objects were absent from samples purified from uninfected cells ( S2 Fig ) . They were bounded by two electron-dense thin layers with a clear layer in between , which likely represent the lipid bilayer of the viral envelope . The inner part of the viral particles was made of electron-dense material , likely corresponding to the viral capsid . The preparation also contained some smaller ( 30–35 nm ) non-enveloped objects of the same size and morphology as viral capsids , a small number of larger ( 55–65 nm ) virus-like particles and other unrelated objects ( Fig 2C and 2D ) . We could not find any clear morphological evidence for the presence of envelope glycoproteins at the surface of viral particles using cryo-electron microscopy . We performed immunogold labeling of purified BVDV under negative staining condition , but surprisingly very few viral particles were labeled ( S3 Fig ) . To verify that anti-E2 antibodies actually recognize E2 glycoproteins present on purified viral particles , we analyzed their neutralization potency . No significant difference of neutralization curves ( IC50 = 3 . 44 ng/ml and 3 . 53 ng/ml for unpurified and purified BVDV , respectively ) was observed between unpurified and purified BVDV ( S4 Fig ) , suggesting that E2 glycoproteins were not altered during purification . To confirm that the 50-nm particles actually are BVDV virions , purified material was incubated with an antibody to BVDV glycoprotein E2 coupled to magnetic beads and the immunocaptured material was visualized by cryo-electron microscopy . Magnetic beads covered with anti-E2 antibodies captured 50-nm virus-like particles , thus confirming that these objects are BVDV virions exposing E2 glycoprotein at their surface ( Fig 3A ) . Occasionally , larger virus-like particles could also be observed ( Fig 3A ) , suggesting that they are viral particles of larger size . Control beads with no antibody ( Fig 3B ) or with an irrelevant antibody ( Fig 3C ) did not bind any particles . This result indicates that the 50-nm particles , as well as larger particles of the same morphology expose BVDV E2 glycoprotein at their surface and thus confirms that these particles actually are BVDV virions . Accordingly , these objects were not observed in control fractions purified from non-infected cells ( S2 Fig ) , and very few of them were found in the flow-through fraction of the chromatography step , which is enriched in contaminating material ( S5 Fig ) . Importantly , the contaminating material principally consisted of vesicles that are structurally different from core-containing 50-nm particles of the purified fraction . The purity of a virus preparation was evaluated by counting objects visible in cryo-electron microscopy according to their morphology . Examples of objects that were counted are indicated in Fig 2C and 2D . Among 1871 objects counted , 82% were 50-nm enveloped virions , 7% non-enveloped core-like particles and 2% larger virions; 1% of the material was made of small vesicles about the same size as a virion with no internal core and 2% were larger vesicles . Finally 6% were unrelated objects ( Fig 2E ) . This suggests that this purified fraction contained at least 82% of viral particles , if one considers only the 50-nm enveloped particles , and up to 91% of viral particle-related objects , if one considers small non-enveloped objects as virions that have lost their envelope during the final purification step , and larger virus-like particles as viruses containing a larger capsid . This indicates a residual contamination of about 9% . In order to have a more defined view of BVDV particles , 160 images of viral particles with a diameter of 50 nm were aligned . A few of them are shown in Fig 2F . The projection resulting from this alignment is shown in Fig 2G . The average image clearly showed the lipid bilayer and the internal capsid structure . As observed on individual viral particles , there was no evidence of spikes or of any other proteins at the membrane surface . Moreover , the lipid bilayer of the envelope was clearly visible as two distinct electron-dense layers on most particles ( Fig 2C , 2D and 2F and S1 Fig ) and on the average image ( Fig 2G ) , suggesting that this membrane is poor in trans-membrane proteins , as observed on HIV virions [20] or HCV pseudoparticles [21] . We also tried to purify infectious particles of cytopathic strain NADL using the same protocol . However , the final pellet mainly contained contaminating vesicles of different sizes and shapes and very few 50-nm capsid-containing virus-like BVDV particles ( S6 Fig ) . Therefore this purification protocol is probably not suited for purifying BVDV particles from cytopathic strains , which generate a large number of membranous contaminants during infection . To try to better define NADL particles from contaminating vesicles , we treated purified NADL particles at pH 5 . 1 in the presence of a reducing agent . Such a treatment has been shown to induce fusion of BVDV bound to the surface of cells [22] . However , no change of morphology was observed ( S7 Fig ) , suggesting that the structure of envelope proteins were not modified , or that they were present at low density on the virions and therefore difficult to observe . Pestiviruses are endowed with envelope glycoproteins that are engaged in covalent homo- and heterodimers [23 , 24] . During the purification , the detection of a single E2-immunoreactive band of about 75 kDa in non-reduced virus samples ( Fig 1A ) suggested the presence of virion-associated covalent E1E2 heterodimers , but no band corresponding to E2 monomer ( ~55 kDa ) or covalent homodimer ( ~110 kDa ) could be observed . To assess whether our anti-E2 antibody could detect E2 monomers and homodimers , we probed a lysate of infected MDBK cells under reducing and non-reducing conditions . Three bands were detected with sizes of about 55 , 75 and 110 kDa , consistent with the presence of E2 monomer , covalent E1E2 heterodimers and covalent E2 homodimers respectively in infected MDBK cells ( Fig 4A ) . None of these bands was detected in non-infected cells lysate . Only one band was detected in reduced sample ( Fig 4A ) . This band had an apparent size close to that of the 55-kDa band detected under non-reducing condition , confirming its being E2 monomer . A slight difference of migration of E2 monomer under reducing and non-reducing conditions was observed , as previously reported [23] . This likely reflects changes in the conformation of the glycoproteins induced by the reducing agent . A more compact conformation of the native protein , resulting from the presence of intramolecular disulfide bridges , would explain its faster migration . These data confirmed that the anti-E2 antibodies we used are able to detect E2 monomer , homodimers and heterodimers . Purified virus samples were probed under reducing and non-reducing conditions . Anti-E1 and anti-E2 antibodies detected a band of the same size ( ~75 kDa ) under non-reducing condition ( Fig 4B ) . Under reducing conditions , an anti-E1 antibody revealed a band of 25 kDa , the expected size of E1 monomer , and an anti-E2 antibody revealed a band of 55 kDa , which is the size of E2 monomer observed in cell lysates . This confirmed that the 75-kDa band is a covalent E1E2 heterodimer . On the other hand , we detected bands of different size and aspect with anti-Erns antibody than with E1 and E2 antibodies in non-reduced samples , indicating the absence of covalent E1Erns or E2Erns heterodimers . Single bands of ~90–100 kDa and ~45–50 kDa were detected under non-reducing and reducing conditions , respectively , consistent with the presence of covalent Erns homodimer in pestivirus virions , as reported previously [25 , 26] . To further assess the potential presence of E2 homodimer and E2 monomer on BVDV particles , lysates of partially purified BVDV ( flotation gradient fraction ) and of infected MDBK cells containing similar amounts of E1E2 heterodimer were analyzed by immunoblot under non-reducing condition . No band corresponding to E2 homodimer or E2 monomer could be observed even after prolonged exposure of the blot ( Fig 4C ) . Scanning and quantification of the bands indicated that the amounts of E2 homodimer and monomer on BVDV particles are less than 0 . 3% of E1E2 heterodimer . In contrast , E2 homodimer was detected in all steps during the purification of NADL particles ( S8 Fig ) . It is not clear if this difference resulted from the difference of viral strain or if E2 homodimer was contributed by the large amounts of cell-derived vesicles generated by this cytopathic strain of BVDV . Nevertheless , the variation of homodimer:heterodimer ratio during the purification rather suggests that homodimer and heterodimer were probably present on different populations of objects , probably including both virions and contaminating vesicles . The protein content of purified BVDV particles was analyzed by SDS-PAGE and colloidal Coomassie blue staining . As a control , we analyzed the protein content of an equivalent fraction originating from non-infected MDBK cells . An intense band migrating faster than the 17-kDa marker was present in the infected sample but not in the control ( Fig 5A ) . The migration of this band was similar to the one detected by immunoblot with an anti-core antibody ( Fig 5B ) and to previous reports of BVDV core size [25] . The band of core appeared in less than a minute after addition of the staining solution , whereas other bands began to appear after several hours of staining , indicating that they were present in much lower amounts than core in the sample . Core was the only protein detected by Ponceau red staining , confirming its being the major protein present in the fraction ( Fig 5B ) . The other bands detected in colloidal Coomassie blue staining were far less intense and were also detected in similar amounts in the non-infected control sample , suggesting that they are components of co-purified contaminating vesicles . A smeared band was detected in the infected sample , but not in the control , with a migration similar to that of Erns monomer . However , surprisingly , no band could be clearly assigned to E1 or E2 , when comparing virus and control samples , even though they were detected by immunoblot ( Fig 5B ) , suggesting that BVDV particles have low amounts of envelope glycoproteins , or that BVDV envelope glycoproteins were not efficiently stained with colloidal Coomassie blue . To further characterize the protein content of purified BVDV particles , proteins were analyzed by mass spectrometry . Proteins that were found at least twice out of 4 independent experiments , including one with proteins deglycosylated by PNGase F , are mentioned in Table 1 . Two viral proteins , core and Erns , were identified , corresponding to the bands shown in Fig 5A , but E1 and E2 were not detected using Mascot searches . The other proteins identified were cellular proteins . They corresponded to proteins detected both in purified virus samples and in control fractions by Coomassie blue staining , and are likely components of contaminating vesicles detected in cryo-electron microscopy experiments . To further assess the presence of E1 and E2 on purified BVDV , we compared the theoretical tryptic digest peak list of E1 and E2 with the experimental peak lists of samples corresponding to masses close to 25 or 55 kDa , respectively . Three masses corresponding to E2-derived peptides were found in a sample of ~55 kDa containing Erns and lactadherin ( S9 Fig ) . Due to the low intensity of these three peaks no MSMS experiment was carried out . These data strongly suggest the presence of E2 , but do not strictly prove it . We performed a similar analysis for E1 , but could not identify it . E1 runs as a ~25 kDa protein at a position very close to that of CD9 , which is a major contaminant . We hypothesize that the presence of large amounts of CD9-derived peptides could have interfered with the detection of E1-derived peptides present in low amounts . The absence of detectable bands for E1 and E2 in Coomassie-stained gels and the lack of detection of E1 in mass spectrometry experiments further suggest that E1E2 content of BVDV envelope is low . In contrast to HCV , no apolipoprotein was found associated to purified BVDV during the mass spectrometry experiments . Among the proteins detected , major cytosolic proteins , such as actin , tetraspanins network-associated proteins , such as CD9 and CD9P1 , ESCRT proteins , such as ALIX , and phospholipid-binding proteins , such as lactadherin and annexin A2 , have been previously found in purified exosomes from other cell types [27 , 28] . This suggests that contaminants found in purified BVDV fractions are exosomes . To assess whether some of these host factors could be virion components , we compared their distribution in a sucrose density gradient with the distribution of E2 . The bands were quantified and plotted over fraction density . This revealed the presence of at least two populations of contaminating vesicles in addition to viral particles , one enriched in annexin A2 , and the other one containing CD9 and ALIX ( Fig 5C ) . CD9P1 and lactadherin could not be probed due to the lack of antibodies recognizing bovine proteins . The buoyant density of annexin-A2 peaked at 1 . 108 g/cm3 while virion marker E2 peaked at 1 . 127 g/cm3 . The peak of the second population of contaminating vesicles was shared by fractions with densities of 1 . 127 g/cm3 and 1 . 136 g/cm3 . Despite their distinct distributions , CD9 , ALIX and , to some extent , annexin A2 overlapped E2 in the gradient ( Fig 5C ) . We also assessed whether some of these host factors would preferentially co-purify with the viral particles during the chromatography step . Cellulose-sulfate beads mimic heparan sulfate and preferentially bind viral particles . Therefore , host factors should co-purify with virions , if they are inserted in viral particles and should not bind beads if they are part of contaminating vesicles . Annexin A2 was detected in fractions of a chromatography experiment . Most annexin A2 did not bind cellulose-sulfate beads and was recovered in the flow-through fraction ( Fig 5D ) . A minor amount of annexin A2 was detected in the 0 . 4 M NaCl virus elution fraction , where core was detected . A similar analysis was performed for ALIX , and CD9 ( Fig 5E ) . Again most of the cellular proteins did not bind cellulose-sulfate beads , and a minor fraction was recovered in the virus fraction . All these results strongly suggest that most of these host proteins are part of contaminating vesicles , but we cannot completely rule out the possibility that a minor fraction of some of these proteins could be associated with viral particles . To determine if the low buoyant density of BVDV particles reflects an atypical lipid composition , as it was recently shown for hepatitis C virus particles [6] , we quantified lipids of purified viruses . In parallel , we also quantified the lipid composition of infected and non-infected MDBK cells , in order to reveal any potential changes of lipid metabolism induced by BVDV infection , and to estimate the enrichment or reduction of lipids in the virions . Lipids were extracted from samples in the presence of internal lipid standards and subjected to nano-electrospray ionization MS/MS analyses on a QTRAP5500 . Lipid species detection was performed in either precursor on neutral loss scanning mode , selecting for lipid class specific fragment ions that were generated by collision-induced dissociation under high vacuum . Data evaluation and quantitation of lipid species was done using the LipidView software . Twenty lipid classes ( 13 glycerophospholipids [GPLs] , 3 sphingolipids [SLs] , 3 neutral lipids and cholesterol ) were included in the analysis . The GPLs analyzed were diacyl-phosphatidylcholine ( PC ) , 1-alkyl , 2-acylglycerophosphocholine or 1-acyl , 2-alkylglycerophosphocholine ( ePC ) , phosphatidylethanolamine ( PE ) , 1-alkyl , 2-acylglycerophosphoethanolamine or 1-acyl , 2-alkylglycerophosphoethanolamine ( ePE ) , 1-alkenyl , 2-acylglycerophosphoethanolamine ( also referred to as ethanolamine plasmalogen ( pl-PE ) ) , phosphatidylserine ( PS ) , 1-alkyl , 2-acylglycerophosphoserine or 1-acyl , 2-alkylglycerophosphoserine ( ePS ) , phosphatidylglycerol ( PG , the mass of which is indistinguishable from its structural isomer lyso-bisphosphatidic acid , or LBPA ) , 1-alkyl , 2-acylglycerophosphoglycerol or 1-acyl , 2-alkylglycerophosphoglycerol ( ePG ) , phosphatidylinositol ( PI ) , 1-alkyl , 2-acylglycerophosphoinositol or 1-acyl , 2-alkylglycerophosphoinositol ( ePI ) , phosphatidic acid ( PA ) and 1-alkyl , 2-acylglycerophosphate or 1-acyl , 2-alkylglycerophosphate ( ePA ) . The three SLs analyzed were ceramide ( Cer ) , sphingomyelin ( SM ) , and hexosyl-ceramide ( HexCer ) . The three neutral lipids were cholesteryl ester ( CE ) , triacylglycerol ( TAG ) and diacylglycerol ( DAG ) . Among the 20 lipid classes analyzed , a total of 398 different molecular lipid species were quantified ( S1 Table ) . The lipidomes of BVDV particles and of total cellular membranes revealed a ~2 . 3 to 3 . 5 fold enrichment of cholesterol , SM and HexCer in the viral envelope as compared to cellular membranes ( Table 2 and Fig 6 ) . Cholesterol and SM account for 51 and 17 mol% of BVDV lipids , respectively . Together with HexCer , they contribute to more than 70 mol% of envelope lipids , while their abundance in MDBK cellular membranes is close to 30 mol% . This enrichment in cholesterol and SLs is mirrored by a 1 . 5 to 5 fold reduction of all GPL classes . In contrast , the lipid composition of infected and non-infected MDBK cells did not reveal any significant differences . Storage lipids CE , DAG and TAG were found in similar amounts in purified virus and in a control fraction obtained from non-infected cells . Therefore , they most likely represent components of contaminating material ( S10 Fig ) . The same conclusion applies to Cer and PG , which were also measured in similar amounts in purified virus and control fractions . In addition to the differences observed for lipid classes , some differences could also be observed between individual lipid species of BVDV envelope and MDBK membranes . There is a tendency toward more saturated PC species in BVDV samples as compared to MDBK cells . PC species containing 0 , 1 or 2 double bonds in acyl chains make up 82 . 5% of all PC species of BVDV and about 55% in MDBK cells ( Fig 7A ) . Additionally , there is a shift toward shorter acyl chains . PCs with a sum of 34 carbon atoms in acyl chains were relatively more abundant in BVDV than in MDBK cells , at the expense of species with a sum of 36 carbon atoms or more in acyl chains ( Fig 7B ) . As a result , PC 34:2 and PC 36:2 , the major PC species represent 46 . 8% of all PC species in BVDV and 30 . 2% or 31 . 3% in infected and non-infected MDBK cells , respectively ( Fig 7C ) . The tendency towards less unsaturated acyl chains is conserved , although less pronounced in PE and PI but not in other major GPL classes or in SLs ( S11 Fig ) . We also compared BVDV lipid content with other viruses [6 , 29–31] . Strikingly , BVDV was most similar to influenza virus , when the repartition of major lipid families such as cholesterol , SLs and GPLs , are compared . The distribution of major lipid families of its envelope was very different from that of HCV , another member of the Flaviviridae family , essentially because CE is the major lipid found in HCV [6] and is most probably absent from BVDV . Even when the storage lipid CE was omitted , BVDV and HCV distributions were still dissimilar , BVDV ( 51 . 0% Chol , 29 . 3% GPLs ) being much closer to influenza ( 52 . 3% Chol , 29 . 1% GPLs ) [31] than to HCV ( 37 . 2% Chol , 43 . 3% GPLs ) [6] , VSV ( 43 . 6% Chol , 36 . 9% GPLs ) , SFV ( 42 . 7% Chol , 37 . 2% GPLs ) [30] or HIV ( 45 . 2% Chol , 36 . 4% GPLs ) [29] ( Fig 8A ) . All viruses had similar SLs content ( 18 . 5 to 20% ) . Remarkably , when GPLs distributions were compared , BVDV and influenza were very different from each other . The major GPL of BVDV was PC ( 61 . 6% of all major GPLs ) , which is the main GPL found in the ER membrane [32] . In contrast , PC was detected in very small amounts ( 6 . 4% ) in influenza virus [31] . Influenza virus envelope was enriched in ePE and PS ( 31 . 3 and 43% , respectively ) , while they represent only 0 . 9% and 3 . 8% of BVDV envelope GPLs ( Fig 8B ) . In this respect , BVDV GPL content is closer , although clearly different , to that of HCV [6] , another virus probably budding at the ER membrane , than to any other virus included in this analysis . To assess the functional importance of sphingomyelin and cholesterol , the two major lipids enriched in BVDV virions , we made use of sphingomyelinase and methyl-β-cyclodextrin , respectively . BVDV was pre-incubated with increasing amounts of sphingomyelinase ( up to 10 U/ml ) , and then diluted in culture medium before infection of MDBK cells . A dose-dependent inhibition of BVDV infection was observed ( Fig 9A ) . To ensure that this inhibition did not result from an action of sphingomyelinase on MDBK cells , a control experiment was performed with no pre-incubation with the virus before dilution and infection , and no inhibition of BVDV infection was detected ( panel A in S12 Fig ) . Alternatively , cells were pre-treated with diluted sphingomyelinase in the absence of virus and then infected in the absence of sphingomyelinase , or infected in the absence of sphingomyelinase and then treated with diluted sphingomyelinase post infection . Again no impact on BVDV infection was measured with these conditions ( panel A in S12 Fig ) . These results indicate that lipid modifications induced in BVDV envelope by sphingomyelinase treatment impaired BVDV infection . BVDV was also treated with increasing concentrations of methyl-β-cyclodextrin , in order to extract cholesterol from its envelope , and then diluted before infection of MDBK cells . Again , a dose-dependent inhibition of BVDV infection was observed ( Fig 9B ) . Control experiments with no pre-incubation did not show any inhibition of BVDV infection ( panel B in S12 Fig ) , indicating an action of methyl-β-cyclodextrin on viral particles and not on cells . Importantly , when BVDV treated with 2 . 5 mg/ml methyl-β-cyclodextrin was further incubated with cholesterol-loaded methyl-β-cyclodextrin before infection of MDBK cells , in order to replenish the cholesterol content of the viral envelope , the infectivity was partially restored ( Fig 9C ) . In contrast , BVDV treated with 5 mg/ml methyl-β-cyclodextrin could not be rescued by cholesterol addition . This suggests that a strong cholesterol extraction from the viral envelope irreversibly inactivates the virus , whereas a partial cholesterol extraction impairs BVDV infection but does not irreversibly disrupt the viral envelope . Examination of BVDV by cryo-EM confirmed the integrity of virions in samples treated with 1 U/ml sphingomyelinase or 2 . 5 mg/ml methyl-β-cyclodextrin ( S13 Fig ) . This indicates that the loss of infectivity is not due to the viral particles being destroyed and indeed results from the depletion of specific lipids from the envelope . In contrast , the sample treated with 5 mg/ml methyl-β-cyclodextrin , which did not allow infection rescue by cholesterol replenishment ( Fig 9C ) , contained no virus ( S13 Fig ) . Only a few non-viral intact vesicles , probably derived from cholesterol-poor contaminating membranes , and a number of large aggregates containing membranous material were observed ( S14 Fig ) . This observation explains the lack of cholesterol-mediated rescue with this dose of methyl-β-cyclodextrin . These results confirmed the functional importance of sphingomyelin and cholesterol in BVDV entry . The lipid composition of BVDV envelope suggests that it might bud in membrane domains enriched in cholesterol and SM . Membranes enriched in cholesterol and sphingolipids have a tendency to produce lipid raft domains [33] . These structures are thought to be insoluble and coalesce in cold non-ionic detergents and can be isolated by flotation in a sucrose density gradient as proteo-lipidic complexes named detergent-resistant membranes , or DRMs [34] . To determine if BVDV envelope glycoproteins can be found in DRMs , BVDV-infected cells were lysed on ice in TNE containing Triton X-100 and the lysate was submitted to flotation in a sucrose density gradient in order to resolve DRMs from soluble proteins . Most of caveolin ( a DRM marker ) was found in floated fractions , whereas the transferrin receptor ( a protein excluded from DRMs ) was not ( Fig 10A ) . We also probed the distribution of calnexin , a trans-membrane protein of the ER , and did not observe any signal in floated fractions , as expected , indicating that the ER membrane was well solubilized . Under these experimental conditions , E2 was not found in DRM fractions from infected cells lysates ( Fig 10A ) and in floated fractions from BVDV particles lysates ( Fig 10B ) . This indicates that , despite a high cholesterol and SM content of BVDV envelope , E2 does not appear to be included in raft-like structures of the ER membrane or the viral envelope .
In this study , we report on the first purification of BVDV particles and the description of their morphology and their molecular composition . Due to moderate viral titers and to the presence of contaminating vesicular material of cellular origin , the four-step purification we set up yields a virus ~90% pure ( Fig 1 and Fig 2 ) . The contaminating material included two distinct vesicle populations with different buoyant densities and protein content ( Fig 5 ) . One of these vesicle populations with a density of about 1 . 14 g/cm3 and containing exosome markers such as the tetraspanin CD9 and the ESCRT protein ALIX can be identified as exosomes originating from intraluminal vesicles of multivesicular bodies [27 , 28] . The other population , which is lighter and contains annexin A2 , might have another cellular origin . Importantly , both of these cell-derived vesicle populations are also present in conditioned medium of non-infected MDBK cells , clearly indicating that they are unrelated to BVDV infection and viral morphogenesis . Nevertheless , the presence of limited amounts of contaminating vesicles did not impede morphological and biochemical studies of BVDV particles . One key issue to achieve this purification was the use of a non-cytopathic strain in order to minimize the amounts of contaminating material in supernatants of infected cells . For this reason , attempts to purify BVDV particles from the cytopathic strain NADL were unsuccessful ( S6 Fig ) , despite higher titers . Purified BVDV virions appeared in electron microscopy as spherical particles with a diameter of 50 nm ( Fig 2 ) , in keeping with previous observations of cell-associated BVDV particles [16 , 23] . In cryo-electron microscopy , virions are characterized by an electron-dense core surrounded by a lipid bilayer , which displays a smooth surface with no spikes ( Fig 2 and S1 Fig ) . This could reflect the presence of envelope glycoproteins lying parallel to the lipid bilayer , like E dimers of flavivirus mature virions [3 , 35] . However , acidic treatment of purified BVDV virions did not induce the formation of spikes ( S7 Fig ) , in contrast to what has been reported for flaviviruses [36] . This difference of response to an acidic treatment may result from the pH-insensitivity of BVDV [22] , or the structural organization of BVDV envelope glycoprotein E2 , which is very different from class II glycoprotein E of flaviviruses [10 , 11] . Alternatively , this may also reflect the presence of low amounts of envelope glycoproteins at the surface of BVDV particles , a hypothesis that is also supported by biochemical data ( Fig 5 and S9 Fig ) . Interestingly , the vast majority of viral particles observed in cryo-electron microscopy have a uniform size of approximately 50 nm , except for a low number ( about 2% ) that are larger with a diameter up to 65 nm ( Fig 2 and S1 Fig ) , which corresponds to a volume twice as large . We cannot exclude that these larger particles could be aberrant forms related to cell culture conditions . However , BVDV is known to be able to incorporate pieces of host RNA into its genome [14] . We can speculate that a size flexibility of the virion would be an advantage for the virus to accommodate larger genomes . To our knowledge , such a size flexibility of the virion has not been observed for flaviviruses . This suggests different structural organization of the capsid of pestiviruses and flaviviruses . The polymorphic nature of a viral particle may arise from non-icosahedral structure , as in rubella virus [37] , or from variations in quasi-symmetry or triangulation number of icosahedral structure , as in hepatitis B virus capsid [38] . Whether BVDV capsid structure is isosahedral or not must await studies with higher resolution cryo-EM techniques . The purified virus preparation also contained non-enveloped particles corresponding to putative naked capsids ( Fig 2 and S1 Fig ) . This is most likely an artifact of the last step of purification , for two reasons: non-lipidated particles , if infected cells were to secrete them , should have a buoyant density quite different from that of enveloped particles and be discarded during the two steps of ultracentrifugation . However we always detected a single peak of capsid-containing material in these two ultracentrifugation steps . Moreover , cryo-electron microscopy examination of virus fractions obtained after the flotation gradient but without chromatography on sepharose-sulfate beads did not reveal any naked capsid-like particle . This strongly suggests that non-enveloped capsids are generated during the last step of purification . One surprising finding of this study is the apparent low glycoprotein content of BVDV particles . E1 and E2 were never detected in colloidal Coomassie blue-stained gels , while core was very abundant and even some proteins of the ~10% contaminating material were easily detected ( Fig 5 ) . On the other hand , Erns was detected as a smear , however apparently in lower amounts than core ( Fig 5 ) . This suggests that BVDV particles have a higher content of Erns than E1 and E2 , a conclusion that is also supported by results of mass spectrometry analysis . E1 was not detected in four independent experiments , and E2 could be identified only in low amounts ( S9 Fig ) , while Erns could easily be identified . The deglycosylation of viral particles with PNGase did not improve E1 and E2 detection in mass spectrometry ( Table 1 ) , suggesting that their low levels of detection did not simply result from hindrance caused by their carbohydrate moieties . These lower amounts of E1 and E2 were previously suggested from immunogold labeling of BVDV particles in electron microscopy [15] . Lower levels of E1E2 are surprising because E1 and E2 are required for pestivirus entry and Erns is not [39 , 40] . However , we should keep in mind that it has been reported for other viruses that a very low number of functional envelope proteins may be required for entry [41–44] . E1 and E2 could be detected in purified BVDV particles by immunoblotting ( Fig 1 , Fig 4 and Fig 5 ) , by neutralization ( S4 Fig ) and by immunocapture ( Fig 3 ) . As reported earlier [23 , 45] , E1 and E2 were found as covalent heterodimers . In contrast , cell lysates of infected cells showed a mixture of monomers , covalent heterodimers and covalent E2 homodimers ( Fig 4 ) . This suggests that a mechanism operates during morphogenesis that allows covalent heterodimers to be specifically incorporated in the nascent particle . Alternatively , E1 and E2 disulfide bridges could be reshuffled during assembly . The presence of covalent heterodimers is consistent with the report that only covalent E1E2 heterodimers are required for BVDV entry , and that covalent E2 homodimers are inactive [40] . In previous studies , E2 covalent homodimers were detected in unpurified pestivirus particles in addition to covalent E1E2 heterodimers [23–25] . It is not yet clear whether the presence of E2 homodimers in virions is strain-specific , or if it merely reflects the presence of contaminating vesicles of cellular origin . For example , material secreted by MDBK cells infected by cytopathic strain NADL contained both homodimers and heterodimers . Although we did not succeed in purifying NADL particles , we observed that the homodimer:heterodimer ratio varied during purification ( S8 Fig ) , suggesting that they are present in different populations of vesicles in the starting material . A major finding of this study is the unexpected lipid composition of BVDV envelope ( Fig 6 , Fig 7 and Table 2 ) . The lipid composition of the viral envelope does not match any known intracellular membrane . Its cholesterol and SM content are somewhat similar to what could be found in the plasma membrane of some cells , and in envelopes of viruses budding at the cell surface ( Fig 8 ) . However PS and ethanolamine plasmalogens , two phospholipids enriched in the plasma membrane and viruses budding at the plasma membrane [29–31 , 46] were found in low amounts in BVDV envelope , suggesting that the lipid bilayer of the viral envelope is not likely to derive from the plasma membrane . Similarly , the very low ceramide and PG/LBPA content of BVDV envelope does not support a budding event in multivesicular endosomes . A high content of cholesterol and sphingolipids is often associated with lipid rafts [33] . However , ethanolamine plasmalogens , another marker of lipid rafts [47] , are not enriched in BVDV envelope ( Fig 6 and Table 2 ) . Moreover , viral glycoprotein E2 was not found in DRMs of infected cells or purified virus ( Fig 10 ) . These observations are not in favor of BVDV budding in lipid raft microdomains of MDBK cells . Several lines of evidence are in favor of BVDV budding into the ER . Envelope glycoproteins are located at the ER membrane of infected cells [48 , 49] . Infectious BVDV particles accumulate intracellularly upon brefeldin A treatment [50] . Individual budding events at the ER membrane have been observed by electron microscopy [16] . In contrast , the BVDV envelope lipid composition , and especially its cholesterol and SM content , is quite different from a typical ER membrane [32 , 51] , even though the lipid content of the ER of MDBK cells has not been determined to our knowledge . Interestingly , SM has been recently shown to be involved in the morphogenesis of West Nile virus [52] , a flavivirus that also buds in the ER , suggesting a potential role of SM in viral budding in the ER . On the other hand , if one only considers GPLs , it is worth noting that BVDV envelope is quite similar to ER membrane [32 , 51] and to the envelope of HCV , another virus budding in the ER [6] , which both contain high amounts of PC and PE ( Fig 8 ) . These data suggest that the distribution of GPLs could be more representative of the membrane where the virus buds than the full lipid repertoire . Remarkably , BVDV and influenza virus , two viruses budding from cellular membranes so different as ER and apical PM , reach similar proportions of GPLs , SLs and cholesterol ( Fig 8 ) . Their high cholesterol and SLs contents are likely to render their envelopes very sturdy and help them to resist to harsh conditions found outside their hosts . It is tempting to hypothesize that this is an example of convergent evolution driven by similar modes of transmission . Influenza lipid content is very similar to apical PM content and the formation of its envelope only involves minor sorting events of selected lipid species [31] . On the other hand , BVDV envelope lipid content is highly divergent from a typical ER membrane composition . Three different hypotheses could explain this observation . A first hypothesis is that the viral infection could alter the cellular metabolism in such a way as to modify the lipid composition of the cellular membrane where the virus buds , here the ER membrane . However this hypothesis is unlikely , because we did not detect any difference of lipid content between infected and non-infected cells ( Table 2 ) . Actually , BVDV replication has been associated to very discrete membrane rearrangements [53] . This absence of any major membrane rearrangement is in line with the absence of lipid content modification during BVDV infection . A second hypothesis is that BVDV would bud into microdomains of the ER membrane enriched in cholesterol and SM , whether they are normally present in the ER membrane or their formation is induced during BVDV infection . Although ER is known not to be enriched in SM and cholesterol , evidence for the presence of DRMs in the ER has been reported [54 , 55] , suggesting a mosaic composition of the ER membrane . However , our results do not support BVDV budding in lipid rafts ( Table 2 and Fig 10 ) . Alternatively , a third hypothesis is that viral proteins could recruit/concentrate cholesterol and sphingolipids to nascent viral envelope during the budding process . It is unlikely that E2 could drive such a process , because of its low amounts in virions and its low affinity to membranes enriched in cholesterol and SLs . Viral proteins fitted for performing such a function might be core , Erns and/or non-structural proteins involved in the assembly process , such as NS3 [17] . Our data add to the notion that members of different genera of the Flaviviridae family use different mechanisms for their morphogenesis . Unlike flaviviruses , HCV [6 , 7] and BVDV ( this work ) appear to contain a limited number of envelope glycoproteins per virion . BVDV also differs from HCV , which uses the machinery of VLDL formation of hepatocytes [4] . The absence of associated apolipoproteins and the very low amounts of CE and TAG , if any , in BVDV particles clearly indicates that BVDV does not rely on lipoproteins for its morphogenesis , in agreement with previous demonstration that LDL receptor plays no role during BVDV entry [56] . Our results further suggest that major changes of lipid content in the bilayer originating from the ER membrane are involved in the formation of BVDV envelope . Future studies on molecular and cellular mechanisms operating this process of lipid sorting should indicate important aspects of pestivirus morphogenesis .
Dulbecco’s modified Eagle’s medium ( DMEM ) , phosphate buffered saline ( PBS ) , glutamax , horse and goat sera , DAPI , gentamycine , nonessential amino acids and trypsin/EDTA were purchased from Life Technologies . All other chemicals were purchased from Sigma . BVDV-free MDBK cells ( ATCC number CCL-22 ) were propagated in DMEM , high glucose modification , supplemented with 10% heat-inactivated horse serum , nonessential amino acids and 2 mM glutamax at 37°C in a humidified atmosphere containing 5% CO2 , and were passaged by trypsination twice a week . BVDV genotype 1 non-cytopathic strain WAX-N used in this study was obtained from P . P . Pastoret ( University of Liège , Belgium ) , and maintained by passages in MDBK cells . The coding sequence of the N-terminal part of the polyprotein ( Npro-p7 ) , which encompasses the structural proteins , was determined from RT-PCR fragments amplified from total RNA of infected cells using primers listed in table A in S1 Text , and deposited to Genebank ( accession number KR013753 ) . BVDV cytopathic strain NADL was obtained from C . M . Rice ( The Rockefeller University , New York ) . About 104 MDBK cells were seeded in wells of P96 plates and infected 4 hours later with 10-fold dilution series of BVDV samples . The medium was replaced after 1 hour of contact . Cells were fixed at 18 to 20 hpi with 3% paraformaldehyde and processed for immunofluorescent detection of NS3 . Immunofluorescent foci were counted from wells infected with appropriate BVDV dilution , and viral titers were expressed as focus-forming units ( ffu ) per ml . Approximately 3 . 108 MDBK cells were infected in suspension with occasional mixing at an M . O . I . of approximately 0 . 02 for 1 hour at room temperature in a volume of 20 ml DMEM containing 10% horse serum and then diluted to 450 ml of DMEM containing 2% horse serum and plated in 3 cell factories ( Easyfill-2trays , Nunc ) . Cell culture supernatants were collected at 40 hpi and then again 24 h later . Both cell culture supernatants were combined and centrifuged at 5000 rpm , 4°C for 15 minutes in a JS-7 . 500 rotor ( Beckman ) to get rid of cellular debris . Polyethyleneglycol ( PEG ) -6000 ( 7% ) was slowly added to supernatants under stirring in order to allow rapid solubilization . The mixture was incubated at 4°C under slow continuous stirring for 4 hours and centrifuged at 10 , 000 rpm , 4°C for 30 minutes in a JLA-10 . 500 rotor ( Beckman ) . The pellet was resuspended in 7 ml TNE ( TNE is 20 mM TrisCl pH 8 . 0 , 150 mM NaCl , 2 mM EDTA ) and incubated overnight at 4°C . Insoluble material was removed by centrifugation ( 7000 rpm , 4°C for 10 minutes in a JS-7 . 500 rotor ) . The supernatant was loaded over 2 layers of 30% ( 2ml ) and 15% ( 3 ml ) sucrose ( w:v ) in TNE in a 12-ml ultraclear centrifuge tube ( Beckman ) and centrifuged at 35 , 000 rpm , 4°C for 2 hours in an SW 41 rotor ( Beckman ) . Fractions ( 0 . 5 ml ) were collected dropwise after puncturing the bottom of the tube . The fraction containing the major part of infectious virus and core protein was mixed with 1 volume of 60% sucrose in TNE , loaded under a 5–35% sucrose gradient , and centrifuged at 35 , 000 rpm , 4°C for 20 hours in an SW 41 rotor . Fractions ( 0 . 5 ml ) were collected dropwise after puncturing the bottom of the tube . The distribution of the virus in the gradient was measured by titration and/or core immunoblotting . The density of each fraction was measured with a refractometer . One or two fractions of the gradient containing the peak of infectious virus were loaded on a 0 . 5-ml column of sulfated cellulose ( Cellufine Sulfate , Chisso Corporation ) pre-equilibrated with TNE . After virus binding , the column was washed with 3 volumes of PBS and viral particles were eluted in 1 ml of PBS containing 0 . 4 M NaCl . Eluted fraction was centrifuged at 45 , 000 rpm , 4°C for 3 hours in a TLA-55 rotor ( Beckman ) and the pellet of purified virus was resuspended in 50 μl PBS . For electron microscopy analysis , the virus was kept at 4°C in PBS until analysis . For lipid extraction and analysis , two preparations were pooled . The final pellet was resuspended in 155 mM ammonium bicarbonate and kept at -80°C . Firstly a continuous carbon film was deposited on the carbon face of a quantifoil with circular holes cupper grid ( Ted Pella Inc ) . BVDV samples ( 5 μl ) were deposited onto the cupper surface of the grid placed in the automated device for plunge-freezing ( EM GP Leica ) enabling a perfect control of temperature ( 15°C ) and relative humidity ( 70% RH ) . The excess of sample was blotted with filter paper and the grid was plunged into a liquid ethane bath cooled and maintained at -183°C with liquid nitrogen . Specimens were maintained at a temperature of approximately -170°C , using a cryo holder ( Gatan , CA , USA ) and observed with a FEI Tecnai F20 electron microscope operating at 200 kV and at a nominal magnification of 29 , 000 x under low-dose conditions . Images were recorded with a 2k x 2k USC 1000 slow-scan CCD camera ( Gatan ) . Pixel size was 0 . 37 nm . Protein A and protein G-conjugated superparamagnetic beads ( Bio-Adembeads ) were purchased from Ademtech . Before use , the beads were washed with a large volume of working buffer ( PBS ) . The magnetic beads ( 10 μl ) were incubated with anti-BVDV E2 mAb or anti-KDEL mAb ( control ) in a final volume of 200 μl , at 4°C for 30 min . The magnetic bead–protein A/G–antibody complexes were sedimented with a magnet . The pellets were washed three times to remove excess antibody . Beads coated with antibody were incubated with BVDV at room temperature for 30 min ( final volume 100 μl ) . After sedimentation with the magnet , the pellet was washed in order to remove unbound particles . Cryo-TEM lacey carbon cupper grids were then prepared with this suspension , without any further treatment according to the protocol described above . Specimens were maintained at a temperature of approximately -170°C , using a cryo holder ( Gatan , CA , USA ) and observed with a FEI Tecnai12 electron microscope operating at 120 kV and at a nominal magnification of 30 , 000 x under low-dose conditions . Images were recorded with a 4k x 4k slow-scan CCD camera ( FEI ) . Proteins of purified virus or a corresponding fraction from non-infected cells were separated under reducing conditions in a 12% polyacrylamide gel . In one experiment , half of a purified virus preparation was denatured and deglycosylated overnight at 37°C with PNGase F ( P0704 , New England Biolab ) as recommended by the manufacturer , before gel electrophoresis . The gel was stained for 3 days in a solution of colloidal Coomassie blue . Stained bands were excised from the gel , reduced , alkylated with iodoacetamide ( 10 mg/ml in NH4HCO3 20 mM ) and digested overnight with 50 ng trypsin ( Promega ) in 20 mM NH4HCO3 . The resulting peptides mixtures were eluted from the gel , desalted , and spotted on a MALDI plate with freshly dissolved α-cyano-4-hydroxycinnaminic acid ( 10 mg/ml in 50% CH3CN , TFA 1/1000 ) . Mass spectrometry was performed with a MALDI-TOF-TOF Autoflex Speed ( Bruker Daltonics ) . MS and MS/MS data were analyzed using BioTools software . Identification of peptides was performed using Mascot , http://www . matrixscience . com/ . Lipid extractions were performed using chloroform:methanol:37% HCl ( 5:10:0 . 15 , vol:vol:vol , ) except for plasmalogens , which were extracted using chloroform/methanol ( 5:10 , vol:vol ) as extraction solvent . Lipid extractions were done in 10 ml Wheaton vial with Teflon-screw caps as described [60] . Typically , the following lipid standards were added to the solvent prior to extractions: 100 pmol of PC ( 13:0/13:0 , 14:0/14:0 , 20:0/20:0; 21:0/21:0 , Avanti Polar Lipids ) , SM ( d18:1 with N-acylated 15:0 , 17:0 , 25:0 , semi-synthesized as described in [36] ) and d6Chol ( Cambrigde Isotope Laboratory ) , 55 pmol PI ( 16:0/16:0 , 17:0/20:4 , Avanti Polar Lipids ) , 50 pmol PE and PS ( 14:1/14:1 , 20:1/20:1 , 22:1/22:1 , semi-synthesized as described in [60] , DAG ( 17:0/17:0 , Larodan ) and cholesterol ester ( CE , 9:0 , 19:0 , 24:1 , Sigma ) , 40 pmol TAG ( D5-TAG-Mix , LM-6000 / D5-TAG 17:0 , 17:1 , 17:1 –Avanti Polar Lipids ) , 10 pmol Cer and GlcCer ( d18:1 with N-acylated 15:0 , 17:0 , 25:0 , semi-synthesized as described [60] , PA ( PA 17:0/20:4 , Avanti Polar Lipids ) and PG ( 14:1/14:1 , 20:1/20:1 , 22:1/22:1 , semi-synthesized as described in [60] . Neutral extraction solvents were spiked with 100 pmol PC standard mix and plasmalogen PE ( pl-PE ) standard mix containing 66 pmol pl-PE Mix 1 ( 16:0p/15:0 , 16:0p/19:0 , 16:0p/25:0 ) , 93 pmol pl-PE Mix 2 ( 18:0p/15:0 , 18:0p/19:0 , 18:0p/25:0 ) and 129 pmol pl-PE Mix 3 ( 18:1p/15:0 , 18:1p/19:0 , 18:1p/25:0 ) . Semi-synthesis of pl-PE was performed as described in [61] . For lipid analysis , 10–20 μl of purified virions and 1-5x104 cells were subjected to extractions . Mass spectrometric analysis was performed on a QTRAP5500 ( ABSciex ) coupled to a Triversa NanoMate device ( Advion ) as described [60] . Data processing was performed using LipidView ( ABSciex ) , Microsoft Excel and a custom-made data evaluation program ( ShinyLipids ) . Cells were incubated in lysis buffer ( 50 mM Tris-Cl buffer pH 7 . 5 containing 100 mM NaCl , 1 mM EDTA , 1% Triton X-100 , 0 . 1% sodium dodecyl sulfate ( SDS ) , and protease inhibitors ) for 20 min on ice . Nuclei were pelleted by centrifugation . The protein concentration in post-nuclear supernatants was determined by the bicinchoninic acid method as recommended by the manufacturer ( Sigma ) , using bovine serum albumin as standard . Proteins from cell lysates or virus purification fractions were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes ( Hybond-ECL; Amersham ) by using a Trans-Blot apparatus ( Bio-Rad ) . The proteins of interest were revealed with specific primary antibodies , followed by species-specific secondary antibodies conjugated to horseradish peroxidase ( Jackson Immunoresearch ) , and enhanced chemiluminescence detection as recommended by the manufacturer ( Thermofischer ) . Unpurified BVDV was incubated in culture medium for 1 h at 37°C with sphingomyelinase from Bacillus cereus ( Sigma S7651 ) at various concentrations . Treated virus was then diluted 1 , 000 times in culture medium before infecting MDBK cells . Partially purified BVDV ( 15–30% sucrose interface ) was incubated in culture medium for 1 h at 37°C in the presence of various concentrations of methyl-β-cyclodextrin . Treated virus was then either diluted 10 , 000 times in culture medium and used to infect MDBK cells , or diluted 100 times in DMEM containing various concentrations of cholesterol:methyl-β-cyclodextrin complex and incubated for 1 h at 37°C in order to reload methyl-β-cyclodextrin-treated virus with cholesterol . Treated virus was diluted 100 times before infecting MDBK cells . To quantify the impact of sphingomyelinase and methyl-β-cyclodextrin treatments on BVDV infectivity , MDBK cells ( 10000 cells per well of 96-well plate ) were infected for 1 h at 37°C with 100 μl of treated or control virus . The virus was removed and the cells were overlaid with 100 μl of fresh culture medium . Cells were fixed 15 h later with 3% paraformaldhehyde in PBS and processed for immunofluorescent detection of NS3 . Nuclei were stained with DAPI . Samples were observed with a Zeiss Axiophot microscope equipped with a 10 X magnification objective . Fluorescent signals were collected with a Coolsnap ES camera ( Photometrix ) . For quantification , images of randomly picked areas from each well were recorded and processed using ImageJ software . The total number of cells was obtained from DAPI-labeled nuclei . NS3-positive cells were counted as infected cells . The infection was scored as ratio of infected cells to total cells . To assess the impact of sphingomyelinase and methyl-β-cyclodextrin treatments on BVDV structure , purified BVDV eluted from the cellulose-sulfate column was incubated at 37°C with sphingomyelinase ( 1 U/ml ) or methyl-β-cyclodextrin ( 2 . 5 or 5 mg/ml ) or the same volume of PBS ( sphingomyelinase and methyl-β-cyclodextrin solvent ) . Treated virus was then transferred on ice and kept at 4°C until cryo-electron microscopy analysis . A small volume was removed for verifying the impact of treatments on infectivity . Lipid rafts were isolated by standard procedures . Infected MDBK cells were grown 150-mm dish to reach confluence . Cells were rinsed with cold PBS and lysed in 1 ml of TNE containing 0 . 8% Triton X-100 , and supplemented with protease inhibitors ( Roche ) . Lysates were scraped off the dish with a cell lifter , the dish was rinsed with 1 ml of the same buffer at 4°C , and the lysate was homogenized in a dounce homogenizer . The extract was finally brought up to 40% sucrose in a final volume of 4 ml and sequentially overlaid with 5 ml of 30% sucrose and 3 ml of 5% sucrose . Gradients were centrifuged for 18 h at 40 , 000 rpm at 4°C in a Beckman SW41 rotor . Fractions ( 1 ml ) were collected from the bottom of the gradient and immediately supplemented with 20μl of a fresh 50X solution of protease inhibitors . Aliquots of each fraction were analyzed by immunoblotting .
|
Bovine viral diarrhea virus ( BVDV ) is the etiologic agent of mucosal disease and bovine viral diarrhea , two economically important diseases of the livestock . BVDV is a member of the Pestivirus genus in the Flaviviridae family , which also includes Hepacivirus and Flavivirus genera . Members of this family share similar genome organization and replication strategies , but differ about their mode of transmission and particle structure . Whereas most studies have been so far performed on viruses of the Hepacivirus and Flavivirus genera , little is known about infectious particles of pestiviruses . In this study , we set up a novel purification method of BVDV infectious particles and analyzed their morphology by cryo-electron microscopy and their molecular composition by mass spectrometry . Our results provide new insights into the structure and biochemical composition of a pestivirus infectious particle , and have implications for research on molecular mechanisms of their morphogenesis and entry .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"vesicles",
"pathology",
"and",
"laboratory",
"medicine",
"hepacivirus",
"pathogens",
"microbiology",
"viral",
"structure",
"electron",
"cryo-microscopy",
"viruses",
"rna",
"viruses",
"microscopy",
"cellular",
"structures",
"and",
"organelles",
"research",
"and",
"analysis",
"methods",
"lipids",
"medical",
"microbiology",
"microbial",
"pathogens",
"hepatitis",
"c",
"virus",
"hepatitis",
"viruses",
"viral",
"packaging",
"viral",
"replication",
"virions",
"cholesterol",
"biochemistry",
"cell",
"biology",
"flaviviruses",
"virology",
"viral",
"pathogens",
"electron",
"microscopy",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2016
|
Morphology and Molecular Composition of Purified Bovine Viral Diarrhea Virus Envelope
|
Recent whole genome polymerase binding assays in the Drosophila embryo have shown that a substantial proportion of uninduced genes have pre-assembled RNA polymerase-II transcription initiation complex ( PIC ) bound to their promoters . These constitute a subset of promoter proximally paused genes for which mRNA elongation instead of promoter access is regulated . This difference can be described as a rearrangement of the regulatory topology to control the downstream transcriptional process of elongation rather than the upstream transcriptional initiation event . It has been shown experimentally that genes with the former mode of regulation tend to induce faster and more synchronously , and that promoter-proximal pausing is observed mainly in metazoans , in accord with a posited impact on synchrony . However , it has not been shown whether or not it is the change in the regulated step per se that is causal . We investigate this question by proposing and analyzing a continuous-time Markov chain model of PIC assembly regulated at one of two steps: initial polymerase association with DNA , or release from a paused , transcribing state . Our analysis demonstrates that , over a wide range of physical parameters , increased speed and synchrony are functional consequences of elongation control . Further , we make new predictions about the effect of elongation regulation on the consistent control of total transcript number between cells . We also identify which elements in the transcription induction pathway are most sensitive to molecular noise and thus possibly the most evolutionarily constrained . Our methods produce symbolic expressions for quantities of interest with reasonable computational effort and they can be used to explore the interplay between interaction topology and molecular noise in a broader class of biochemical networks . We provide general-purpose code implementing these methods .
Investigations in yeast [1] , [2] led to the hypothesis that in most organisms the recruitment of polymerase to the promoter is the primary regulated step in the activation of gene expression [3]–[6] . However , recent studies of multicellular organisms have revealed a diverse array of other regulatory strategies , including several types of post-initiation regulation [7]–[9] . Zeitlinger et al . [7] generated tissue-specific whole-genome polymerase binding data in Drosophila melanogaster and showed that regulation of polymerase release from the promoter is widespread during development . Their data shows that some 15% of tissue-specific genes bind polymerase to their promoters in all tissues , even though each gene only allows polymerase to proceed through the coding sequence in a specific tissue ( see Figure S1 ) . Differential expression of these genes is made possible by a paused state wherein a polymerase remains stably bound but precisely stopped a short distance from the promoter and awaits a regulated release that is only triggered in the appropriate tissue [7] . Finally , many metazoa have been shown to have , genome-wide , disproportionate amounts of polymerase bound at promoter regions as compared to coding regions [7] , [8] , [10] , [11] . This mechanism has been called promoter proximal pausing . It should not be confused with the stochastic stalling of a polymerase as it transcribes , a phenomenon which has also been termed “polymerase pausing” . Furthermore , there are distinctions to be made between: stalled polymerase , a polymerase which associates in a transient , unstable manner with the promoter but does not proceed into productive transcription; poised polymerase , a polymerase for which the association is stable but has not escaped from the promoter to begin transcription; and promoter proximal paused polymerase , a polymerase that completely escapes from the promoter but “pauses” in a stable , inducible state just downstream of the promoter . It is believed that most genes which have polymerase bound to their promoters in all tissues but expressed in only some tissues fall in the last category; this promoter proximal accumulation of pol II may indicate that regulation of pausing transitions is a general feature of metazoan transcriptional control . We remind the reader that a gene need not use the paused state as a waiting step at which to integrate regulatory information in order to be termed a paused gene , as even constitutive house-keeping genes may be denoted as paused [12] . In this study we will be interested only in the elongation regulated subset of paused genes . For further discussion of terminology and assays which distinguish these conditions , see the Supporting Information . It remains an open question why expression of some genes is controlled further downstream than others . Several groups have postulated that pausing may ready a polymerase for rapid induction [8] , [10] , [13] . ( Here induction refers to the first time at which all the components required for expression of a particular gene become available , and expression is when transcription of the first nascent mRNA transcript begins . ) To motivate this idea , the preloaded , paused polymerase is described as a “loaded gun” ready to shoot off a single transcript as soon as it is induced . Experiments with heat shock genes – the first class of genes for which paused promoters were identified – show evidence of rapid induction consistent with this idea [14] , [15] . However , pre-loading only provides an argument for why the first transcript would be produced more quickly . Surprisingly then , it was also observed by Yao et al . [14] that subsequent polymerases are recruited rapidly to promoters of induced , elongation-regulated genes as well as the first , preloaded Pol II – a phenomenon not accounted for by the loaded gun metaphor . Since most genes must be transcribed several times in order to produce functional levels of mRNA , changes in speed of induction as a whole are likely to be of more physiological consequence than changes in the time at which the first , pre-paused transcript releases . When whole-genome studies extended the observation of pausing to cover many key developmental regulatory genes [7] , further questions arose . While the selective advantage of rapid induction is reasonably apparent for stress response genes , it is harder to explain why rapid induction would be selected for in so many developmental transcription factors and signaling pathway components . An additional hypothesis , suggested by Boettiger and Levine [16] , is that regulation of transcriptional elongation ( for instance , by promoter proximal polymerase pausing ) may have evolved to ensure more coordinated expression across populations of cells . This hypothesis was motivated by the striking correspondence between genes shown experimentally to activate in a synchronous fashion and genes shown to bind polymerase at the promoter independent of activator state but not continue elongation until activator arrival . Recent work by Darzacq and colleagues [17] provides insight into why a regulatory interaction downstream of transcriptional pre-initiation complex ( PIC ) assembly may lead to more coordinated gene expression than does regulation upstream of PIC assembly . Using fluorescently tagged transcription components , they demonstrated that transcriptional initiation is a highly variable process , with only about one in ninety Pol II–gene interactions leading all the way to productive mRNA elongation [17] . Nonproductive interactions each lasted between several seconds and a minute , suggesting that abortion of transcriptional initiation can occur at different stages in assembly of the complex . Regulatory interactions that occur after this noisy assembly process would act only on transcriptionally competent polymerases , and so this mechanism might result in more synchronous expression – a hypothesis we test here . The idea that gene expression itself is intrinsically variable ( rather than variable as a result of extrinsic fluctuations in upstream quantities ) is well established and is a recent focus of theoretical and experimental interest – see [18] and [19] for reviews . Stochasticity can arise at many stages of the process , including from the diffusion of molecules in the cell [20] , noisy gene regulation [21] , chromatin and other conformal rearrangements [22] , random events during elongation [23] , [24] , and random dynamics of translation and degradation of mRNA and proteins [25] . Populations of single-celled organisms have been shown to take advantage of noisy gene expression to achieve clonal yet phenotypically heterogeneous populations [26] . In metazoan development , however , proper growth and development generally relies on coordination and synchrony rather than stochastic switching . For example , certain cells in the Drosophila embryo are induced to become neurons if they are next to a mesoderm cell but not mesoderm themselves [27] , so uneven activation of mesoderm fate could produce early patches of mesoderm , thereby improperly inducing neuronal development in neighboring tissue . Although synchronous behavior is important for metazoa , particularly in development , it is not a universal property of all metazoan genes . For instance , genes with both synchronous and very stochastic patterns of induction have been observed in the Drosophila embryo [16] . The unique challenges of coordinating the behavior of a large number of independent cells may explain why elongation regulation aimed at release from a paused state appears to be much more dominant among metazoa like D . melanogaster and humans than E . coli or S . cerevisiae . Here we investigate mathematically whether the significant change in the coordination of expression observed in experiment [16] can be explained by a change in the regulation network topology which only effects whether regulation occurs before or after PIC assembly , while keeping other details ( reactions and rates ) of the PIC assembly process the same . We also seek to determine which interactions in the transcriptional pathway are most important for determining the coordination of expression , and what effect different topologies have on the speed of induction and variability between sister cells in total number of mRNA synthesized . We do this by constructing continuous-time Markov chain models of PIC assembly with states that correspond to joint configurations of the promoter and the enhancer . The ( random ) time taken for the chain to pass from a “start” state to an “end” state corresponds to the elapsed time between successive transcription events . The models we construct for the two different modes of regulation have a common set of transition rates , but the particular mode of regulation dictates that certain transitions are disallowed , resulting in two chains with different sets of states accessible from the “start” state . We describe this situation by saying that each model is a topological rearrangement of the other . Because the same set of transition rates completely parametrize both chains , ( see Figure 1 ) we can make meaningful comparisons between the two models . Once the Markov chains are constructed , we use the Feynman–Kac formula [28] , model-specific decomposition techniques and computer algebra to find symbolic expressions for features of these first passage times that correspond to the delay between induction and transcription . Although there has recently been much work modeling different sources of stochasticity in gene expression , most models refrain from a detailed representation of the different protein–DNA complexes involved in favor of more abstract approximations [26] , [29]–[34] . Two–state “on–off” Markov chains have been used many times to model stochasticity in transcription ( e . g . [21] , [35] ) , and provide analytic solutions . Such models have been used to explain , for instance , the observation that mRNA copy number does not in general follow Poisson statistics , implying that there are “bursts” of transcription in some sense . This bursting behavior can occur if the gene transitions between an active state ( in which transcription can occur ) , and an inactive state ( in which it does not ) , as shown by Raj et al . [36] . Although more complicated Markov chain models have appeared , often presented via a stochastic chemical master equation [37] , they are usually simulated rather than studied analytically ( see [38] for a review of methods and software ) . A notable recent exception is Coulon et al . [39] , who use matrix diagonalization to study the power spectrum and other properties of several models of regulation . A complementary set of techniques takes a broader view , using the fluctuation–dissipation theorem to work on the scale of small stochastic deviations from the differential equations that capture the average behaviors at equilibrium [29]–[31] , [33] , [39] . We model the intrinsic noise of regulation and polymerase recruitment using biologically-derived Markov chain models . We focus on this particular piece of the larger process of expression in greater detail than has been done previously in order to provide a detailed mathematical investigation of the role of promoter proximal pausing . Unlike simulation methods , our approach provides a tractable way to compute analytic expressions for which interpretation is direct and reliable . Moreover , it does not depend on small-noise or equilibrium assumptions , or require the passage to a continuum limit . Furthermore , the structure of the models we use is determined by biological realism rather than being constrained by mathematical tractability . Our approach is most similar to that of [39] , although our methods are less computationally intensive and produce symbolic expressions which allow us to investigate phenomena in greater depth . In particular , we compare alternate modes of gene regulation and readily evaluate analytically the sensitivity of system properties to changes in rate parameters over a large proportion of parameter space .
As a prelude to describing the actual Markov chain model of transcriptional regulation we analyze , we describe a general approach to modeling promoters , enhancers and their interactions , and illustrate this approach with a toy model of transcription that is not too cumbersome to draw – see Figure 1 . We begin with two separate Markov chains , a promoter chain and an enhancer chain ( Figure 1A ) . The states of the promoter chain are the possible configurations of the components involved in polymerase loading onto the promoter ( e . g . “naked DNA” or “DNA–polymerase complex” ) and the allowable transitions correspond to the arrivals of these components , in whichever order is permissible by the underlying biochemistry . The states of the enhancer chain are the the components involved in enhancer activation ( e . g . the binding of regulatory transcription factors to the appropriate cis-control sequence for that promoter ) . Next , to model the regulatory interaction between enhancer and promoter , we designate a particular configuration of the enhancer as the permissive configuration , and specify a particular transition of the promoter chain as the regulated step . We require the enhancer chain to be in the permissive configuration for the promoter chain to make the transition through the regulated step and we assume that the enhancer remains in the permissive configuration as long as the promoter chain is downstream of that step . ( The specification that the enhancer remains in the bound/permissive state while the process is downstream of the regulated step is not the only possible choice , but it is a reasonable one , and one we do not expect to affect our conclusions . ) We choose the regulated step according to the regulation mechanism that we are modeling . The composite stochastic process that records the states of both the promoter and enhancer chains is our resulting Markov chain model of transcription . Varying the regulated step leads to alternative topologies for this chain . We stress that , as we change the choice of regulated step , the underlying promoter and enhancer chains remain the same . In particular , the same set of rate parameters are used in both schemes and they have the same meaning . This permits meaningful comparison of different methods of regulation . Two possible regulated steps , labeled “IR gated” and “ER gated” , are shown along with the corresponding Markov chains in Figure 1 . Each possible configuration of the components of the transcription complex and associated enhancer elements is represented by a state of the composite chain , and the composite chain jumps from one state to another when a single molecular binding or unbinding event converts one configuration of complexes into another . For simplicity , we assume that each arrival in the end state allows one transcript to be made . After transcription occurs , the transcription complex may dissociate entirely , returning the chain to its initial state , or it may leave behind a partial scaffold , returning the composite chain to an intermediate state ( and possibly leading to successive rounds of reinitiation and thus a “burst” of transcription products – i . e . multiple mRNA molecules being transcribed per promoter opening event ) . Formally , the general composite Markov chain model is constructed as follows . Consider two promoter configurations , say , and , such that a direct transition from the first to the second is possible . Write for the rate at which this transition occurs . For any two promoter configurations for which a direct transition is not possible , we set this rate equal to zero . Similarly , we write for the transition rate from enhancer configuration to enhancer configuration . Denote the permissive enhancer configuration by . Suppose that the regulated step of the promoter chain is the step from state to state . Let be the set of states downstream from , i . e . those states that can only be reached from the unbound state by passing through . Then , the composite Markov chain takes values in a set of pairs of configurations , and it jumps from to at rate , defined as follows:and , otherwise . Denote by the expressing promoter configuration with productively elongating mRNA . We are interested in the passage of the composite Markov chain from certain starting states – either the state in which both promoter and enhancer are unbound or the state to which the system returns after elongation begins – to the final , expressing state . Depending on which transition is regulated , some pairs of promoter and enhancer configurations will be unreachable from the relevant starting states; these pairs are biochemically inaccessible and are never visited , and so need not appear in our depictions or in our generator matrices ( e . g . state 2A in the IR-gated model of Figure 1 ) . Because there are generally only two promoters per gene active at the same time in a given nucleus , binding of a general transcription factor ( TF ) at one locus does not decrease the total concentration of the TF in the nucleus sufficiently to affect the rate of binding at the homologous locus . Furthermore , since the observed timescales of variability in induction are shorter than the expected timescale for protein translation and folding , we neglect any feedback from mRNA synthesis which might modify the transition rates . This allows us , in particular , to assume that the jump rates of the Markov chain are homogeneous in time . We now apply this framework to examine a model of transcription that is more interesting and detailed than the toy model used above for illustrative purposes . Many general transcription factors ( TFs ) , such as the protein complexes TFIIA , TFIIB , etc . , function together in a coordinated fashion to form the pre-initiation complex ( PIC ) necessary for the proper activation of transcription [40]–[42] . Experiments with fluorescently labeled TFs in vivo indicate that the components of this complex assemble on the promoter DNA [17] , [43] rather than float freely in the nucleoplasm , as had been previously argued [44] . The steps of PIC assembly are not fully understood [40] , although some important details are known . We analyze the assembly scheme depicted in Figure 2 , which is largely consistent with available data . The promoter is recognized by TFIID , the binding of which allows TFIIA and TFIIB to join the complex [42] . We choose this complex as the first state in our promoter model ( state 1 of Figure 2 ) , since it is only just after this step that the regulation method may differ . TFIIB facilitates the recruitment of RNA polymerase II ( Pol II ) [42] ( state 2 ) . For many non-paused genes , polymerase is only detected in cells that have an activated enhancer ( the cis regulatory sequence which controls expression ) [7] . We call these genes initiation regulated and require that the enhancer reach its permissive state ( ) before this association can occur . Since Mediator is important for many promoter–enhancer interactions [40] , [45] it has likely also joined the complex prior to polymerase arrival . TFIIE , ( state 3 ) , and TFIIF ( state 4 ) , bind next , possibly in either order . Once both are bound ( state 5 ) , TFIIH must also bind ( state 6 ) before Pol II starts synthesizing RNA and clears the promoter [40] , [41] . TFIIH is displaced upon promoter escape [41] , and if Ser 2 of the Pol II tail is not phosphorylated by CDK9 ( pTEFb ) , transcription pauses 40–50 base pairs downstream of the promoter [15] , [45] , [46] ( state 7 ) . For elongation regulated genes , it is the release from this paused state that is possible only in the presence of an activated enhancer ( permissive state ) – which is generally believed to recruit the necessary CDK9 ( and possibly other factors ) . Phosphorylation of Ser 2 allows the fully competent polymerase to proceed through the gene and produce a complete mRNA ( state 8 ) . The transition rates between configurations depend on the energy of association of the bond created and the concentration of the reacting components . Since we are interested in exploring the differences in which step of PIC assembly is regulated and not the different possible modes of enhancer activation , we use a simple abstracted two-state model of enhancer activation . A single transition switches the enhancer from the inactive state to the permissive state . For instance , a transition to the permissive state could represent the binding of a TF to the enhancer . This is not likely to be completely realistic , but if a particular step in the actual dynamics of transcription factor assembly and enhancer-promoter interaction is rate-limiting ( e . g . the looping rate between a bound enhancer and its target promoter ) , then its behavior will be well approximated by our minimal model , with the transition from active to inactive corresponding to the rate for this limiting step . For many paused genes , it is the phosphorylation event which is believed to be regulated [7] , [45] . However , accumulating data suggests the molecular identity of the release factors may vary between paused genes . For example , some also require the recruitment of TFIIS in order to escape a “backtracked” paused state [47] . We consider any such regulation by release from pausing after PIC assembly to be elongation regulation ( ER ) , and any regulation acting upstream of PIC assembly initiation regulation ( IR ) . Finally , the scaffold of transcriptional machinery that facilitates polymerase binding does not necessarily dissociate when transcription begins . Thus , reinitiation may occur by binding new polymerases ( at step 5 ) which must still reload TFIIH which was evicted during promoter escape in order to proceed to step 6 and so on back to step 8 . Repeated cycles of reinitiation may lead to a burst of mRNAs synthesized from a single promoter opening event . We denote by the probability that the scaffold survives to cycle in a new polymerase ( see Figure 2 ) . The scaffold breaks down before the next polymerase arrives with probability , in which case transcription activation must start again from state 1 . We analyze both the time until the first transcript begins ( for which such bursting is irrelevant ) and the effect of this partial stability of the scaffold on cell–to–cell variation in total mRNA . Our aim is not to present a definitive model of PIC assembly itself . Rather , we seek to understand the impact of different modes of regulation on a reasonable model that incorporates sufficient detail and to develop tools that can analyze effectively models of this complexity . We are interested in the speed and variability of the transcription process , as measured , respectively , by the mean , , and variance , , of the delay between induction of the gene and expression of the first functional mRNA transcript . ( Recall that by induction we mean the first time at which all the components required for expression of a particular gene become available , and by expression we mean the time when transcription of the first nascent mRNA transcript begins . ) We use the mean delay to explore the hypothesis that the mechanism of elongation regulation is faster than that of initiation regulation , even when there is no polymerase initially bound ( as reported in [14] ) . The variance of the delay is related to the degree of synchrony of expression of the first transcripts in a population of identically induced cells ( studied in [16] ) – allowing us to test if synchrony is a functional consequence of elongation regulation . We are also interested in the variation between activated cells of the total amount of mRNA produced in each . If we denote by the random number of transcripts produced up until time , then it follows from elementary renewal theory ( see e . g . Section XI . 5 in [48] ) that has mean approximately and variance approximately . A natural measure of relative variability of is the squared coefficient of variation of , ( i . e . the variance of divided by the squared mean of ) , which is thus approximately . We denote the coefficient by , and refer to it as transcript count variability . The transcript count variability provides a measure of the variation in total number of rounds of transcription initiated by identical cells that have been induced for the same amount of time . Note that has units of time: However , the ratio of this quantity for the IR scheme to its counterpart for ER scheme does not depend on our choice of time scale . For any time , this ratio is approximately the ratio of the squared coefficients of variation of for the two schemes , and thus the ratio provides a way of comparing the relative variability in transcript counts between the two schemes across all times . Such a comparison is of interest because many of the known pausing regulated genes are transcription factors or cell signaling components that act in concentration dependent manners , and hence the precision of the total number of transcripts made directly affects the precision of functions downstream [16] . ( Rather than the coefficient of variation , some authors consider the Fano factor of , defined to be [32] . If has a Poisson distribution , then its Fano factor is 1 , and hence a Fano factor that differs from 1 indicates some form of “non-Poisson-ness” . As such , the Fano factor capture a feature of the character of the stochasticity inherent in the number of transcripts made up to some time , whereas the squared coefficient of variation indicates the ( relative ) magnitude of the stochastic effects . ) We use our model to examine how these three important system properties – speed , synchrony , and transcript count variability – depend on the jump rates and how they differ between an IR and an ER regulation scheme . In both cases , the delay between induction and transcription corresponds to the ( random ) time it takes for the corresponding Markov chain to go from an initial state to a final state . For the chains corresponding to the models shown in Figures 1 and 2 , the moments of , the Laplace transforms of , and hence the probability distributions themselves , can be found analytically as we describe briefly here ( for detailed discussion , see the Supporting Information , Text S1; and Figure S3 ) . Denote by the infinitesimal generator matrix that has off-diagonal entries given by the jump rate from state to state , and diagonal entries given by the negative of the sum of the jump rates out of state . The infinitesimal generator of the chain stopped when it hits state is the matrix obtained by replacing the entries in the row of corresponding to with zeros . Writing for the probability density function of , the Laplace transform of is ( 1 ) In principle , the transform can be inverted to find , as we do in Figure 4D . Also , the moment of can be found from the derivative of : ( 2 ) In particular , the mean and variance of can be computed from the first and second derivatives of . It is not necessary to carry out the differentiation in equation ( 2 ) explicitly , since ( 2 ) becomes ( 3 ) after some matrix algebra , as derived in the Supporting Information . Here , is the submatrix of obtained by removing the final row and column . As shown in the Supporting Information , these expressions can be computed much more efficiently than ( 1 ) or ( 2 ) . Equation ( 1 ) is known as the Feynman–Kac formula [28] , and it reduces our problem in principle to inverting the matrix . This is easy to do numerically for particular rate parameter values , but in order to make detailed general predictions about the consequences of changing the step at which the enhancer regulates transcription we require symbolic expressions for the system properties with the rates as free parameters . However , for even moderately complex chains like that described in Figure 2 , symbolic inversion of the matrix is prohibitively difficult for commonly available software . To overcome this obstacle , we develop new analytic techniques that take advantage of the special structure of these matrices . First , we note that chains modeling transcription often have a block structure , in that we can decompose the state space according to the subset of states that must be passed through by any path of positive probability leading from the initial to the final state ( we call such states pinch points ) ( see Figure 2 ) . A schematic of this decomposition is shown in Figure S3 . The models of initiation regulation we consider are amenable to this approach . In order for the ER model to be amenable to this approach , we assume that by the time the PIC assembly has reached the regulated step , the enhancer chain is in ( stochastic ) chemical equilibrium . Concretely , if is the stationary probability that the enhancer is in the permissive state , then at each time the promoter chain jumps to state 7 ( of Figure 2 ) we suppose it jumps to state 7B with probability and to state 7A with probability . ( To evaluate the effect of this approximation , we investigate how our results change after removing the parameter vectors in which the enhancer chain is slow to equilibrate and hence when this approximation is the worst . ) A similar decomposition for elongation regulated genes is possible using spectral theory , but the computational savings are not as great as for the pinch point decomposition . We provide a detailed description of these techniques and the accompanying proofs ( plus implementations coded in MATLAB ) in the Supplemental Text S2 . Our approach has several advantages . Firstly , once we have derived symbolic expressions for features of interest , it is straightforward to substitute in a large number of possibilities for the transition rate vector in order to understand how those features vary with respect to the values of the transition rates . This would be computationally impossible using simulation and at best very expensive using a numerical version of the naive Feynman–Kac approach . Secondly , we are able to differentiate the symbolic expressions with respect to the transition rate parameters to determine the sensitivity with respect to the values of the parameters . It would be even more infeasible to use simulation or a numerical Feynman–Kac approach to perform such a sensitivity analysis .
To get an initial sense of the differences between these two schemes of regulation , we first compared the transcriptional behaviors for a best-guess set of parameters , guided by measurements of promoter binding and escape rates by Darzacq et al . [17] and Degenhardt et al . [49] in vivo and observations in embryonic Drosophila transcription . These data do not allow us to uniquely estimate all 14 binding reaction rates in our model of PIC assembly , but they do constrain key properties , including the time scale of the rate-limiting reactions and the ratio of forward to backward reaction rates for both early binding events and later promoter engagement events . We chose parameters to be consistent with these measurements , and chose enhancer activation and deactivation rates to be consistent with induction times estimated in Drosophila [16] ( which are also in the range recently reported in human cell lines [49] ) . We used the following rate parameters for the model of Figure 2: We found the probability density of the amount of time it takes the system to go from induced to actively transcribing , shown in Figure 3A , by numerical inversion of the Laplace transform ( equation 1 ) . With these rate parameters , the mean time between induction and the start of transcription for an elongation regulated scheme is around 5 minutes , with a standard deviation of about 4 minutes , whereas an initiation regulated scheme with the same rate parameters has a mean of 16 minutes and a standard deviation of 12 minutes , consistent with experimentally estimated initiation times in Drosophila [16] . We also described the number of mRNA produced over a given period of time at one choice of ( the probability the GTF scaffold dissociates before the return of the next polymerase ) . Setting , we found the distribution of the time delay between the beginning of the production of subsequent transcripts under each model . Using this distribution , we simulated the number of mRNA produced during a 600 minute period in 2000 independent cells , under both the IR and the ER scheme ( for the common vector of rate parameters listed above ) . The resulting distributions of mRNA numbers are shown in Figure 3B and C . To depict the amount of variability this represents , Figures 3D and E show a cartoon of the results – for each cell pictured , we sampled a random number of mRNA as above , which are shown red dots randomly scattered within the cell . To emphasize the variability , we then colored cells blue that have less than two-thirds the mean mRNA number and colored cells red that have more than three halves the mean mRNA number . In this example , is 2 . 8 times larger in the ER model than in the IR model , so these simulations also give a sense of how a given ratio of transcript count variabilities for the two schemes corresponds to a difference in cell-to-cell variability of transcript counts , a topic we explore in more detail below . Our predictions for the time of expression and the number of transcripts in the previous subsection depended on the chosen parameter values such as the association rate of different GTFs and the average burst size of the gene expression . The values of such parameters can , for the most part , be only very approximately estimated . Moreover , they may be expected to vary considerably between different genes and different species . Since a single vector of parameters simultaneously specifies our models for the two regulation mechanisms , we can systematically explore all possible combinations of promoter strength and enhancer activation rates and ask in each of these cases how the two mechanisms compare in terms of speed , synchrony and variability in transcript counts . To compare the two kinds of regulation of the model in Figure 2 , we sampled 10 , 000 random vectors of transition rates and substituted them into our analytic expressions for , , and , with each rate chosen independently and uniformly between 0 and 1 ( we could also have used a regular grid of parameter vectors ) . Since we will use ratios of the relevant quantities to compare models , and these ratios are all invariant under a common linear rescaling of time , the fact that all rates are bounded by 1 is no restriction – we are effectively sampling over all of parameter space . ( For instance , the ratio of mean expression times of the two models does not change after multiplying every rate parameter by 100 . ) Furthermore , independent draws of new sets of 10 , 000 parameter vectors and substitutions give nearly identical results , confirming that our results are not sensitive to the specifics of the sample . Additionally , discarding parameter vectors for which the enhancer dynamics are significantly slower than for the promoter chain ( i . e . or is smallest ) does not qualitatively change any of the results , validating our treatment of the enhancer chain when analyzing the ER scheme . In Figure 4A–C we plot the histogram of ratios for the mean delay , variance in delay , and transcript count variability for the 10 , 000 randomly selected parameter combinations sampled uniformly across parameter space . We found that at all sampled choices of rate parameter , and therefore in the vast majority of parameter space , the time to the first transcription event after induction is smaller and less variable ( i . e . more synchronous ) for elongation regulation than for initiation regulation in the realistic model of Figure 2 . Thus , both the experimentally reported speed [14] and synchrony [16] for elongation regulated genes can be expected purely from effects of regulation topology without invoking changes in promoter strength or in the composition of the PIC . We emphasize that this conclusion is still consistent with the possibility that a particular initiation regulated gene is expressed in a more synchronous pattern or with more rapid kinetics than some other elongation regulated gene: it is only necessary that the rate parameters are also sufficiently different . However , for the fixed set of rates associated with a given gene , the network topology of the ER scheme always improved synchrony and speed in our model of transcription relative to the corresponding IR scheme for the parameter vectors we sampled . There is a plausible intuitive explanation for why elongation regulation is almost always faster than initiation regulation ( Figure 4A ) . When the regulation acts downstream , there are multiple paths which the system can take to before it reaches the regulated step – ( i . e . either the enhancer can reach the permissive state first or the polymerase can load ) , as illustrated for the simple model in Figure 1A and B . The system moves closer to the endpoint with whichever happens first , whereas the IR regulated scheme must wait for enhancer activation before proceeding . The combination of this intuition and our strong numerical evidence suggests a provable global inequality . However , recall that for the toy model IR is faster over about 6% of parameter space , and one can reduce the realistic model to the toy model by making appropriate transitions very fast . For example , for the toy model the choice of parametersleads to a 5 fold increase in speed of the IR scheme relative to the ER scheme . Figure S2 shows the distribution of values for each parameter in the parameter sets where the IR scheme is faster ( see also Text S1 ) . This allows us to find parameter vectors where IR is faster than ER for the realistic model , for instance , produces in the realistic model a 10 fold increase in speed for the IR scheme relative to the ER scheme . However , such reversals of the typical ordering must occur over less than one ten-thousandth of parameter space . The fact that the typical ordering is not universal and hence not the consequence of some analytically provable domination of one model by the other demonstrates the necessity of our numerical exploration of parameter space . The effect of the regulatory scheme on the variation in the total amount of expression among cells is perhaps the most interesting and also experimentally untested consequence of regulating release from the paused state . As discussed above , we compute a factor for each scheme and compare the schemes by examining the ratio of the resulting quantities . If the ratio is larger than one at a particular set of parameter values , a population of cells using the IR scheme with those rate parameters will show more variability in mRNA concentrations between cells ( relative to the average over all cells ) than if they were using the ER scheme with the same rate parameters . In this case , we say that the ER scheme is more consistent than the IR scheme . We explored the logarithm of this ratio ( equivalently , the difference of the logarithms of the respective quantities ) at four different values of ( the probability the scaffold does not disassemble; see Figure 2 ) ; several of the resulting distributions are shown in Figure 5 . When the complex is very stable , so that all polymerases find a preassembled scaffold to return to ( , Figure 5A ) , the ER scheme is more consistent for most rate parameters , but the differences are small . In fact , in nearly all cases at which differs by a factor of at least 2 , the IR scheme is the more consistent . When the scaffold is still stable but less so ( , Figure 5A; mean burst size 10 ) , the ER scheme still almost always produces more consistent numbers of transcripts among cells than the IR scheme , and the differences are much larger . If the scaffold is less stable ( , Figure 5C; mean burst size 1 . 4 ) , the ER scheme is still more often more consistent than the IR scheme . When we consider the simplest case with no bursting ( , Figure 5D ) , the ER scheme produces less variation in total transcript ( smaller ) for most of parameter space . Moreover , the distribution is strongly skewed to the right , to the extent that for the 20% of parameter space where there is more than a 1 . 5 fold difference between the two regulatory mechanisms the ER scheme is always less variable . We have found that , regardless of the value of , the ER scheme is more consistent over most of parameter space . However , for that difference in consistency to be substantial , must not be too close to 1 . This is at first surprising , because if the scaffold remains assembled , so that the chain returns to state 5 of Figure 2 , an IR scheme seems to have a clear “advantage” – it does not have to wait for the enhancer to arrive , whereas the ER scheme does , and one might expect that this added stochastic event would only increase variability . Consideration of how each chain depends on its starting state suggests an intuitive explanation for this difference . The IR scheme differs more in the amount of time it takes to reach the synthesis state when started with or without a scaffold ( state 5 or state 1 ) than does the ER scheme . Intermediate values of allow the possibility of some cells making many bursts by reverting to state 5 after each synthesis while other cells make dramatically less by reverting to state 1 after each synthesis . In contrast , under the ER regulation scheme , cells that start again from state 1 or from state 5 have relatively more similar synthesis times , and thus relatively less variation . The similar synthesis times result from the fact that ER is faster starting from state 1 , for the reasons discussed above , and slower than IR when starting from state 5 , because of the extra regulatory step before synthesis . Consequently , an ER scheme reduces the noise associated with very stable transcription scaffolds ( see [30] , [32] , [34] for a discussion of this noise ) . To further understand why elongation regulation results in faster , more synchronous , and more consistent gene expression over a wide range of parameters we investigated alternative post-initiation regulatory schemes . This allows us to explore how changing certain properties of the model of PIC assembly ( the promoter chain ) will affect the results: Is the difference large because there are many steps between the IR step and the ER step , or is it because there is no allowed transition leading backward out of the state immediately before the regulated step ? To explore these questions , we made modifications to the toy model of Figure 1 which we are able to analyze without the assumption of enhancer equilibrium . First note that , as is shown in Figure 1C , the ER model is still faster , less variable , and more reliable ( smaller , , and ) than the IR model over approximately 95% of parameter space . ( It is also reassuring that the results are so similar to those for the more realistic model . ) We performed the same analysis after adding a reverse transition from state 3 back to state 2 ( see Figure S4A–B ) . The results are shown in Figure S4C , and demonstrate that there is strikingly little difference between the two models of regulation . This suggests that the absence of a backwards transition from the state immediately preceding the regulated transition is an important factor in producing the differences between the models we observed above . In the ER scheme of Figure 1 , PIC assembly becomes “caught” in state 3 , awaiting arrival of the enhancer . ( Similarly , the ER scheme of Figure 2 gets “caught” in state 7 ) . After adding a transition , PIC assembly may run up and down the chain many times before it is in state 3 at the same time the enhancer is in the permissive configuration , and this counteracts any benefits in speed or reliability that may have been gained otherwise . ( It is not obvious that this will happen: the ER scheme of Figure S4B still has “more routes” from state 1A to state 4 than the IR model , so it may run counter to intuition that the IR model could be so often faster . ) This furthermore suggests that regulating after a state in which PIC assembly is “caught” reduces variation – some polymerases may run from state 1 to 8 smoothly and fire very quickly , while others may go up and down the assembly process many times before they actually escape the promoter and make a transcript ( as is suggested by the data of Darzacq et al . [17] ) , and this will substantially spread out the times at which the first transcript is created . We also investigated the case in which the transition is regulated and observed a similar pattern – see Figure S4D–F . This investigation supports the intuition that it is the stability of the paused state , not simply the parallel assembly of enhancer complex and promoter complex , that is most important in understanding the different behavior of the two regulatory schemes . It also suggests that these differences should be specific to genes that are regulated through paused ( as opposed to poised or stalled ) polymerase . Small variations in rate parameters between cells will occur if the number of TF or Pol II molecules is small , so it is of interest to investigate how robust the properties of each regulation scheme are to such variation and which jump rates affect each scheme the most . To measure this sensitivity , we compute the gradient of a quantity of interest ( e . g . the mean induction speed ) with respect to the vector of jump rates , square the entries , and normalize so that the entries sum to one , giving a quantity we refer to as relative sensitivity that is analogous to the “percent variation explained” in classical analysis of variance . Our analytic solutions for the quantities of interest make this computation possible . For example , let denote the mean transcription time of the chain when the vector of transition rates is . Then , the relative sensitivity of to each rate is . The larger this quantity is , the larger is the relative effect a small change in has on . To explore the sensitivity across parameter space , we computed relative sensitivities for each of the three system properties to all 16 parameters at each of the 10 , 000 random vectors of transition rates described above . Each of the system properties showed surprisingly similar sensitivity profiles , so we only discuss the results for the mean time to transcription . Marginal distributions of sensitivity of mean time to transcription to each parameter are shown in Figure 6 . Corresponding plots for the variance of transcription time and for transcript count variability are shown in Figures S5 and S6 . As one might expect , for a given parameter vector the parameters to which the behavior of the models are most sensitive are generally those that happen to take the smallest value ( and are thus rate-limiting ) : for each parameter vector , we recorded the sizes of the two parameters with the highest and second highest sensitivity values and found that their sample means were and , respectively ( whereas the sample mean of a typical parameter value will be very close to ) . However , just how small a given transition rate must be before it controls the system properties depends on where the corresponding edge lies in the topology of the network . As shown in Figure 6 , some parameters are relatively important throughout a large region of parameter space in both the ER and IR schemes , while others only dominate the response of the system in a small portion and some never appear . Two further observations are evident from this analysis . First , we see which transitions in the process of activating the gene are most sensitive to small fluctuations ( due to small number of TF molecules or changes in binding strength ) . As is apparent from Figure 6 , just 4 of the 16 promoter chain jump rates dominate the sensitivity , and these are the same for both IR and ER schemes ( , , , and ) . The relative importance among those 4 jump rates depends on the position in parameter space , primarily through their relative sizes . Furthermore , although the ER and IR schemes have otherwise similar sensitivity profiles , the IR scheme is additionally sensitive to variation in the rate of enhancer–promoter interactions , . As this interaction between potentially distant DNA loci is likely rate-limiting for gene expression , the robustness of the elongation regulated scheme to fluctuations of this rate may provide a further explanation for why elongation regulated genes appear to exhibit considerably more synchronous activation . It suggests additionally that the rate of enhancer–promoter interactions is under more selective pressure for IR genes , where it has a large effect on their expression properties , than it is for ER genes , which may exhibit very similar expression properties despite having different enhancer interaction rates . Second , we also observe that the complex assembly steps which may occur in arbitrary arrival order , namely the recruitment of TFIIE or TFIIF ( governed by the jump rates , , , and ) are considerably more tolerant to stochastic variation than sequential assembly steps such as the initial recruitment of the polymerase ( ) , the arrival of the last component of the complex , TFIIH ( ) , or promoter escape ( ) . Although between–cell variation in the total concentration of these intermediate , non-sequential binding factors will affect their binding rate parameters , it will not greatly change properties of the time to expression , thus suggesting an additional benefit of ER . This observation leads to the conclusion that the regulatory processes controlling the concentration of factors arriving in arbitrary order and the binding affinities of such factors may be under less evolutionary pressure than the corresponding quantities for factors associated with other transitions .
Speed , synchrony , degree of cell–to–cell variability , and robustness to environmental fluctuations are important features of transcription . They are properties of the system rather than of a particular gene , DNA regulatory sequence , or gene product taken in isolation , and optimizing them can , for instance , reduce the frequency of mis-patterning events that arise due to the inherently stochastic nature of gene expression . Understanding how these properties emerge , the mechanism by which they change , and the tradeoffs involved in optimizing them all require tractable models of transcription . Through a study of stochastic models of transcriptional activation , we demonstrated that the increased speed and synchrony of paused genes , reported by Yao et al . [14] and Boettiger et al . [16] respectively , are expected consequences of the elongation regulation shown by such genes . We also predicted that ER genes produce more consistent numbers of total transcripts than IR genes . This hypothesis can be tested directly using recently developed methods ( see [19] , [50] for reviews and the Supporting Information for more details ) . We furthermore explored what aspects of ER make this possible . From an examination of the effect of scaffold stability we proposed that elongation regulation should reduce the noise-amplifying nature of bursty expression . By investigating alternative models of post-initiation regulation , we also determined that our predictions depend critically on the stability of the transcriptionally engaged , paused polymerase , and would not be expected from polymerases cycling rapidly on and off the promoter ( i . e . polymerase stalling ) . Our investigation required us to introduce a general probabilistic framework for analyzing system properties of protein–DNA interactions . Stochastic effects , resulting from molecular fluctuations , are increasingly understood to play important roles in gene control and expression ( see [18] for a review ) . We can now determine quantitatively how an element's location in a network affects the general properties of that network , even when the rate constants and concentrations of the network components are unknown . In particular , we quantified the extent to which system properties are sensitive to each rate parameter , something which might predict the evolutionary constraint on that component . Most previous approaches to the analysis of protein–DNA interactions have either relied on simulations , which require some knowledge of numerical rate values , or use the fluctuation–dissipation theorem assuming the system is near equilibrium and the noise is small . Our methods avoid the limitations of those approaches and also make analysis of realistic models , as done in [39] , significantly more feasible . Finally , our approach is not restricted to investigating the assembly of transcriptional machinery , but may also prove useful in studying stochastic properties of a variety of regulatory DNA sequences ( such as enhancers ) . Different assembly topologies , such as sequential versus arbitrary association mechanisms for the component TFs [40] , may account for some of the observed differences in sensitivities and kinetics between otherwise similar regulatory elements . As new technologies allow better experimental determinations of these mechanisms , a theoretical framework within which one can explore their potential consequences will become increasingly important .
|
Gene activation is an inherently random process because numerous diffusing proteins and DNA must first interact by random association before transcription can begin . For many genes the necessary protein–DNA associations only begin after activation , but it has recently been noted that a large class of genes in multicellular organisms can assemble the initiation complex of proteins on the core promoter prior to activation . For these genes , activation merely releases polymerase from the preassembled complex to transcribe the gene . It has been proposed on the basis of experiments that such a mechanism , while possibly costly , increases both the speed and the synchrony of the process of gene transcription . We study a realistic model of gene transcription , and show that this conclusion holds for all but a tiny fraction of the space of physical rate parameters that govern the process . The improved control of cell-to-cell variations afforded by regulation through a paused polymerase may help multicellular organisms achieve the high degree of coordination required for development . Our approach has also generated tools with which one can study the effects of analogous changes in other molecular networks and determine the relative importance of various molecular binding rates to particular system properties .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"computational",
"biology/transcriptional",
"regulation",
"genetics",
"and",
"genomics/gene",
"expression",
"biophysics/theory",
"and",
"simulation",
"biochemistry/theory",
"and",
"simulation",
"biochemistry/transcription",
"and",
"translation",
"mathematics/statistics",
"biophysics/transcription",
"and",
"translation"
] |
2011
|
Transcriptional Regulation: Effects of Promoter Proximal Pausing on Speed, Synchrony and Reliability
|
Peroxisomes are subcellular organelles involved in lipid metabolic processes , including those of very-long-chain fatty acids and branched-chain fatty acids , among others . Peroxisome matrix proteins are synthesized in the cytoplasm . Targeting signals ( PTS or peroxisomal targeting signal ) at the C-terminus ( PTS1 ) or N-terminus ( PTS2 ) of peroxisomal matrix proteins mediate their import into the organelle . In the case of PTS2-containing proteins , the PTS2 signal is cleaved from the protein when transported into peroxisomes . The functional mechanism of PTS2 processing , however , is poorly understood . Previously we identified Tysnd1 ( Trypsin domain containing 1 ) and biochemically characterized it as a peroxisomal cysteine endopeptidase that directly processes PTS2-containing prethiolase Acaa1 and PTS1-containing Acox1 , Hsd17b4 , and ScpX . The latter three enzymes are crucial components of the very-long-chain fatty acids β-oxidation pathway . To clarify the in vivo functions and physiological role of Tysnd1 , we analyzed the phenotype of Tysnd1−/− mice . Male Tysnd1−/− mice are infertile , and the epididymal sperms lack the acrosomal cap . These phenotypic features are most likely the result of changes in the molecular species composition of choline and ethanolamine plasmalogens . Tysnd1−/− mice also developed liver dysfunctions when the phytanic acid precursor phytol was orally administered . Phyh and Agps are known PTS2-containing proteins , but were identified as novel Tysnd1 substrates . Loss of Tysnd1 interferes with the peroxisomal localization of Acaa1 , Phyh , and Agps , which might cause the mild Zellweger syndrome spectrum-resembling phenotypes . Our data established that peroxisomal processing protease Tysnd1 is necessary to mediate the physiological functions of PTS2-containing substrates .
Peroxisomes are subcellular organelles that are involved in the catabolism of very-long-chain fatty acids ( VLCFAs ) , branched-chain fatty acids , D-amino acids , polyamines and the biosynthesis of bile acids [1]–[3] . Abnormalities of peroxisomal biogenesis or enzymes cause dysfunctions of the peroxisomal metabolism [3] . Clinically , peroxisomal disorders are divided into two large groups: Zellweger Syndrome spectrum ( ZSS ) and deficiency of peroxisomal enzymes [3] . ZSS is caused by defects of PEX ( peroxisomal biogenesis factor ) gene family members that interfere with or abrogate the biogenesis resulting in abnormally shaped peroxisomes or peroxisome deficiency [3]–[5] . In the case of ZSS peroxisome-targeted proteins are present in the cytosol , but most peroxisomal matrix proteins are not properly processed [6] , [7] . ZSS includes neonatal adrenoleukodystrophy , infantile Refsum disease , rhizomelic chondrodysplasia punctata ( RCDP ) type 1 and Zellweger syndrome , the most severe form [3] . RCDP type 1 disease is caused by mutations in PEX7 that interfere with its function as a receptor in targeting PTS2-containing proteins ACAA1 ( acetyl-CoA acyltransferase 1 ) , AGPS ( alkylglycerone phosphate synthase ) and PHYH ( phytanoyl-CoA 2-hydroxylase ) to the peroxisomes . The mutated PEX7-mediated effects result in the accumulation of VLCFAs , phytanic acid and a reduced plasmalogen synthesis [8] . Peroxisomal matrix proteins are imported from the cytoplasm into peroxisomes through PTSs [9] , [10] . The majority of peroxisomal enzymes have a C-terminal PTS1 signal [SA]-K-L [11] , that has been extended to a dodecamer motif [12] . PTS1 is recognized by the cytosolic soluble receptor Pex5p that carries the cargo to the peroxisomal membrane . A few peroxisomal proteins are targeted via the N-terminal PTS2 signal [RK]-[LVI]-[X5]-[HQ]-[LAF] [13] which is recognized by Pex7 . Recently we identified the peroxisomal processing protease , Tysnd1 ( trypsin domain containing 1 ) [14] . Tysnd1 is localized in peroxisomes and cleaves PTS2-containing prethiolase Acaa1 , PTS1-containing proteins Acox1 ( acyl-CoA oxidase 1 , palmitoyl ) , Hsd17b4 ( hydroxysteroid ( 17-beta ) dehydrogenase 4 ) and ScpX [14] , the longer mRNA product of gene Scp2 ( sterol carrier protein 2 , liver ) [15] . Scp2 lacks the thiolase domain-containing amino acids 1–404 of ScpX , but shares its C-terminal sequence ( 405–547 ) [15] , [16] . Since Acox1 , Hsd17b4 and ScpX are pivotal in the peroxisomal β-oxidation of VLCFAs , we created Tysnd1−/− mice to investigate the in vivo functions of Tysnd1 , and to assess whether the phenotype would resemble any of the clinical features of human single peroxisomal enzyme deficiencies .
Tysnd1 was disrupted by targeted constitutive deletion of exons 2 and 3 , encoding amino acids 392–496 of peptidase cysteine/serine , trypsin-like domain ( 333–537 ) , using CRE/LoxP technology ( Figure 1A and 1B ) . The Tysnd1−/− mice were obtained by crossing Tysnd1+/− mice . The ratio of homo- and heterozygote mice followed Mendel's law ( 1 . 08 vs . 2 . 18 for 285 homozygote mutants and 574 heterozygotes , compared with 1 . 0 for 263 wild-type mice ) . Tysnd1 mRNA and protein expression were completely disrupted in the Tysnd1−/− mice ( Figure 1C and 1D ) . The phenotypes of female and male Tysnd1−/− mice of different litters were analyzed at 7–26 weeks of age using Japan Mouse Clinic ( JMC ) pipeline 1 [17] , which includes modified-SmithKline Beecham , Harwell , Imperial College , Royal London Hospital phenotype assessment ( modified-SHIRPA ) [18] at eight weeks of age . Tysnd1−/− mice of both sexes did not display any anomalies with regard to body weight ( eight weeks ) , body mass index ( BMI , eight weeks ) , haematology ( nine weeks ) , urine composition ( ten weeks ) , anatomy ( 26 weeks ) , clinical blood biochemistry ( eleven and 18 weeks ) , insulin tolerance ( 13 weeks ) , oral glucose tolerance ( 14 weeks ) , blood pressure ( 21 weeks ) , open field behaviour ( seven weeks ) and dual energy X-ray absorptiometry ( 22 weeks ) compared with wild-type mice ( data not shown ) . Tysnd1−/− male mice fed with high-fat diet between week 5 and 15 after birth did not display significant changes in body weight ( Figure S1A ) , body length ( Figure S1B ) and BMI ( Figure S1C ) compared with age-matched male wild-type controls . Home cage activity [19] tested with 34 weeks old male mice was significantly reduced in male Tysnd1−/− mice ( data not shown ) . The test was a component of a JMC-independent energy metabolism screen . Western blots of liver extracts prepared from 18 weeks old male Tysnd1−/− mice showed only increased amount of unprocessed Acaa1 , Acox1 , ScpX and Hsd17b4 , whereas liver extracts from age- and gender-matched Tysnd1+/+ and Tysnd1+/− mice contained both processed and unprocessed forms ( Figure 1E ) . Electron microscopy ( EM ) image analysis ( Figure S2A ) of 29 weeks old male Tysnd1−/− liver sections revealed that the number of peroxisomes per counted area almost doubled compared with age- and gender-matched wild-type mice ( Figure S2B ) . The slight increase of peroxisome size in Tysnd1−/− liver was statistically not significant ( Figure S2C ) . The peroxisomal β-oxidation activity measured by [1-14C]lignoceric acid oxidation in liver homogenates of 15 weeks old Tysnd1−/− male mice was reduced to approximately 60% of male wild-type mice of same litters ( Figure 1F ) . The decreased β-oxidation activity did not affect blood serum VLCFA levels in adult ( 38–39 weeks ) Tysnd1−/− male mice maintained on CE-2 diet compared with controls ( data not shown ) . Repeated , independently conducted mating of wild-type female mice with male Tysnd1−/− mice produced no offspring ( Table S1A ) . Mating pairs consisting of female Tysnd1−/− and wild-type or heterozygote males led to normal pregnancies and litters . The results lent support to our earlier formulated claim [14] that Tysnd1 may affect male fertility at the sperm level . The analysis of spermatogenesis revealed sperms with abnormally round-shaped heads in the seminiferous tubules of Tysnd1−/− mice ( Figure 2A ) . Epididymal sperms of Tysnd1−/− mice showed coiled axonemes ( Figure 2B ) , abnormal anterior acrosome lacking the acrosomal cap ( Figure 2D ) . The defects were confirmed by staining with peanut agglutinin ( PNA ) lectin-conjugated FITC ( Figure 2E ) and by EM image analysis ( Figure 2F and 2G ) . Since the anterior acrosome anomaly may affect the acrosome reaction we conducted sperm penetration assays using in vitro fertilization under intact cumulus mass and zonae pellucidae-free conditions ( Table S1B ) . When the cumulus mass was intact , fertilization by Tysnd1−/− sperms was significantly reduced . Phospholipids are primary components of cellular membranes . Since the deciduation of sperm acrosomes occurred only in Tysnd1−/− mice ( Figure 2D and 2E ) , we hypothesized that alterations in the phospholipid composition causes the fragility of acrosomal membrane . Plasmalogens are major components of the acrosomal membrane [20] . The first two steps of plasmalogen synthesis , which are catalyzed by Gnpat ( glyceronephosphate O-acyltransferase ) and Agps occur in the peroxisomes [21] . We assessed the effect of Tysnd1 loss on plasmalogens by measuring the ethanolamin or choline plasmalogen species composition of the vinyl ether bound fatty alcohol at sn-1 and the ester bound fatty acids at sn-2 position of the glycerol backbone using whole testes and epididymides extracts of ten to eleven weeks old Tysnd1−/− and wild-type males . Although we did not find significant differences in total plasmalogen levels between Tysnd1−/− and wild-type mice ( Figure S3 ) , we detected differences in the ratio and composition of certain plasmalogens . In testes ( Figure 3A and 3B ) palmitic acid-oleic acid ( 16∶0–18∶1 ) prevailed among both choline and ethanolamine plasmalogens with 16∶0–18∶1 slightly decreased in Tysnd1−/− mice . Choline- and ethanolamine-type plasmalogens were reduced 8 . 0% and 14 . 7% , respectively . Among Tysnd1−/− epididymal ethanolamine plasmalogens we observed a 16 . 7% decrease in 16∶0–18∶1 and a 10 . 5% decrease in 16∶0–20∶4 palmitic acid-arachidonic acids compared with wild-type controls ( Figure 3 ) . Epididymal choline plasmalogens were only reduced at 16∶0–20∶4 levels ( 16 . 5% decrease ) ( Figure 3 ) . In contrast , plasmalogen containing DPA ( docosapentanoic acid; 22∶5 ) or DHA ( docosahexaenoic acid; 22∶6 ) at sn-2 position were slightly higher in testes and epididymides of Tysnd1−/− mice than in wild-type mice ( Figure 3 ) . The Tysnd1 substrates Hsd17b4 and ScpX catalyze the oxidation of branched-chain fatty acids . In human , defects of Hsd17b4 cause various neurological abnormalities [22] , limb abduction and hypotonia . In Scp2- and Hsd17b4-deficient mice phytanic acid accumulates in the liver when its precursor phytol is orally administered [23] , [24] . We tested whether Tysnd1−/− mice would show a similar abnormal biochemical profile . After determining phytanic acid levels in the blood serum of 38–39 weeks old male Tysnd1−/− mice we found that phytanic acid levels were significantly higher in Tysnd1−/− mice than in age-matched male wild-type mice ( Figure S4A ) . In the absence of visible macroscopic abnormalities , we performed a phytol ( 15 mg/day ) overloading experiment over a period of 13–14 days with eight weeks old mice . All female Tysnd1−/− mice on the phytol-containing diet died after one week and were not further analyzed . Male Tysnd1−/− mice lost approximately 20% of their body weight compared with 5% in wild-type mice ( Figure S4B and S4C ) . Phytol-fed male Tysnd1−/− mice had beige livers ( Figure 4A ) that contained a two times greater amount of total fat ( Figure S4D ) and triglycerides ( Figure S4E ) than controls , indicating the onset of a fatty liver phenotype . The liver parenchyma of Tysnd1−/− mice appeared inflamed and was infiltrated by giant cells ( Figure 4B ) . Phytanic acid accumulated in the sera of phytol-administered male Tysnd1−/− mice more than 100-fold compared with controls ( Figure 4C ) . Plasma pristanic acid , a metabolite of phytanic acid was elevated in ten weeks old male Tysnd1−/− mice that were fed for ten days with phytol , but statistically not significant when compared to wild-type mice ( Figure S4G ) . The blood serum of phytol-fed Tysnd1−/− mice showed a significant accumulation of VLCFAs ( Figure 4D ) compared with wild-type mice and controls on a carboxy methyl cellulose ( CMC ) diet . After phytol feeding , the ratios of tetracosanoic acid ( C24∶0 ) docosanoic acid ( C22∶0 ) , pentacosanoic acid ( C25∶0 ) and hexacosanoic acid ( C26∶0 ) to docosanoic acid ( C22∶0 ) increased between Tysnd1+/+ mice from 0 . 66 for C24∶0 , 0 . 017 for C25∶0 and 0 . 0011 for C26∶0 to 1 . 41 , 0 . 062 and 0 . 0034 in Tysnd1−/− mice , respectively . The reasons for the somewhat peculiar higher C25∶0/C22∶0 than C26∶0/C22∶0 ratios in both , wild-type and Tysnd1−/− mice are unknown . The increase in VLCFAs after phytol feeding of Tysnd1−/− mice was accompanied by a significantly reduced liver peroxisomal β-oxidation rate compared with CMC-fed controls and phytol-fed wild-type mice ( Figure 4E ) . Liver mitochondrial β-oxidation was not affected after phytol feeding of Tysnd1−/− mice ( Figure S4F ) . EM images of liver tissue sections of CMC-fed Tysnd1−/− mice showed enlarged peroxisomes ( Figure 4F ) compared with wild-type mice . In addition , we observed in phytol-fed Tysnd1−/− mice a significant decrease in the number of liver peroxisomes and an increase in autophagosomes ( Figure 4F and 4G ) . Since the mitochondria appeared intact ( Figure 4F; Figure S4F ) , the autophagosomes are likely to be of peroxisomal origin and of the macropexophagy type . The observed changes indicate that phytol administration to Tysnd1−/− mice causes substantial peroxisomal dysfunctions . Clinical blood biochemical analyses of Tysnd1−/− mice administered with phytol showed significantly elevated levels of lactate dehydrogenase ( LDH ) , aspartate transaminase ( AST ) and alanine aminotransferase ( ALT ) that are indicative of liver damage ( Table S2 ) . Gnpat , Agps , Far1 ( fatty acyl CoA reductase 1 ) and Far2 ( fatty acyl CoA reductase 2 ) are enzymes involved in plasmalogen synthesis . Phytanic acid oxidation depends on Phyh , whereas Amacr ( alpha-methylacyl-CoA racemase ) is involved in the β-oxidation of pristanic acid [25] . We co-transfected COS-7 cells with Agps and Tysnd1 or Phyh and Tysnd1 while increasing the amount of Tysnd1 to evaluate its substrate processing . The unprocessed forms of Agps and Phyh decreased ( Figure 5A , 5D ) in proportion to the increase of Tysnd1 . Since the amount of processed Phyh increased , Tysnd1 seems to process Phyh directly . The amount of processed Agps increased only in presence of proteasome inhibitor MG132 indicating possible degradation via the ubiquitin-proteasome pathway ( Figure 5A , 5B ) . Western blot analyses of testes and liver extracts support in vivo processing of Agps and Phyh by Tysnd1 . The processed forms of Agps and Phyh were present in wild-type and heterozygous mice , but absent in the extracts of Tysnd1−/− mice ( Figure 5C , 5E ) . Furthermore we tested whether Tysnd1 can process Gnpat , Amacr , Far1 and Far2 . Peroxisomal membrane-bound Far1 and Far2 are involved in plasmalogen synthesis , but localized to peroxisomes in an apparently PTS1/PTS2-independent manner [26] . Tysnd1 co-transfection experiments of COS-7 cells with Gnpat , Amacr , Far1 and Far2 demonstrated that the amount of unprocessed Gnpat and Far2 decreased ( Figure S5A , S5C ) . The in vivo evidence of Gnpat and Far2 as Tysnd1 substrates remained ambiguous due to faint Western blot signals ( data not shown ) . Far1 and Amacr were not affected by Tysnd1 co-transfection , implying that these enzymes are not substrates of Tysnd1 ( Figure S5B , S5D ) . Assuming that Tysnd1 processing of peroxisomal proteins is essential for the their localization to peroxisomes , we assayed the localization of each Tysnd1 substrate , Acaa1 , Phyh , Agps , Acox1 , Hsd17b4 , and ScpX by co-transfecting expression vector constructs for the substrate-GFP fusion proteins with peroxisomal location marker DsRed2-Peroxi ( PTS1 ) into primary hepatocytes of six weeks old Tysnd1−/− and wild-type male mice . The subcellular localization was assessed by confocal laser-scanning microscopy 27–28 hours after transfection ( Figure 6A–6F ) . In wild-type mice hepatocytes all six GFP fusion proteins unequivocally co-localized with DsRed2-Peroxi , indicating their peroxisomal localization ( Figure 6A–6F; Figure S6A–S6F ) . In contrast , in Tysnd1−/− hepatocytes PTS2- containing GFP fusion proteins Acaa1 , Phyh and Agps ( Figure 6A–6C; Figure S6A–S6C ) co-localized to a noticeably lesser degree with DsRed2-Peroxi than PTS1-containing GFP fusion proteins Acox1 , Hsd17b4 and ScpX ( Figure 6D–6F; Figure S6D–S6F ) , which showed mainly peroxisomal localization . In Tysnd1−/− hepatocytes , successfully transfected with Acaa1-GFP , some of it was observed to co-localize in part with peroxisomes and some remained outside the peroxisomes ( Figure 6A; Figure S6A ) . Most of Phyh-GFP appeared as punctuated structures with some co-localized with DsRed2-Peroxi ( Figure 6B; Figure S6B ) . Almost all Agps-GFP did not co-localize with DsRed2-Peroxi , indicating mostly non-peroxisomal localization ( Figure 6C; Figure S6C ) . In a control experiment we also tested the suitability of DsRed2-Peroxi ( PTS1 ) as peroxisomal co-localization marker in Tysnd1−/− hepatocytes by comparing its co-localization with anti-Pmp70 Alexa Fluor 488 , an antibody against peroxisomal membrane marker Abcd3 ( ATP-binding cassette , sub-family D ( ALD ) , member 3 ) also called Pmp70 . Both markers co-localized with peroxisomes , confirming that the peroxisomal localization properties of DsRed2-Peroxi itself were not affected in Tysnd1−/− hepatocytes ( Figure S7A ) . Since co-transfection of DsRed2-Peroxi with Tysnd1 substrates expressed as GFP fusion proteins might strain the peroxisomal protein import capacity we evaluated the co-localization of singly transfected Acaa1-GFP , Phyh-GFP and GFP-Hsd17b4 with anti-Pmp70 Alexa Fluor 568 . Under singly-transfection conditions , Acaa1-GFP ( Figure S7B ) and Phyh-GFP ( Figure S7C ) co-localized with peroxisome marker anti-Pmp70 in both Tysnd1−/− and Tysnd1+/+ hepatocytes . Similarly for GFP-Hsd17b4 , we did not observe any differences in the peroxisomal co-localization with anti-Pmp70 between Tysnd1−/− and Tysnd1+/+ hepatocytes ( Figure S7D ) . Altogether , the localization of PTS2 matrix proteins in Tysnd1−/− hepatocytes was only affected when the PTS1-containing DsRed2-Peroxi marker was co-transfected . Western blot analysis of the liver subcellular fractions showed that a considerable amount of unprocessed Acaa1 , Phyh , and Agps were localized to the cytosol-enriched fraction of Tysnd1−/− liver . In wild-type mice liver extracts Acaa1 and Phyh were almost exclusively detected in the peroxisome-enriched fraction , and Agps was not detectable in wild-type mice livers ( Figure 6G ) . The results of Agps processing by Tysnd1 in COS7 cells supplemented with MG132 proteasome inhibitor indicate that the processed form of Agps is prone to degradation ( Figure 5B ) . The unprocessed forms of PTS1-containing proteins Acox1 , Hsd17b4 , and ScpX were enriched in the liver peroxisome- and cytosol-enriched fractions of Tysnd1−/− mice compared with wild-type mice ( Figure 6G ) . The results are consistent with the increased amount of unprocessed Acaa1 , Acox1 , ScpX , and Hsd17b4 observed on Western blots of liver extracts ( Figure 1E ) and the expression pattern of peroxisomal membrane marker Pmp70 in Tysnd1−/− ( Figure 6G ) . Pex5 , the peroxisomal PTS1 receptor , was present in the cytosol- and peroxisome-enriched fractions of both Tysnd1+/+ and Tysnd1−/− liver , with the majority located in the cytosolic fraction ( Figure 6G ) . Pex7 , the PTS2 receptor was barely detected in the peroxisomal fractions of Tysnd1+/+ and Tysnd1−/− liver . We interpret the partial peroxisomal localization of PTS1- and PTS2-containing proteins observed in the fractionation experiment ( Figure 6G ) as the effect of an overall accumulation of peroxisomal proteins in Tysnd1−/− primary hepatocytes . Loss of Tysnd1 might mediate the overload of the peroxisomal targeting signal receptors . To test this hypothesis , we co-transfected Tysnd1+/+ and Tysnd1−/− hepatocytes with Acaa1-GFP ( 4 µg ) and HA-Acox1 ( 1 µg and 4 µg ) , and compared the co-localization of Acaa1-GFP ( PTS2 ) with anti-Pmp70 Alexa Fluor 568 to that in singly-transfected hepatocytes . In Tysnd1−/− hepatocytes Acaa1-GFP localized in part with peroxisomes and in part with a different cellular compartment ( Figure S7E ) . In contrast , even after co-transfection of Tysnd1+/+ hepatocytes with 1 µg HA-Acox1 ( Figure S7E ) or 4 µg HA-Acox1 ( data not shown ) , Acaa1-GFP still co-localized with Pmp70 Alexa Fluor 568 in peroxisomes , indirectly lending support to our hypothesis .
Tysnd1 processes both PTS1-targeted ( Acox1 , Hsd17b4 , and ScpX ) and PTS2-targeted ( Agps , Phyh , and Acaa1 ) enzymes that are involved in peroxisomal β-oxidation of VLCFAs , phytanic acid metabolism , and plasmalogen synthesis . In Tysnd1-deficient mice limited peroxisomal targeting and accumulation of unprocessed substrates reduced the metabolic activities of the aforementioned three pathways . As shown in Figure 1E and Figure 6G , the amount of the examined peroxisomal matrix proteins and total Pmp70 was elevated in Tysnd1−/− mice , indicating spontaneous peroxisome proliferation ( Figure 6A ) . We interpret the unexpectedly strong Pmp70 signal in the cytosol-enriched fraction ( Figure 6G ) of Tysnd1−/− liver extract as possible interference with the chaperoning function of Pex19 , which binds co-translationally to newly synthesized Pmp70 and is docked by Pex3 to the peroxisomal membrane [27] , [28] . Since Pmp70 was reported to aggregate and degrade in absence of Pex19 [27] , we speculate that Tysnd1 deficiency might somehow indirectly interfere with the insertion of Pmp70 into the peroxisomal membrane via Pex19-Pex3 docking , leaving soluble Pmp70 in the cytoplasm . However , without further investigation we cannot exclude other mechanisms that account for the presence of Pmp70 in the cytosol-enriched fraction . EM image analysis of liver peroxisomes in Tysnd1−/− mice showed a significant increase in their number ( Figure S2B ) together with a slight enlargement ( Figure S2C ) . The proliferation of peroxisomes in Tysnd1−/− mice seems to be the result of compensatory changes caused by their impaired peroxisomal functions . The resulting phenotype of Tysnd1−/− mice resembles biochemically mild variants of RCDP type 1 disease ( RCDP1 ) with somewhat decreased plasmalogen and increased phytanic acid levels . RCDP1 is caused by nonsense mutations in PEX7 which specifically prevent the import of PTS2-containing proteins PHYH , ACAA1 , and AGPS [29]–[31] into peroxisomes . Pex7−/− mice , a model for RCDP1 [8] show reduced plasmalogen synthesis . In Pex7−/− mice the α-oxidation of phytanic acid is impaired , resulting in low phytanic acid levels under normal diet conditions , but a significant accumulation after oral phytol administration . At birth only 50% of pups were alive , and the surviving male mice developed testicular atrophy with infertility , dwarfism by delayed endochondral ossification and eye cataracts in adults . In contrast , Tysnd1−/− mice displayed a rather mild phenotype with most pups reaching adulthood without dwarfism and normal eyes until we stopped maintaining and monitoring the mice at one year of age ( data not shown ) . The underlying phenotypic differences are attributed to the abrogation of PTS2-containing peroxisomal protein import in Pex7−/− mice versus diminished import with inadequate peroxisomal localization in Tysnd1−/− mice . Similar findings were reported for mice deficient in Gnpat , which is also involved in plasmalogen synthesis . Male Gnpat−/− mice are aspermic and infertile [32] , whereas infertile Tysnd1−/− mice produce malformed sperms . Biochemically , classical RCDP1 patients show strongly reduced plasmalogen , elevated phytanic acid and low pristanic acid levels [33] . Clinical symptoms include severe growth and mental retardation , congenital cataracts , chondrodysplasia and rhizomelia . A biochemical and neurological study of eleven patients diagnosed with RCDP1 included three female patients who were clinically diagnosed with a mild form of RCDP1 [34] . One of the patients who displayed autistic behaviour patterns and developed epilepsy at age 21 had only weakly elevated phytanic acid and reduced plasmalogen levels [34] . It is therefore possible that TYSND1 deficiency in human might cause phenotypes that are clinically diagnosed as a mild RCDP1 variant accompanied by male infertility . Repeated unsuccessful mating of male and female Tysnd1−/− mice helped us to discover the abnormal sperms , which prompted us to conduct a plasmalogen analysis . Plasmalogens are phospholipids that are enriched in myelin [21] , [35] , testes and spermatozoa membranes [20] , [36] where they are involved in anti-apoptotic functions [37] and spermatogenesis [38] . Developing gametes mature in the epididymis where the remodelling of ether lipids that constitute the sperm cell membrane occurs [39] . In Tysnd1−/− hepatocytes Agps , a rate-limiting enzyme [40] in the peroxisomal steps of plasmalogen synthesis showed almost no peroxisomal co-localization with DsRed2-Peroxi when transfected as GFP construct ( Figure 6C ) . The changes in the composition of choline- and ethanolamine-type components of epididymal plasmalogens ( Figure 3 ) most likely result in fragile sperm cell membranes and missing or defect acrosomes . The acrosome contains digestive enzymes that break down the outer ovum membrane , zona pellucida . Therefore , the acrosome-deficient sperms of Tysnd1−/− mice are unable to penetrate through the cumulus to fertilise the egg . Plasmalogens are also involved in neurodegenerative diseases [41] . In Alzheimer model mice levels of Agps protein and plasmalogen synthesis are reduced [42] . Considering that phytanic acid accumulates in the plasma , liver and brain of phytol-fed Pex7-deficient mice [8] and RCDP patients [29] , [43] the significantly lower home cage activity ( data not shown ) of male Tysnd1−/− mice at 34 weeks of age might be an indicator of behavioural anomalies due to neuronal changes mediated by reduced plasmalogen and elevated phytanic acid levels . Neurological and biochemical analyses of the central and peripheral nervous system are on-going and will be reported elsewhere . Phytanic acid , a natural agonist of PPARα induces peroxisome proliferation and hypertrophy [44] . Since blood serum levels of phytanic acid in male Tysnd1−/− mice fed with normal diet ( CLEA Japan ) were elevated ( Figure S4A ) , the metabolic tolerance to phytol , a precursor of phytanic acid was tested in a phytol feeding experiment . Serum phytanic acid levels of male Tysdn1−/− mice increased 100-fold compared with phytol-administered male wild-type mice . The phytol-intolerant phenotype and hepatic lipidosis seen inTysnd1−/− mice is similar to Phyh−/− mice . However male Phyh−/− mice are born without any abnormalities , and they are fertile in contrast to Tysnd1−/− mice . Ferdinandusse et al . [45] showed that phytol-fed Phyh−/− mice developed ataxia due to accumulation of phytanic acid in the cerebellum . Although Tysnd1−/− mice received twice the amount of phytol as the Phyh−/− mice [45] we did not observe ataxia , indicating only a mild impairment of phytanic acid metabolism in Tysnd1−/− mice . The study of an E . coli-expressed PHYH with a mutation in the N-terminal PTS2 region demonstrated that the unprocessed form of PHYH is active , but may affect in vivo its solubility and/or transport into peroxisomes [46] . We showed that unprocessed Phyh had a much stronger signal in the cytosol-enriched fraction of Tysnd1−/− mice ( Figure 6G ) than in the peroxisomal fraction . If unprocessed Phyh is active in vivo , it is probably the diminished amount of unprocessed Phyh inside peroxisomes that reduces phytanic acid metabolism in Tysnd1−/− mice . The peroxisomes of phytol-administered Tysnd1−/− mice showed a concomitant decrease in their number and an increase in their size . Probably , the extreme Pparα-mediated hypertrophy induced autophagy of peroxisomes ( Figure 4F , 4G ) , also termed pexophagy [47] , [48] . The accumulation of unprocessed peroxisome-targeted proteins Acox1 , Hsd17b4 , ScpX and Acaa1 ( Figure 1E ) in male phytol-administered Tysnd1−/− mice indicates the induction of a compensatory mechanism that counteracts the Tysnd1 deficiency-mediated reduced metabolic functions of peroxisomes and inflammatory liver changes . The inflammatory liver changes seen in Tysnd1−/− mice administered with phytol resembled the liver phenotypes of Scp2-deficient mice [23] . Two out of nine male mice suddenly died at days 12 and 13 after phytol administration . Possibly , atrioventricular changes induced by high phytanic acid levels led to cardiac arrest as reported for Scp2-knockout mice [49] . All female Tysnd1−/− mice administered with phytol-containing diet died during the experiment . BALB/c and C57BL/6J females are known to have low amounts of liver ScpX , which catalyses the thiolytic cleavage of branched-chain 3-ketopristanoyl-CoA during the degradation of pristanic acid [50] . Tysnd1 loss of function in female mice seems to exacerbate the effect of reduced amounts of ScpX and the toxicity of accumulating phytanic acid . Although plasma pristanic acid levels of phytol-fed Tysnd1−/− mice were statistically not significantly ( p = 0 . 105 ) elevated ( Figure S4G ) , the significant increase in phytanic acid ( Figure 4C ) suggests that the loss of Tysnd1 affects Phyh , the primary peroxisomal α-oxidation enzyme and subsequent enzymes in the hierarchy of β-oxidation reactions as evidenced by the reduced peroxisomal β-oxidation rate in liver ( Figure 4E ) . An in vitro study showed that there are no differences in the enzymatic activity of processed and unprocessed forms of human AGPS [51] . Since peroxisomes lack DNA and protein synthesis capabilities , all peroxisomal proteins are synthesized in the cytosolic compartment and post-translationally sorted to the peroxisome . Insufficient cleavage of substrates interfered with their peroxisomal localization as observed in primary hepatocytes of Tysnd1−/− mice ( Figure 6A–6C; Figure S6A–S6C , Figure S7E ) . Reduced peroxisomal targeting neither occurred in wild-type hepatocytes ( Figure 6A–6F; Figure S7B–S7D ) nor when peroxisomal protein Acox1 was co-expressed ( Figure S7E ) or overexpressed ( data not shown ) , indicating that the peroxisomal import of PTS2-containing proteins was to some extent impaired in Tysnd1−/− hepatocytes while accumulating in the cells . The overall accumulation of peroxisomal proteins may lead to the saturation of the Pex5- and Pex7-mediated peroxisomal protein transport capacity . RNAi knockdown of Tysnd1 in Hela cells reportedly [52] ( data not shown ) resulted in normal peroxisomal localization of Agps without apparent effect on the import system of peroxisomal proteins . We interpret the discrepancy between earlier reported results and ours , obtained from primary hepatocytes of Tysnd1−/− mice , as an effect of residual Tysnd1 protein expression after insufficient knock-down of Tysnd1 by RNAi . In mammals the import of PTS2-containing proteins into the peroxisome depends on the binding to Pex7 and the direct interaction of Pex7-bound PTS2 protein with the long isoform of Pex5 ( Pex5pL ) [53] , [54] . The N-terminal PTS2 signal is cleaved when the Pex5pL-Pex7-PTS2 protein complex has been transported into the peroxisome [9] . In Pex7−/− mice , Agps is absent in the liver and brain , suggesting that the precursor form of PTS2-containing proteins are unstable and prone to degradation [8] when they are not bound to Pex7 . In contrast , the precursor form of Agps in Tysnd1−/− mice hepatocytes was not degraded ( Figure 6G ) . These results strongly suggest that the binding of Pex7 to PTS2-containing proteins and subsequent association with Pex5pL in Tysnd1−/− mice is somehow impaired , which seems to affect Pex7 recycling and the degradation of PTS2-containing proteins [55] . Contrary to other models of peroxisome biogenesis disorders in which mislocalized peroxisomal proteins undergo accelerated degradation , the abundance of these proteins in our mouse model suggests that hitherto unidentified factors play a role in stabilizing these proteins and these will be the subject of future studies that may elucidate novel aspects about peroxisomal biogenesis . The precise mechanism how the cleaved PTS2 signal is detached from Pex7 and Pex5pL is still unknown , but it might resemble the recently identified mechanism of Ubp15p-mediated Pex5 detachment in yeast from the PTS1 signal . Ubp15p , a ubiquitin hydrolase , cleaves off the ubiquitin moieties from the PTS1 receptor Pex5p [56] . The released Pex5p becomes available for a new round of matrix protein import from the cytosol [56] . Based on the current knowledge of Pex7- and Pex5pL- dependent PTS2-containing protein import into mammalian peroxisomes and the interference with the peroxisomal localization of Acaa1 , Agps and Phyh caused by defective Tysnd1 processing , we propose a model which would leave , in the absence of Tysnd1-proteolytic removal of the PTS2 sequence , most PTS2-containing proteins bound to Pex7 in association with Pex5pL ( Figure 7 ) . The Pex7-Pex5pL-PTS2 protein complex returns to the cytosol , thereby limiting the rate of PTS2-containing protein import . Decreasing levels of free Pex5pL are predicted to affect also Pex5pL/Pex5pS heterodimerization and indirectly the import of PTS1-containing proteins by Pex5pL/Pex5pS heterodimeric oligomers [54] , which may narrow PTS1 import to Pex5pS homodimers . As a result PTS1- and PTS2-containing peroxisomal matrix proteins would accumulate in the cytoplasm as seen in Tysnd1−/− mice . In conclusion , Tysnd1−/− mice show reduced β-oxidation and phytanic acid metabolism . The changes in the plasmalogen composition , which we consider a contributing factor , but not necessarily the cause of male infertility in Tynsd1−/− mice are thought to be secondary effects of altered cellular acyl-CoA pools mediated by the reduced β-oxidation . Faulty peroxisomal targeting of novel Tysnd1 PTS2-containing substrates Phyh and Agps , and the previously reported substrate Acaa1 decreases their activities . Since Tysnd1 acts as a protease that affects the function of its substrates in the mouse model , we anticipate a new human peroxisomal disease entity caused by impaired TSYND1 functions that trigger a combination of mild dysfunctions among TYSND1 processing-dependent peroxisomal enzymes .
Heterozygous Tysnd1 knock-out mice were generated by TaconicArtemis ( Cologne , Germany ) under an ArteMice CONSTITUTIVE service contract . Briefly , genomic fragments of C57BL/6 BAC DNA ( RP23-302P6 ) were subcloned into pTysnd1 FINAL Seq ( MK141 ) vector . The deletion of exons 2 and 3 was confirmed by Southern blotting . C57BL/6N embryonic stem ( ES ) cells were electroporated with the construct and cultured . After confirming homologous recombination by Southern blotting , the ES cells were microinjected into C57BL/6J blastocysts . Germline transmission was achieved by crossing chimeric males with C57BL/6J females . Heterozygous mice breeding produced viable homozygous Tysnd1 null mice . Since the mice were of a hybrid C57BL/B6J , B6N background , heterozygous F1 mice were back-crossed for two generations with C57BL/6J mice . The mice were housed under specific pathogen free conditions at 23°C with a 12 h light-dark cycle . All mice used in this study were maintained and handled according to the protocols approved by the Animal Research Committee of Saitama Medical University . Genotyping of male and female mice was performed by multiplex PCR under the following conditions: 5 min at 94°C; 35 cycles of 30 s at 94°C , 30 s at 55°C , 1 min at 72°C; and a final extension for 10 min at 72°C . The P1 forward primer ( 5′-cctggctcctcactgtttgc-3′ ) was combined with two reverse primers P2 ( 5′-ctacacttaaccagatgtgctttcc-3′ ) and P3 ( 5′-gtcatagtagtggccagaacc-3′ ) . Primers P1 and P2 amplify the wild-type allele ( 237 bp amplicon ) . Primer pair P1 and P3 amplify the null allele with an amplicon size of 339 bp . Livers of Tysnd1−/− , wild-type and heterozygote mice were immersed in RNA-later solution ( Takara ) . Total RNA was isolated using the RNeasy Miniprep kit ( Qiagen ) . Real-time RT-PCR was carried out using the same primers ( Table S4 ) and method as described [14] Mice were fed ad libitum with standard rodent diet CE-2 ( Clea Japan ) . During experiments with high-fat diet mice were given D12492 Rodent Diet containing 60 kcal% fat ( Research Diets ) . Control mice were fed with D06041501 Rodent Diet containing 10 kcal% fat ( Research Diets ) . The diet was changed one to two weeks prior to oral phytol administration from CE-2 to D06041501 control diet . Orally administered phytol ( 15 mg/day/mouse ) was suspended in 0 . 25% sodium carboxy methyl cellulose . Fundamental and in depth screens of JMC pipeline 1 were performed with 7–26 weeks old mice as previously reported [17] . Modified-SHIRPA which is a component of JMC pipeline 1 was routinely conducted for eight weeks old mice . It involved 42 tests identical to the first stage of the original SHIRPA protocol [18] . Home cage locomotor activity , which is not included in JMC pipeline 1 was tested as part of an energy metabolism screen as previously described by Cao et al . [19] . Custom-made rabbit polyclonal antibodies against Tysnd1 [14] , Acaa1 also called prethiolase [14] , Acox1 , ScpX , Scp2 , Hsd17b4 and Agps were obtained from Scrum . Anti-Phyh ( 2858-1-AP; Proteintech Group , Inc . ) , anti-Pex7 ( 20614-1-AP; Proteintech Group ) and anti-Pex5 ( GTX109798; GeneTex ) polyclonal antibody were purchased from Funakoshi . Monoclonal anti-Pmp70 ( SAB4200181; Sigma Aldrich ) and rabbit anti-Gapdh antibodies ( G9545; Sigma-Aldrich ) were purchased from Sigma-Aldrich . The antigenic sequences and their positions are shown in Table S3 . MN9 is a monoclonal antibody raised against mouse spermatozoa that has been previously characterized [57] . Protein-coding regions of potential Tysnd1 substrate candidate genes were amplified by PCR using cDNA products from a male liver of C57BL/6J mice , RIKEN Mouse Genome Encyclopedia DNAbook ( DNAform ) . Primer information is shown in Table S4 . For co-expression studies the PCR products of candidates were cloned into pcDNA3 . 1 or pcDNA3 . 1/V5-His-TOPO vectors ( Invitrogen ) . For subcellular localization experiments the PCR products of candidates targeted by C-terminal PTS1 or N-terminal PTS2 were cloned into pcDNA3 . 1/NT-GFP-TOPO and pcDNA3 . 1/CT-GFP-TOPO vectors ( Invitrogen ) , respectively . N-Tysnd1-Flag-C vector was used as previously described [14] COS7 cells were grown under 5% CO2 at 37°C in Dulbecco's modified Eagle's Medium ( DMEM ) ( GIBCO ) containing 10% ( v/v ) heat-inactivated fetal bovine serum ( Bio West ) and 0 . 1 mM non-essential amino acid ( GIBCO ) . Primary cell cultures were prepared from livers of six weeks old male mice . Excised livers were drained of blood by flushing the hepatoportal vein with pre-circulation medium ( HBSS ( − ) without phenol red , with HEPES , pH 7 . 2 ) and refluxed with circulation medium ( 1 mg/ml collagenase type IV in HBSS ( − ) ) . Once the liver texture became soft , tissue cells were suspended and incubated with circulation medium for 20 min at 37°C . The cell suspension was filtered through a 70 µm nylon mesh washed with DMEM and cultured with DMEM containing 10% FBS , 100 U/ml penicillin and 0 . 1 mg/ml streptomycin . One day before transfection experiments , COS7 cells or primary cell cultures were seeded in plates and grown until 90–95% confluence . Transfections of COS7 cells and primary culture cells were performed according to the manufacturers' protocols with Lipofectamine 2000 ( Invitrogen ) and LTX ( Invitrogen ) , respectively . Liver or testis samples were homogenized in radioimmunoprecipitation assay ( RIPA ) buffer containing protease inhibitor cocktail ( Roche ) , followed by ultrasonic fragmentation and centrifugation . Peroxisomal fractions of liver were obtained as described by Omi et al . [58] . The cytosolic and peroxisomal fractions were confirmed by Western blotting using anti-Gapdh and anti-Pmp70 antibodies , respectively ( Table S3 ) . The protein extracts were separated by SDS-polyacrylamide gel electrophoresis and transferred to Immobilon-P membranes ( Millipore ) at least in duplicates . The Western blots were probed with anti-rabbit antibodies ( see Antibodies ) conjugated to anti-Rabbit IgG HRP ( W401B31549303 , Promega ) or anti-mouse IgG-HRP ECL ( NA931 , GE Healthcare Life Sciences ) . The recognized proteins were detected by using ECL plus or ECL advance Western blotting detection kits ( GE Healthcare Life Sciences ) . Livers of 15 weeks old male mice were excised and homogenized in 0 . 25 M sucrose and 5 mM MOPS using a Potter-Elvehjem homogenizer . After centrifugation of the homogenate at 3 , 000×g for 1 min the supernatant was incubated with 37°C for 60 min with reaction buffer as described elsewhere [59] , [60] . [1–14C]lignoceric and [1–14C]palmitic acids ( Muromachi Yakuhin ) were used to measure peroxisomal and mitochondrial β-oxidation activities , respectively . The reaction was stopped by adding 1/5 volume of 1N KOH . Fatty acids were heated at 60°C and acidified in 6% perchloric acid for 1 hour on ice . After centrifugation at 7 , 700×g for 10 min , the [1–14C]lignoceric or [1–14C]palmitic acids were removed with 1 . 5 ml chloroform-methanol ( 2∶1 ) and 14C-labeled water-soluble metabolites were analyzed in a liquid scintillation counter ( Beckman , LS 6500 ) . Spermatic fluid was extracted from the caudal region of the epididymis . The samples were incubated for 15 min at 37°C with 1∶10 diluted mitochondrial stain MitoFluor Red 589 ( Molecular Probes ) . After washing with 1×PBS the samples were mounted on glass slides using DAPI nuclear stain-containing mounting solution ( Vector Laboratories ) . Digital images were taken with Axiovert 200 M ( Zeiss ) . Total lipids were extracted with chloroform/methanol ( 1∶2 by vol . ) from freeze-dried mouse epididymides and testes . Plasmalogens were analyzed by liquid chromatography/electrospray ionization tandem mass spectrometry ( LC/ESI-MS/MS ) . Liquid chromatography separation was performed using an Accela UPLC system ( Thermo Fisher ) with a BEH C8 column ( 1 . 7 µm , 100 mm×2 . 1 mm i . d . ; Waters ) at 60°C and a flow rate of 450 µl/min . Mobile phase A consisted of water containing 5 mM ammonium formate . Mobile phase B consisted of acetonitrile . MS analysis was performed using a TSQ Quantum Access Max ( Thermo Fisher ) equipped with an HESI probe in positive ion mode . The number of molecular species measured by LC-MS/MS was thirty-nine for both PlsCho and PlsEtn when limited to those containing at least one percent of total plasmalogens . The collision energy was 32 eV for PlsCho and 20 eV for PlsEtn . Tissue samples were immersed overnight in 4% paraformaldehyde in phosphate buffered saline ( PBS , pH 7 . 4 ) , washed with PBS and dehydrated in ethanol solutions containing 70% to 100% ethanol . The dehydrated samples were cleared in xylene/ethanol and xylene , and embedded in paraffin ( Sakura Fine Tech Japan ) . All samples were cut into 8–10 µm sections using a large sliding microtome ( Yamato Kohki ) and mounted on glass slides . The sections were deparaffinized by passing them through xylene , graded ethanol series ( 100–70% ) and PBS . After staining with hematoxylin and eosin the sections were dehydrated in ethanol and embedded in glycerol . EM analysis of sperm and testis samples derived from two Tysnd1−/− and two Tysnd1+/+ ten weeks old male mice were performed as described before [61] . The liver samples were fixated by immersion in 2 . 5% glutaraldehyde plus 4% paraformaldehyde in 0 . 1 M phosphate buffer ( pH 7 . 2 ) at 4°C for 2 hr , and postfixated for 1 h in a solution of 1% osmium tetroxide in phosphate buffer . The samples were then dehydrated in graded ethanol and embedded in epoxy resin . Ultrathin sections were cut with an ultramicrotome ( Leica ) at a thickness of 60 nm , mounted on 200-mesh nickel grids and counterstained with 2% uranyl acetate and 1% lead citrate . The sections were examined under a JEM 1010 transmission electron microscope ( JEOL ) with accelerating voltage of 80 kV . Blood serum VLCFAs and phytanic acid levels Tysnd1−/− and wild-type mice were measured by using GS/MS as previously reported [62] . VLCFAs were determined from the serum of ten weeks old mice . Phytanic acid was measured in the serum of both ten and 38–39 weeks old mice . Plasma pristanic acid levels of ten weeks old male Tysnd1−/− and wild-type mice were determined by UPLC-MS/MS as described in Text S1 . Primary hepatocytes of six weeks old male Tysnd1+/+ or Tysnd1−/− were cultured on cover glasses in 6-well plates and each of the GFP-fused Tysnd1 substrate expression vector ( Table S4 ) was either singly transfected or co-transfected with the C-terminal PTS1-containing peroxisomal marker DsRed2-Peroxi ( Clontech ) . All transfections except for Acaa1-GFP and GFP-ScpX were performed in duplicate . Twenty-four to 36 hours after transfection the living cells were analyzed . The cells used for immunostaining were first fixed in 4% paraformaldehyde containing PBS , then blocked in 2% skim milk and incubated with the primary antigen and anti-Pmp70 ( peroxisomal membrane marker ) . Immune complexes were visualized using Alexa Fluor 568 or Alexa Fluor 488 labeled goat anti-rabbit IgG antibodies ( Molecular Probes ) . Cells were observed with a TCS SP2 confocal laser-scanning microscope ( Leica ) . The fluorescence of GFP or Alexa Fluor 488 was measured at 488 nm excitation . DsRed2-Peroxi and Alexa Fluor 568 fluorescence was measured at 543 nm excitation .
|
Peroxisomes are subcellular organelles that are present in almost all eukaryotic cells . The syllables “per-oxi” reflect the oxidative functions of these single-membrane-bound organelles in various metabolic processes , including those of very-long-chain fatty acids and branched-chain fatty acids . In an earlier study we identified a protease named Tysnd1 that is specifically located in the peroxisomes and processes the enzymes catalyzing the peroxisomal β-oxidation of very-long-chain fatty acids . In this study , we identified two novel Tysnd1 substrates , Agps and Phyh , which are involved in plasmalogen synthesis and phytanic acid metabolism , respectively . To further investigate the in vivo function of Tysnd1 , we analyzed Tysnd1 knock-out mice . Mice that lack Tysnd1 showed reduced peroxisomal β-oxidation activity and an altered plasmalogen composition , as well as an abnormal phytanic acid metabolism . Male infertility is one of the major phenotypic manifestations of Tysnd1 deficiency . Our data support the idea that Tysnd1 affects the localization and activity of some of its substrates inside peroxisomes . Altogether , our Tysnd1-deficient mouse model expands the current peroxisome biology knowledge with regard to the molecular pathogenic mechanisms that may be relevant to some patients with Zellweger syndrome spectrum disorders .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"medicine",
"urology",
"medicinal",
"chemistry",
"anatomy",
"and",
"physiology",
"chemical",
"biology",
"histology",
"model",
"organisms",
"chemistry",
"biology",
"chromatography",
"metabolic",
"disorders",
"biochemistry",
"physiology",
"genetics",
"molecular",
"cell",
"biology",
"genetics",
"and",
"genomics"
] |
2013
|
Tysnd1 Deficiency in Mice Interferes with the Peroxisomal Localization of PTS2 Enzymes, Causing Lipid Metabolic Abnormalities and Male Infertility
|
Epstein-Barr virus ( EBV ) , a lymphomagenic human herpesvirus , colonises the host through polyclonal B cell-growth-transforming infections yet establishes persistence only in IgD+ CD27+ non-switched memory ( NSM ) and IgD− CD27+ switched memory ( SM ) B cells , not in IgD+ CD27− naïve ( N ) cells . How this selectivity is achieved remains poorly understood . Here we show that purified N , NSM and SM cell preparations are equally transformable in vitro to lymphoblastoid cells lines ( LCLs ) that , despite upregulating the activation-induced cytidine deaminase ( AID ) enzyme necessary for Ig isotype switching and Ig gene hypermutation , still retain the surface Ig phenotype of their parental cells . However , both N- and NSM-derived lines remain inducible to Ig isotype switching by surrogate T cell signals . More importantly , IgH gene analysis of N cell infections revealed two features quite distinct from parallel mitogen-activated cultures . Firstly , following 4 weeks of EBV-driven polyclonal proliferation , individual clonotypes then become increasingly dominant; secondly , in around 35% cases these clonotypes carry Ig gene mutations which both resemble AID products and , when analysed in prospectively-harvested cultures , appear to have arisen by sequence diversification in vitro . Thus EBV infection per se can drive at least some naïve B cells to acquire Ig memory genotypes; furthermore , such cells are often favoured during an LCL's evolution to monoclonality . Extrapolating to viral infections in vivo , these findings could help to explain how EBV-infected cells become restricted to memory B cell subsets and why EBV-driven lymphoproliferative lesions , in primary infection and/or immunocompromised settings , so frequently involve clones with memory genotypes .
Epstein-Barr virus ( EBV ) , an orally transmitted herpesvirus widespread in human populations , first replicates in a permissive cell type in the oropharynx and then colonises the B cell system through a growth-transforming infection that drives the clonal expansion of latently-infected cells [1]–[3] . This growth transformation can be studied in vitro where infection of resting B cells occurs via CD21 receptor-mediated virus entry and leads to the outgrowth of permanent lymphoblastoid cell lines ( LCLs ) expressing all eight EBV latent cycle proteins ( six nuclear antigens EBNAs 1 , 2 , 3A , 3B , 3C and –LP , and two latent membrane proteins LMPs 1 and 2 ) [3] . Cells displaying these same markers of viral transformation are present in the tonsillar lymphoid tissues of infectious mononucleosis ( IM ) patients undergoing primary EBV infection [4] , [5] . Already , however , there is heterogeneity within these expanding B cell clones in IM tonsils [5] , [6] , with some cells apparently down-regulating viral antigen expression and switching out of cell cycle , thereby establishing a latent reservoir that can evade detection by the host T cell response . A key finding was that the cells constituting this reservoir , whether in the blood of convalescent IM patients or of long-term EBV carriers , lie within the IgD− CD27+ memory B cell subset and not in IgD+ CD27− naïve cells [7]–[9] . Furthermore , in IM cases where infected cell numbers were sufficient to allow single cell analysis , these cells carried somatically-mutated immunoglobulin ( Ig ) gene sequences typical of antigen-experienced memory cells [9] , as do many of the EBV-driven lymphoproliferative disease lesions that arise in immunocompromised patients where T cell control is relaxed [10]–[14] . The physiologic process of memory selection involves IgM+ IgD+ CD27− naïve B cells encountering cognate antigen in lymphoid tissues and , with antigen-specific T cell help , proliferating to form germinal centres ( GCs ) . Here Ig variable gene sequences are subject to successive rounds of somatic hypermutation ( SHM ) to generate intra-clonal diversity before being re-expressed , usually in isotype-switched forms [15] . Both SHM and isotype-switching are critically dependent upon activation-induced cytidine deaminase , AID [16] , [17] , but are nevertheless distinct reactions that can take place independently of one another [18] , [19] . The small fraction of GC progeny cells with improved affinity for antigen are then specifically selected by T cell-derived survival signals , emerging as IgD− CD27+ memory B cells; the great majority of these are also IgM− and have switched isotype to IgG or IgA , ( “switched memory” cells ) [20] . Given this understanding of the physiology of memory cell selection , different views have emerged as to how EBV might selectively colonise the IgD− CD27+ memory cell pool . One view is that the virus first infects naïve cells in vivo and , through mimicking the activation signals normally induced by cognate antigen , drives these cells to initiate a GC reaction; the virus-infected clonal descendents of that reaction thus acquire both the genotype and phenotype of memory cells via the natural process of GC transit , albeit with virus-coded LMPs 1 and 2 substituting for affinity-based survival signals [1] . A second view , based mainly on the analysis of EBV-infected B cell clones within IM tonsillar tissues , is that memory B cells are preferentially infected , or possibly have a proliferative/survival advantage during the phase of virus-driven B cell expansion , and that their progeny subsequently re-assume memory characteristics with no requirement for GC transit [2] . An added complication to this debate is the more recent finding [21] that EBV is harboured not just in the conventional IgD− CD27+ memory pool of healthy virus carriers but also at lower levels in a second distinct memory population comprising IgD+ CD27+ “non-switched memory” cells [20] , [22] , that had hitherto been largely ignored in EBV studies . Although such cells do carry somatically mutated Ig genes , opinion is divided as to whether the IgD+ CD27+ subset arises ontogenetically and is entirely independent of GC activity ( as their presence in certain GC-null immune-deficiency states would imply [23] , [24] ) or is populated at least in part by the products of abortive/incomplete GC reactions involving Ig gene mutation without isotype switching [20] , [25]–[28] . EBV's ability to colonise this subset in healthy individuals can therefore be explained in different ways depending upon one's view of non-switched memory B cell origins . Interestingly however , in T cell-deficiency states where GC development is blocked and there are no conventional switched memory B cells , EBV is sequestered in the tiny population of Ig gene-mutated , non-switched memory cells that exists in such patients rather than in the numerically dominant naive population [29] , [30] . The chances of incoming virus selectively targeting such a small population by direct infection seems remote , again raising the possibility that , in naive B cells , EBV infection per se may be able to impose a memory genotype and/or phenotype on these cells without recourse to GC signals . Given these uncertainties , it is surprising that little attention has been given to studying naïve , non-switched-memory and switched-memory ( here designated N , NSM and SM respectively ) B cell subsets as targets of EBV infection in vitro . Specifically , can naïve cells acquire aspects of the memory cell Ig phenotype or genotype as a result of virus transformation ? Here we show that EBV infection per se does not alter the Ig phenotype of naive cells , although such infected cells remain susceptible to Ig class switch induction by surrogate T cell signals . However , we did find evidence for Ig gene mutation among naive B cell transformants; thus , both under limiting dilution and bulk culture conditions , selection for successful LCL outgrowth from naïve B cell infections frequently involved clones with mutated Ig genotypes that appear to have arisen through virus-induced SHM in vitro .
Circulating B cells , typically representing 4–8% of adult PBMC populations , were isolated by CD19 bead selection to purities consistently >98% , as shown by staining for the pan-B cell marker CD20 ( Figure 1A ) . Such preparations were co-stained with fluorochrome-labelled Abs to IgD and CD27 in order to identify the N ( IgD+ CD27− ) , NSM ( IgD+CD27+ ) and SM ( IgD− CD27+ ) B cell subsets . Naïve B cells always represented the major subset , accounting for 60–70% of total B cell numbers , with the other two subsets each constituting between 8–25% . All three subsets were isolated to high purity by FACS sorting . Figure 1B illustrates the sort gates used and Figure 1C shows the results of re-analysing IgD and CD27 staining on the sorted populations . By these criteria , purity was always >99% for N , >96% for NSM and >98% for SM cell preparations . IgH sequence analysis further confirmed the purity of these sorts . Thus , combining results from naive cell sorts from 12 different individuals , 188 of 195 IgH sequences amplified from the N cell population were deemed to be non-mutated ( i . e . less than 2 nucleotide changes from germline ) . In six of the above experiments , IgH sequencing was also extended to include the other subsets; as expected , the great majority of sequences amplified from the NSM and SM cell populations ( 33/37 and 64/66 respectively ) were clearly mutated ( Figure 1D ) . Expression of the CD21 receptor and levels of virus binding were determined for the above B cell subsets from 4 successive PBMC samples . Cell surface staining with a mAb to the EBV receptor CD21 reproducibly showed that N , NSM and SM cells express CD21 at similar levels ( Figure 2A ) . Likewise these cells all bound virus to similar amounts , as measured by exposure to a standard virus dose at 4°C to prevent internalisation of the virus , followed by extensive washing and quantitation of EBV genome copies bound per cell by quantitative PCR ( data not shown ) . We then assayed matched N , NSM and SM preparations for transformability using two different experimental designs . In one set of experiments , N , NSM and SM cells were exposed to a standard virus dose ( 50moi ) before being seeded at a range of cell densities in replicate wells in 96-well plates; in other experiments , N and NSM cells were exposed to a range of virus dilutions before seeding into a 96-well plate at a standard cell number . In both cases , end points in the transformation assay were scored after 6 weeks by microscopic inspection of cultures for characteristic foci of EBV-transformed cells . Typical results from such experiments are shown in Figures 2B and 2C , and are expressed as the percentage of replicate wells scoring positive at the limiting cell seeding or limiting virus dose . Whilst absolute values for transformation efficiency varied between experiments , for any one individual donor the three B cell subsets always gave very similar yields of transformed cultures . Thereafter , cultures scoring positive on plates set up under limiting conditions were first split into duplicate wells and , where possible , further expanded over the following 3–6 weeks to yield a limiting dilution ( LD ) -LCL for analysis . Note that only a proportion of the wells transformed under limiting conditions could be successfully expanded in this way; overall , however , that proportion was not greatly different comparing cultures derived from N ( 34% ) , NSM ( 30% ) and SM ( 24% ) cell infections . We therefore conclude that the N , NSM and SM subsets from peripheral blood are equally susceptible to EBV infection and transformation in vitro . In several transformation experiments , we also set up LCLs under non-limiting conditions by infecting 2×106 N , NSM and SM B cell preparations with EBV at 50moi and then culturing these cells in bulk , with subsequent expansion to bulk LCLs . To investigate whether EBV induced transformation altered Ig isotype expression in the resultant lines , we first performed RT-PCR analysis of IgH transcripts using isotype-specific primer combinations . Figure 3A shows typical results obtained , using reference B cell lines of known isotype restriction and the Ig-negative Jurkat T cell line as internal controls . Bulk lines derived from the N and NSM cell subsets were consistently positive for IgM transcripts and weakly positive for IgD but negative for the other isotypes . By contrast , bulk lines of SM cell origin expressed IgG and IgA transcripts , often accompanied by a weak signal for IgM but never for IgD; note that the weak IgM signal accords with the fact a very small proportion of cells within sorted IgD− CD27+ memory populations are so-called “IgM-only” cells with an IgM+ IgD− CD27+ phenotype [25] . The clear inference from these transcriptional data , that viral transformation had not induced detectable class switching , was strongly supported by the results of staining with mAbs specific for IgM , IgD , IgG and IgA heavy chains . Figure 3B shows typical results where both the N-derived and NSM-derived bulk LCLs retained an IgM+ IgD+ phenotype , whereas the SM-derived LCL was dominated by IgG+ and IgA+ cells . Note also that , as others have reported [31] , EBV transformation induces N cells to express CD27 such that all three groups of LCLs shared a CD27+ phenotype ( data not shown ) . Table 1 summarises the Ig isotype data both from bulk LCLs of N , NSM or SM origin generated as above , and from the LD-LCLs expanded from cultures of all 3 B cell subsets as described earlier . Again , all N- and NSM-derived cell lines from bulk and LD cultures were IgM+ IgD+; by contrast the SM-derived bulk lines were predominantly mixtures of IgG+ and IgA+ cells , sometimes with a small IgM+ component , whereas SM-derived LD cultures were either only IgG+ , only IgA+ or mixtures of both . We then asked whether EBV-transformed cells , though not induced to undergo isotype switching by viral infection , were still capable of switching given signals that mimic physiologic T cell help . CD40L and IL-4 stimulation is known to induce isotype switching to IgG in freshly isolated N cell preparations in vitro [32]–[34] , with switching also to IgA in the presence of IL21 [35] . Early passage bulk cultures ( 2–3 weeks post-infection ) of N- and NSM-B cell origin were exposed to these inducing signals for 10 days; parallel cultures being maintained under normal conditions as a control . Figure 3C shows the results from an N-derived culture , typical of that seen generally with early passage cultures of N or NSM origin . Untreated cultures were uniformly IgD+ and lacked IgG and IgA , whereas significant fractions of cells in cultures exposed to CD40L+IL4+IL21switched to IgG or IgA , with concomitant loss of IgD . We now examined the Ig genotype of all LD-LCLs established from the N , NSM and SM cell preparations whose pre-infection Ig genotypes had been analysed in Figure 1D . For each LD-LCL , we sequenced several cloned IgH PCR products , identified the constituent IgH V , D and J alleles and then assigned the sequences to individual CDR3 clones . By these criteria , the majority of LD-LCLs ( whether derived from N , NSM or SM cells ) were dominated by a single cellular clone; in addition , some were oligoclonal with 2 or 3 different clonotypes . Individual sequences within each identified clone were assigned as germline or mutated as described above . Table 2 shows results from one such experiment , here focusing only on infections of the N cell subset . The individual clonotypes detected within each LD culture are identified through their IgH V , D and J alleles and their signature CDR3 amino acid sequence . In this experiment , of 25 cultures analysed , there were 13 monoclonal cultures ( LCL6-1 to 6-13 ) and 2 biclonal cultures ( LCL6-14 , 6-15 ) with non-mutated genotypes . Surprisingly , however , there were also 7 monoclonal cultures with mutated genotypes ( LCL6-16 to 6-22 ) , plus 3 biclonal cultures ( LCL6-23 to 6-25 ) with both mutated and non-mutated genotypes . Figure S1 presents the actual sequences for three of the above cultures; LCL6-2 , LCL6-18 and LCL6-20 with 0 , 5 and 7 mutations , respectively . The above pattern of results , with a number of N-derived LD-LCLs showing significant levels of IgH mutation , was observed in 8 successive experiments , each involving a different naive B cell preparation . Combining data from all 8 experiments , we analysed 594 IgH sequences from 140 LD cultures of EBV-infected N cell preparations , and within these identified 198 distinct clones . The majority of these cultures ( 89/140 ) yielded only germline sequences , usually with a single CDR3 clonotype or in some cases with two co-resident clonotypes . However , the other 51/140 N-derived cultures yielded mutated sequences , in most cases in the absence of any detectable germline sequence . Overall , 72 of 198 ( 36 . 3% ) individual clonotypes identified within these cultures were represented by mutated sequences . As a comparator , in 6 of the above experiments we also analysed 38 NSM-derived and 55 SM-derived LD cultures from infections of the memory cell subsets . Within these we identified 54 and 96 resident clones , respectively , most cultures being dominated by a single or by two co-resident clonotypes . Not surprisingly , given the predominance of mutated IgH sequences present in these B cell subsets pre-infection , all of the NSM-derived and all but two of the SM-derived cultures carried mutated IgH sequences . Figure 4 summarises the overall IgH genotype data from the N , NSM and SM cell preparations pre-infection ( open bars ) and from the derived LD-LCL cultures ( shaded bars ) . The histograms show the total number of unique clonotypes observed in relation to their degree of divergence from germline , sequences with 0 , 1 or 2 changes deemed non-mutated . Note that the degree of IgH sequence divergence among NSM-derived and SM-derived LD-LCLs is not markedly different from that within the matched pre-infection populations . More importantly , however , mutated IgH sequences are much more common in LD-LCLs derived the N cell subset than in their pre-infection counterparts . We were interested to know whether the frequent appearance of clones with mutated Ig genotypes in LD-LCLs derived from N cell infections reflected a growth rate advantage that these cells enjoyed , perhaps one that might be linked to differences in the degree to which cells were leaving the EBV-transformed latent state and entering lytic cycle . Early passage freezings of the emergent N-derived LCLs from individual experiments were therefore taken from cryostorage and compared in proliferation assays . Figure 5A shows typical results from one such experiment comparing 5 mutated and 4 germline clones from the same donor . The emergent LD-LCLs varied considerably in their growth rates , but this was not related to their Ig genotype status; clones with mutated genotypes ( closed symbols ) and clones with germline genotypes ( open symbols ) showed a similar spread of growth rates . Furthermore , as is clear from the immunoblots in Figure 5B , neither Ig genotype status nor growth rate showed any obvious correlation with expression levels of EBV latent proteins ( here illustrated using the key transformation-associated proteins EBNA1 , EBNA2 and LMP1 ) or with the degree of lytic cycle entry as detected from levels of the immediate early lytic protein BZLF1 . Thus , although cells with mutated Ig genotypes are well represented among LD-LCLs from naive cell infections , this is not because they have an inherently faster growth rate than cells carrying germline IgH sequences . The above evidence for mutated Ig genotypes in N-derived LCLs came entirely from mono- or bi-clonal populations expanded from individual limiting dilution cultures , i . e . from wells scored positive at the end of the 6 week transformation assay . Because many such wells still contained just small foci of transformed cells at that time , limitations on cell numbers precluded any prospective analysis of these cultures during their early period of expansion; as a result , sampling of LD-LCLs for genotypic analysis was often delayed until 9–12 weeks post-infection . As a second approach therefore , we turned to the analysis of resident clonotypes within bulk N cell cultures , harvesting aliquots of the same culture at regular intervals up to 12 weeks post infection . This allowed us to compare the clonal composition within an expanding EBV-infected culture with that seen in a matched culture driven to expand by repeated exposure to a non-viral proliferative trigger , CD40L/IL4 . In our hands , this latter protocol induces expansions within the first 9 weeks which are at least as the equal of those driven by EBV , after which proliferation slows to a halt . Note that such mitogen-driven proliferation has never been reported to induce Ig gene mutation [36]–[38] . We first used the approach of IgH CDR3 spectratyping to gain an overall picture of CDR3 length distribution ( i . e . clonality ) in the two types of culture . Data from one such experiment are shown in Figure 6A . Clearly the CD40L/IL4-activated culture remains polyclonal despite 9 weeks of in vitro expansion , retaining a similar Gaussian distribution of CDR3 lengths to that seen in the original starting population . In contrast , the parallel EBV-transformed culture is dominated by sub-populations with distinct CDR3 lengths at both 6 and 9 weeks post-infection . The same samples were analysed by IgH sequencing as before , and the corresponding data are shown in Table 3 . Thus all 18 sequences amplified from the initial N cell population ex vivo were unique and 17/18 of these were germline; likewise all 17 sequences amplified from the bulk culture after 9 weeks expansion by CD40L/IL4 were unique and non-mutated . However , the parallel EBV-transformed culture analysed at 9 weeks contained only 5 distinct clonotypes , three of which appeared from their frequent detection to represent numerically dominant cell clones . Of these three dominant clones , one ( in this case , the most frequent ) had a germline IgH sequence while the other two had 3–4 mutations relative to the nearest germline sequence . In some experiments , greater cell yields in the N-subset sort allowed more frequent sampling of the EBV-infected bulk cultures . The CDR3 spectratyping data from one such experiment are shown in Figure 6B , clearly showing that the population remains broadly distributed in terms of CDR3 size up to 4 weeks post-infection but becomes much more focused by 6 weeks and further focused by 9 and 12 weeks . The corresponding IgH sequence data from this experiment are presented in Table 4 . They show that the EBV-infected N cell culture is composed of multiple unique clones both at 2 weeks , when all amplified sequences were germline , and also at 4 weeks , at which point the first mutated sequences appear as minor components . By 6 weeks , three clonotypes ( one of which was previously seen at week 4 ) together account for about half the amplified sequences and by 9–12 weeks the culture is dominated by just one of these clonotypes , with another remaining as a minor component . In this case , the dominant IgH clonotype in the N-derived bulk LCL is clearly mutated . Such EBV-infected bulk cultures were generated from six independent donors , in three cases with a matched CD40L/IL4-stimulated B blast culture . Figure 7 presents a summary of the IgH genotype data from the two types of cultures expressed as histograms; the height of the open bar indicates the percentage of sequences with a particular level of mutation frequency that were amplified from the bulk cultures at different times , while the shaded area of the bar reflects the proportion of those amplified sequences that were members of a clonal family i . e . were detected twice or more . Thus , in starting polyclonal populations , all IgH sequences are unrelated to one another and the great majority are non-mutated . This remains the pattern seen at all times in the CD40L/IL4-expanded B blasts , whereas in EBV-infected cultures , the situation changes with time post-infection . Early on these cultures are also polyclonal and non-mutated but , at around 4 weeks post-infection , mutated IgH sequences appear in significant numbers . From 4 to 12 weeks post-infection , the cultures then become increasingly dominated by a small number of individual clones; in some cases ( e . g . Table 3 ) the most abundant clonotype is non-mutated , while in other cases ( e . g . Table 4 ) it is mutated , much as seen earlier in the limiting dilution culture experiments . To ask whether the above Ig sequence changes seen in vitro might be products of SHM , we first screened N , NSM and SM subsets pre- and post-infection for expression of AID , an enzyme which is essential ( but itself not sufficient ) for SHM to occur [39] . In each case , AID transcription was undetectable before infection and was activated by EBV . Figure S2A shows the relevant data from N cell preparations . Interestingly , even though all cells in the culture were actively infected and proliferating by day 7 post-infection , AID levels rose only slowly , not reaching their steady state level until day 35 , kinetics that are at least compatible with the delayed appearance of mutated IgH sequences in such cultures . However , we found no correlation between AID expression and Ig gene mutation status . Thus AID transcript levels were similar in N-derived LCLs with and without mutation as well as in NSM- and SM-derived lines; interestingly these AID levels were not only lower than seen in freshly-isolated GC B cells and in 4 Burkitt lymphoma ( BL ) -derived cell lines , included as SHM-positive controls , but also much lower than those in CD40L/IL4-stimulated B blasts which lack detectable SHM ( Figure S2B , C ) . Moreover N-derived LCLs with and without IgH mutations gave similar results in quantitative RT-PCR assays specific for alternatively-spliced AID mRNAs and for the Polη and UNG co-factors involved in the SHM process [40] , [41] ( Figure S2C ) . As a second approach , we asked whether the IgH gene mutations seen in N-derived LCLs had the hallmark of SHM targeting by mapping the location of all IgHV sequence changes seen in pre-infection N , NSM and SM populations and in their derived LCLs . To avoid distortion of the LCL data by numerically dominant clones , here any one mutated clonotypic sequence identified within an LCL only contributes once to the cumulative data . The overall findings are summarised in Figure 8 , where the height of the bars indicates the number of times a change in a particular IgHV codon was identified . Sites in the IgHV sequence known to be favoured by the SHM machinery ( hot-spots ) [42] , [43] are identified by filled bars . Focusing first on the data from pre-infection B cell subsets , as already described we found very few mutations in the N-subset , while the NSM and SM subsets showed the expected distribution of mutated sites , with frequent involvement of known SHM hotspots and avoidance of coldspots . Not surprisingly , the NSM- and SM-derived LCLs showed a similar distribution of mutations as in their pre-infection populations . However , it was striking that the same pattern of distribution was also seen among the mutated IgH clonotypes found in N-derived LCLs , at least consistent with these changes being a product of SHM . Such findings nevertheless leave open the possibility that the mutated clonotypes frequently detected in N-derived LCLs have not arisen from authentic naive cells induced into SHM in vitro but from memory cells already carrying IgH mutations that were present as minor contaminants of the original N cell preparations . Note that the transformation assays on sorted B cell subsets ( Figure 2 ) implied that memory cell contaminants would enjoy no competitive advantage in such a situation . Furthermore , the IgM+ IgD+ phenotype of the mutated LCLs discounted their being derived from isotype-switched SM cells . However we could not formally discount a contribution from NSM contaminants from the Ig phenotype since both N-derived and NSM-derived transformants would give the same IgM+ IgD+ LCL signature . The possibility of resolving this issue by genotype was raised by a recent report [44] that a proportion of NSM cells carry mutations in the Bcl6 intronic major mutation cluster ( MMC ) , mutations that ( as in conventional memory cells ) are thought to arise through SHM mis-targeting during a cell's residence in GCs where Bcl6 is highly expressed [44] , [45] . If such mis-targeting is indeed dependent upon active bcl6 transcription , then this would not be expected to occur during EBV-induced B cell transformation in vitro because , as already reported [46] and as we confirmed in the present work ( Figure S3 ) , Bcl6 expression is suppressed by growth-transforming EBV infection . We therefore selected 18 LD-LCLs that had a mutated Ig genotype and were derived from N cell preparations , amplified a 718bp region of the Bcl6 MMC , then cloned and sequenced multiple independent amplification products . As internal controls , we included parallel amplifications from sorted naive and memory B cell populations ex vivo . As shown in Table S1 , 17/18 Bcl6 MMC sequences amplified from naive B cells were germline , whereas 10/24 sequences from memory cells were mutated; such findings are in close accord with an earlier report [44] . Turning to the Ig-mutated LD-LCLs from N cell infections , we detected both germline and mutated allelic sequences in just 2 lines; the great majority of lines ( 16/18 ) only ever yielded germline sequences . Overall , therefore , the bcl6 data are again consistent with the majority of these Ig-mutated LD-LCLs being truly derived from naive cells . To pursue the question another way , we reasoned that if the mutated clonotypes arising in EBV-infected N cell cultures were truly being generated from naive precursors in vitro , then we should be able to detect evidence of intra-clonal sequence diversification within such cultures . In this regard , 28 of the 72 LD-LCLs in which the dominant clonotypic sequence was mutated also contained variants ( typically with 1 to 3 nucleotide changes ) of that sequence at the single point of harvesting . More informative , however , are the data from bulk N-cell infections sampled over time . Summing data from all 6 experiments , each involving a different N-cell donor , we identified a total of 17 mutated clonotypes that were present on two or more occasions in the same culture between 4 and 12 weeks post-infection; of these , 14/17 clonotypes showed evidence of intra-clonal sequence variation , with a total range of 1–12 sequence changes per clonotype . Two such examples are presented in Figure 9 , where in each case the variants can be linked into family trees with different branches arising from a co-resident germline parental sequence . In one case ( Figure 9A ) , we found the parental IgH sequence and 9 clonally-related variants with up to 8 sequence changes; in another case ( Figure 9B and Table 4 ) , we found the parental IgH sequence and a further 5 related variants with up to 6 sequence changes . The detailed sequences used to construct these trees are shown in Figures S4A and S4B .
As originally stated , the different views as to how EBV selectively colonises the SM B cell subset in vivo hinged on whether this was GC-dependent , involving preferential infection of naïve cells that the virus then induced into memory via GC transit [1] , or GC-independent , involving preferential infection/expansion of pre-existing memory cells [2] . Neither view can fully accommodate the more recent finding that EBV also colonises the IgD+ CD27+ NSM subset , not just in healthy carriers [21] but also in patients congenitally devoid of GCs and therefore of SM cells; in such patients the NSM subset is hugely outnumbered by naïve B cells ( and is therefore very unlikely to be a preferential target of primary infection ) yet it harbours essentially all the latent virus [29] . Given the difficulties posed by such in vivo findings , the present work took the reductionist approach of focusing on experimental infection of N , NSM and SM cells in vitro . This showed that , while EBV itself does not induce Ig isotype switching , N-derived LCLs remain susceptible to switching induced by surrogate T cell signals . More importantly , in at least a proportion of N cells , EBV infection induces IgH sequence changes which bear the hallmarks of SHM; furthermore , B cell clones with such changes frequently become numerically dominant as the emerging LCL evolves towards monoclonality , typically beginning between 4–6 weeks post-infection . These findings not only suggest alternative routes whereby EBV might become embedded in B cell memory in vivo but also strengthen the argument that EBV-induced SHM could contribute to clonal evolution in EBV-associated lymphoproliferative lesions/lymphoma . It was first necessary to check the susceptibility of the different B cell subsets to EBV infection/transformation . This has been a surprisingly neglected issue after an early study , examining B cell subsets within EBV-infected tonsillar B cell cultures up to 48 hr post-infection , found no differences in the infectability of cells with different surface Ig isotypes [47] . More recently , Dorner et al . also reported that naïve ( CD27− ) and total memory ( CD27+ ) B cells from tonsils were equally infectable in short-term assays , although subsequently naïve cells grew slightly quicker and gave slightly better LCL yields from limiting dilution seedings [48] . While naïve ( CD27− ) cells from peripheral blood resembled their tonsillar counterparts , they were more infectable than peripheral blood memory ( CD27+ ) preparations; this could not be explained at the level of receptor ( CD21 ) /co-receptor ( HLAII ) expression but was ascribed to heterogeneity among circulating memory cells in expression of another putative co-receptor , α5β1 integrin [49] . Comparisons with the present work are difficult because Dorner et al . employed an unusual recombinant EBV strain ( lacking one of the latent proteins , LMP2A ) whose low titre preparations necessitated the use of spinoculation to achieve measurable rates of infection . The present work , using wild-type EBV and conventional infection protocols , did not detect any significant difference in virus binding , infectability or transformability under limiting conditions between peripheral blood N , NSM and SM preparations ( Figure 2 ) . Likewise , another recent report comparing naïve ( CD27− ) and total memory ( CD27+ ) B cells from blood also found no significant difference in transformability [50] . Interestingly we noticed that , in contrast to the ease with which LCLs grow out from positive wells at the high end of transformation assays , only a subset of positive wells arising under limiting conditions could be expanded to establish LCLs . This likely reflects the fact , recently noted by others [51] , [52] and also apparent from our CDR3 spectratyping of EBV-infected bulk cultures ( Figure 6 ) , that the process of LCL establishment is associated with significant clonal selection , a hurdle which cultures with small seed populations may fail to overcome . The basis of this selection remains to be determined . However , from the point of view of the present work , we can conclude that N , NSM and SM populations show similar levels of attrition at this stage . With respect to virus-induced changes in cellular phenotype , as many have observed [52]–[56] , EBV-transformed LCLs converge on a similar “lymphoblastoid” phenotype , irrespective of the precise differentiation stage of the target B cell; importantly , that phenotype includes the memory marker CD27 , which is induced on N-derived LCLs to levels similar to those retained on memory LCLs [31] . By contrast , EBV transformation does not lead to convergence of Ig isotype expression . Thus we detected no Ig isotype switching in N- or NSM-derived LCLs , whether analysed by transcript-specific RT/PCR assay or by protein expression ( Figure 3A , B and Table 1 ) . This accords with early work on the IgM+ IgD+ cell fraction from peripheral blood where EBV infection induced IgM but not IgG or IgA production [57] . However it apparently contradicts another study [58] in which switch circles , considered an early marker of class switching , and some switch transcripts were detected in naïve B cells after EBV infection in vitro . In that same study , however , evidence of isotype switching at the protein level was only given for clones of the EBV-negative BL line , Ramos , that had been transfected to express LMP1 [58] , the EBV latent cycle protein that mimics many of the effects of CD40 ligation [59] , [60] . Likewise another study [61] linking EBV with the induction of switch recombinase activity was based on viral infection of a B lymphoma line BJAB or on EBV-positive BL cells . In neither study , therefore , was there definitive evidence of EBV-induced isotype switching in the setting of normal B cells; indeed others have used EBV-infected B cells as isotype-stable substrates in order to study switching induced by exposure to cytokines and/or CD40 ligation [62] , [63] . We indeed confirmed that EBV transformation still leaves early passage N- and NSM-derived LCLs responsive to exogenous signals , CD40L/IL4/IL21 , that mimic T cell-derived switch signals in vivo ( Figure 3C ) . Thus EBV infection itself can induce naïve B cells to acquire an NSM-like surface phenotype ( IgD+ CD27+ ) and , with appropriate T cell help , a SM-like phenotype ( IgD− CD27+ ) . Key findings arose when the study turned to virus-induced changes Ig genotype . In limiting dilution seeding experiments , designed to generate transformed populations of limited clonality , we were surprised to find that >30% N-derived LD-LCLs contained mutated IgH sequences , a result seen consistently across experiments on 8 different B cell donors ( Figure 4 ) . Combining data from all these mutated clonotypes gives a mean of 9 . 2 nucleotide substitutions per IgH sequence , a value substantially higher than the background that would expected from PCR amplification error and in stark contrast to the mean values of 0 . 4 substitutions per sequence seen in the original N cell-sorts and of 0 . 5 substitutions per sequence seen in CD40L/IL4-expanded N cell cultures . Interestingly the 9 . 2 value is below the mean number of mutations ( 15 . 2 ) we observed in SM cell preparations but similar to the mean ( 7 . 2 ) seen in NSM cells and their derived LCLs . This similarity , and the fact that all N-derived transformants were IgM+ IgD+ , meant that if such mutated clonotypes were arising from pre-existing memory cells in the original N cell sorts , then the source of contamination must be NSM cells . However , such a possibility seems at odds both with the high purity of N cell preparations , shown to be >99% by IgD/CD27 staining and 97% by Ig gene sequencing ( Figure 1 ) , and with the fact that NSM cells were not more transformable than N cells when compared in parallel assays ( Figure 2 ) nor more likely to survive the clonal selection that then occurs with transition from transformed cell focus to established LCL . Having said that , we cannot entirely discount the possibility that some of these mutated clonotypes derive from NSM cells contaminating the N cell sorts , particularly in the rare cases of clones that also carry bcl6 mutations . Were NSM cells to be the source of even a small fraction of the Ig-mutated LCLs detected in the present work , this would be worthy of further attention since it implies that NSM-derived transformants enjoy an advantage over EBV-infected naive cells that is only apparent when the two are competing in mixed culture . However we would argue that this is not the main explanation for our findings . Rather , the evidence suggests that many of the mutated IgH clonotypes seen in LCLs derived from N cell preparations have arisen in vitro from naïve precursor cells . This rests on the observation of ongoing intra-clonal diversification within a culture . Early in the work we noted that clonotypic sequences varying by 1–3 nucleotides could be amplified from some LD-LCLs , but such examples , coming from single harvests made up to 12 weeks post-infection , are difficult to interpret in isolation . We therefore set up EBV-infected bulk N cell cultures that could be harvested prospectively , with parallel cultures of CD40L/IL4-stimulated N cell blasts as controls . Such experiments , carried out on N cell preparations from 6 different donors , led to the following conclusions . Firstly , the EBV-infected populations remain polyclonal for 4 weeks but thereafter move quite rapidly to oligo- or mono-clonality ( N . B . a previous CDR3 spectratyping study also reported clonal dominance within 12 weeks [52] ) ; by contrast parallel cultures of CD40L/IL4-stimulated blasts remain entirely polyclonal despite extensive proliferation ( Figure 6 , Tables 3 , 4 ) . Secondly , mutated IgH genotypes regularly appear in EBV-infected N cell cultures ( but only rarely in the CD40L/IL4 cultures ) and , in 5 of 6 experiments , grew out to become either the dominant or significant sub-dominant clones . Thirdly , and most importantly , in some cases family trees could be drawn tracing intra-clonal diversification with time in the emerging LCL ( Figure 9 ) . To date , almost all other reports of intra-clonal diversification in an LCL have come from one late passage , monoclonal antibody-producing line studied long after its establishment [64]–[66]; interestingly , however , a recent study [31] found that 1 of 3 naïve B cell-derived limiting dilution LCLs ( included as controls in a study of GC cell transformants ) showed significant levels of IgH gene diversification , a result which here we show to be representative of a general trend . It is striking that the IgH mutations observed in this study are frequently situated at known hotspots of SHM targeting ( Figure 8 ) . This , and the known ability of EBV to induce AID and other factors involved in the SHM process [58] , [64] , [67] , leads us to suggest that the changes are indeed AID-mediated . However we cannot say why EBV-induced , AID-driven IgH gene mutation was detected in only a proportion of N-derived LCLs in these experiments , when the expression of AID and related co-factors UNG and DNA polη was equally well induced in all lines ( Figure S2 ) . Mutation may be a stochastic event or possibly dependent on the cellular context , i . e . there may be differences among naïve B cells in their inducibility to IgH gene mutation , just as naive populations are heterogeneous in their response to isotype-switch signals in vitro [35] . These findings potentially bear upon two key aspects of EBV biology in vivo . The first concerns the means whereby the virus sequesters in memory B cell populations . It is clear from Ig genotyping of infected cells both in the transient lymphoproliferations seen during primary infection in IM tonsils [6] and in the progressive lymphoproliferative lesions arising in acutely immunosuppressed transplant recipients [10] , that at least some naive B cells do become infected in vivo . Hence the subsequent absence of the virus from naive populations in the immunocompetent host must reflect either clearance of these EBV-infected naive B cells by the T cell response ( which could happen were they unable to make the transition to an antigen-negative resting state ) or the cell's acquisition of a memory Ig genotype/phenotype . Our data raise the possibility that EBV infection per se could induce an NSM Ig genotype/phenotype without germinal centre transit , while T cell signals , perhaps in the extra-follicular environment [5] , [21] could induce switching to SM status . The second aspect of EBV biology touched on by this work concerns the mechanisms whereby virus infection contributes to B-lymphomagenesis . It is increasingly clear that the relatively low efficiency with which a B cell , driven into the first cell cycle by EBV infection , achieves outgrowth to an LCL ( 1–10% by some calculations [68] ) , cannot be fully ascribed to the inherent limitations of in vitro culture . Successful outgrowth reportedly requires escape from the oncogenic stress and attendant DNA damage response to which hyper-proliferating cells are subject within the first week ( i . e . first 5–6 divisions ) post-infection [69] and also survival through a transient period of genetic instability peaking around 4 weeks post-infection [51] . It remains to be seen to what extent virus-driven SHM activation contributes to these phenomena . However our findings on clonal evolution in N cell cultures from 4–8 weeks post-infection would suggest that naive cells in which SHM has targeted the Ig locus have a competitive edge in outgrowth to form an LCL . One attractive possibility is that the competitive advantage comes not from Ig mutation per se , but from mutations in other growth/survival-promoting genes that are coincidentally targeted by the SHM machinery . If such a mechanism , thought to be involved in the genesis of many lymphomas of GC/post-GC origin [70]–[72] plays some role in determining clonal selection during LCL outgrowth , then it represents a second way , in addition to expression of EBV's growth-transforming latent genes , whereby the virus may contribute to lymphomagenesis .
These studies were approved by the University of Birmingham and the South Birmingham Research Ethics Committee , UK ( 07/H/1207/271 ) . All blood donors provided written informed consent . Peripheral blood mononuclear cells ( PBMC ) were isolated by Ficoll-Isopaque density centrifugation of adult buffy coat samples ( National Blood Service , UK ) and B cells positively selected by immunomagnetic cell isolation using CD19 Pan B Dynabeads ( Life Technologies ) followed by bead detachment . B cell purity was assessed by staining with PE-Cy5-conjugated mouse anti-human CD20 ( 1∶40 , Dako ) monoclonal antibody ( mAb ) and FACS analysis on a Coulter Epics XL-MCL flow cytometer . To isolate B cell subsets , purified B cells were co-stained with FITC-labelled anti-IgD ( 1∶40 , Dako ) and PE-labelled anti-CD27 ( 1∶20 , BD Pharmingen ) antibodies ( Abs ) . IgD+ CD27− ( naïve ) , IgD− CD27+ ( switched memory ) and IgD+ CD27+ ( non-switched memory ) B cell subsets were simultaneously collected during a single FACS sorting procedure on a MoFlo sorter ( Beckman Coulter ) . Small aliquots of sorted cells were used to re-analyse purity of subsets thus isolated . Expression of the EBV receptor , CD21 [73] , [74] , was assessed by FACS analysis after staining B cells and subsets with PE-Cy5 mouse anti-human CD21 mAb ( 1∶40 , BD Pharmingen ) . For virus binding assays , 2×105 B cells from each subset were incubated with B95 . 8 strain EBV at 100 moi ( 100 EBV genome copies/cell ) for 3 hours at 4°C and the number of cell-bound virus genomes determined by real time quantitative PCR as described [75] . The transformation efficiency of different B cell subsets was assayed by two different experimental assays . In one assay , 1×106 N , NSM and SM B cells were exposed to a standard virus dose ( 50 moi ) for 1 hour at 37°C before seeding at two-fold limiting dilutions ( from 1000 cells/well ) into replicate wells of a 96-well plate containing human fibroblast feeder cells . In the second assay , 2×105 N and NSM cells were exposed for 1 hour at 37°C to two-fold virus dilutions ( from 50 moi ) before seeding 1000 cells into replicate wells of a 96-well plate containing human fibroblast feeder cells . In both cases , cultures were maintained in RPMI 1640 medium supplemented with 2 mM glutamine and 10% v/v FCS for 6 weeks at which time wells with typical foci of EBV-transformed lymphoblastoid cells were scored positive . At that time , all positive cultures at the end-point of the titration ( i . e . conditions where only a fraction of replicate wells had transformed ) were split , initially into duplicate microtest wells , then the duplicates pooled into a 1 ml well and thence into a 5 ml culture , in order to establish limiting dilution ( LD ) LCLs for analysis . The proportion of cultures which could be successfully expanded in this way over the following 4–6 weeks was recorded; as an internal control , when positive cultures from the high end of the titration ( i . e . conditions giving 100% positive wells ) were likewise treated , all could be successfully expanded to LCLs . In other experiments , bulk LCLs of N , NSM and SM origin were generated by infecting 2×106 cells of each type with a standard virus dose ( 50 moi ) , seeding into 2 ml wells and then expanding into a 25 ml flask; flask cultures were then maintained in exponential growth by passaging twice weekly . Activated B blasts were generated by culturing naive B cell preparations with irradiated CD40L-expressing mouse L cells in the presence of IL-4 [76] . Irradiated L cells were added to 24-well tissue culture plates ( 0 . 2×106 cell/well ) and allowed to adhere over night before the addition of 0 . 5×106 B cells in 3 ml CD40L blast medium ( Iscove's Medium supplemented with 10% human serum and 50 ng/ml IL-4 ) . Cultures were passaged onto fresh L cells twice weekly and maintained for up to 12 weeks . Ig isotype switching was induced by culturing N and NSM B cell-derived LCLs with irradiated CD40L-expressing mouse L cells in the presence of IL4 and IL21 [35] . Irradiated L cells were added to 12-well tissue culture plates ( 0 . 2×106 cell/well ) per well prior to the addition of 2×106 LCL cells in 1 . 5 ml CD40L blast medium supplemented with 50 ng/ml IL-4 and 50 ng/ml IL21 . The cells were split after 3–4 days in culture , and harvested after 7 days for Ig staining . LCLs were stained with fluorochrome-labelled mAbs against CD23 ( Immunotech ) and CD27 ( BD Pharmingen ) to determine the expression of these surface markers . Cells were also analysed for surface Ig by staining with fluorochrome-labelled isotype-specific mAbs to IgD , IgM , IgG and IgA ( Dako ) . All mAbs were used at 1∶20 and staining analysed using a Coulter Epics XL-MCL flow cytometer and WinMDI software . Total RNA was extracted from 2×106 cells using a Nucleospin II RNA isolation kit ( Macherey Nagel ) and cDNA was synthesised using random hexamers and AMV-RT ( Roche ) . Ig heavy chain ( IgH ) sequences were PCR-amplified using a single consensus forward primer FR1c within framework region ( FR ) 1 [77] and one of four IgH constant region reverse primers specific for IgM , IgD , IgG or IgA [78] . Each PCR amplification contained 0 . 8 µM FR1c primer , 0 . 8 µM IgH family-specific reverse primer , 200 µM dNTPs , 1 . 5 mM MgCl2 and 5units Red Hot DNA Polymerase ( ABgene ) . The first PCR cycle consisted of a denaturing step of 94°C for 5 mins , followed by 35 cycles of denaturation at 94°C for 30 s , annealing at 61°C for 60 s and extension at 72°C for 60 s ( 10 mins in the last cycle ) . PCR products were run on a 1 . 5% agarose gel and IgH products ( 600 bp ) identified by ethidium bromide staining . DNA was extracted from up to 2×106 cells using a DNeasy Tissue kit ( Qiagen ) . Ig heavy chain variable ( IgHV ) sequences spanning FR1 , CDR1 , FR2 , CDR2 , FR3 , and CDR3 were PCR-amplified using a single consensus forward primer FR1c 5′AGGTGCAGCTGSWGSAGTCDGG3′ and a mixture of JH family-specific reverse primers JH1/2/4/5 5′ACCTGAGGAGACGGTGACCAGGGT3′ , JH3 5′TACCTGAAGAGACGGTGACCATTGT3′ and JH6 5′ACCTGAGGAGACGGTGACCGTGGT3′ described previously [10] , [77] . PCR amplification was performed using the Expand High Fidelity PCR system ( Roche ) in a reaction containing 1 µg heat denatured genomic DNA , 0 . 8 µM FR1c primer , 0 . 26 µM each JH primer , 200 µM each dNTP , 1 . 5 mM MgCl2 and 3 . 5units Expand High Fidelity DNA polymerase . The first PCR cycle consisted of a denaturing step of 94°C for 5 mins , followed by 30 cycles of denaturation at 94°C for 30 s , annealing at 61°C for 60 s and extension at 72°C for 60 s ( 10 mins in the last cycle ) . PCR products were then separated by electrophoresis on a 8% polyacrylamide gel , and bands corresponding to the approximately 350 bp IgH products excised and the DNA eluted . Finally each purified PCR product was cloned using the pGEM-T Easy Vector System ( Promega ) , and following bacterial transformation , 3–5 independent bacterial clones were subjected to DNA sequencing on a Applied Biosystems ABI3730 DNA analyser ( Functional Genomics , University of Birmingham ) . IgH sequences were analysed using the V-Quest ( http://www . imgt . org ) [79] and VBase ( http://vbase . mrc-cpe . cam . ac . uk ) directories to identify the nearest IGHV , IGHD and IGHJ germline alleles; at this point sequences carrying out-of-frame CDR3 junction regions or chimaeric sequences generated by PCR crossover between two different IgH alleles were discarded from the analysis . Translated CDR3 junction sequences were generated using V-Quest . IgH sequences carrying the same IGHV allele and identical translated CDR3 sequences were assigned to the same CDR3 clone . IgH sequences with 0 , 1 or 2 nucleotide changes in the IGHV region ( between codons 9–92 , based on the Kabat numbering system [80] ) were considered to be naive , and those showing more than 2 nucleotide changes were considered to be mutated [10] . Sequence alignments were generated and visualised using BioEdit ( http://www . mbio . ncsu . edu/bioedit/bioedit . html ) . Total RNA isolation and cDNA synthesis reactions were carried out as described above . AID , uracil N-glycosylase ( UNG ) and DNA polymerase η transcripts were quantified by real time PCR on an ABI Prism 7500 Sequence Detection System using commercially available reagents ( Applied Biosystems ) . Values were normalised against GAPDH mRNA levels in the same cells and then expressed relative to that seen in a reference AID-positive B cell line , Ramos-BL [81] . A primer/probe combination to specifically detect an AID splice variant lacking exon 4 was designed using Primer Express software ( Applied Biosystems ) . Western blotting was carried out as described previously [82] using mAbs to: EBNA1 ( 1H4 ) , EBNA2 ( PE2 ) , LMP1 ( CS1-4 ) , BZLF1 ( BZ-1 ) . Bcl6 was detected using a rabbit polyclonal Ab ( N3 , Santa Cruz ) . A 765 bp fragment of the Bcl6 MMC was PCR-amplified with the primers 5′ CAAATGCTTTGGCTCCAAGTTTTCT 3′ and 5′ AGGAAGATCACGGCTCTGAAAGG 3′ PCR amplification using the Expand High Fidelity PCR system ( Roche ) in a reaction containing 1 µg heat denatured genomic DNA , 1 µM each primer , 200 µM each dNTP , 1 . 5 mM MgCl2 and 3 . 5 units Expand High Fidelity DNA polymerase . The first PCR cycle consisted of a denaturing step of 94°C for 5 mins , followed by 40 cycles of denaturation at 94°C for 30 s , annealing at 60°C for 60 s and extension at 72°C for 60 s ( 10 mins in the last cycle ) . PCR products were then separated by electrophoresis on a 1 . 5% agarose gel , the appropriate bands excised and the DNA eluted . Purified PCR products were cloned using the pGEM-T Easy Vector System ( Promega ) , and following bacterial transformation , 3–5 independent bacterial clones were subjected to DNA sequencing . A 716 bp region of each amplified Bcl6 sequence was aligned to a Bcl6 reference sequence using BioEdit .
|
Epstein-Barr virus ( EBV ) , a growth-transforming virus linked to several B cell lymphomas in man , is usually carried as an asymptomatic latent infection in B lymphocytes . Such virus carriage selectively involves memory , but not naive , B cells . How this selectivity is achieved is poorly understood since we find that naive and memory cell types are equally susceptible to infection and growth transformation to lymphoblastoid cell lines in vitro . Here we ask if EBV-transformation of purified naïve B cells can induce key features of memory cells , namely immunoglobulin ( Ig ) class switching and Ig gene mutation . We find that EBV does not induce Ig class switching ( though the infected cells remain responsive to exogenous switch signals ) but can induce Ig gene mutation . Thus , within 4 weeks of infecting naive B cell preparations , one can often detect cells carrying Ig mutations which appear to have arisen by somatic hypermutation in vitro . Furthermore , in many cases such cells become dominant during clonal evolution of the emergent EBV-transformed cell line . Overall these findings suggest a possible explanation as to why EBV is selectively found in memory B cell populations in vivo and why EBV-positive lymphoproliferative lesions/lymphomas so frequently involve clones with mutated Ig genotypes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"virology",
"immunology",
"biology",
"microbiology"
] |
2012
|
Epstein-Barr Virus Infection of Naïve B Cells In Vitro Frequently Selects Clones with Mutated Immunoglobulin Genotypes: Implications for Virus Biology
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.